Dear chess friends,
It is a pleasure to follow the ongoing successful work of the FIDE Arbiters' Commission in many aspects, including the online magazine for chess arbiters and the Arbiters’ Manual.
The Manual was launched several years ago as an instrument for arbiters and has proven to be of great help to National Federations, their respective Arbiters' councils and everybody in our chess community.
I am confident that this Manual will be instrumental in each Arbiter’s work and will facilitate and enrich his/her skills in order to exercise arbiter’s duties in the best way.
Commission’s daily work and brilliant organization of seminars, webinars and workshops has substantially increased the number and quality of chess arbiters throughout the world, including new Federations.
I l support and welcome the work and future plans of the Arbiters’ Commission and would like to wish all of you, first of all its Chairman and Councillors and all those who contribute to this tremendous work, lots of success, increasing audience and welltrained professional arbiters.
FIDE was founded in Paris on 20 July 1924 and one of its main objectives was to unify the rules of the game. The first official rules for chess were published in 1929 in French.
An update of the rules was published (once more in French) in 1952 with the amendments by the FIDE General Assembly.
There was another edition in 1966, with comments to the rules. Finally, in 1974 the Permanent Rules Commission published the first English edition with new interpretations and some amendments. In the following years the FIDE Rules Commission made some more changes, based on experience from competitions.
Major change was made in 1997, when the ‘more or less’ actual Laws of Chess were split into three parts: The Basic Rules of Play, the Competition Rules and Appendices.
In 2016 the Laws of Chess were split in 5 parts: The Basic Rules of Play, the Competition Rules, the Appendices, the Guidelines and the Glossary of terms of the Laws of Chess
The first part  Articles 1 to 5  is important for all people playing chess, including the basic rules that anyone who wants to play chess need to know.
The second part – Articles 6 to 12  mainly applies to chess tournaments.
The third part includes some appendices for Rapid games, Blitz games, the Algebraic notation of the games and the rules for play with blind and visually disabled players
The fourth part includes guidelines for adjourned games, for Chess 960 games and for games without increment, including Quick play finishes.
The fifth part includes a glossary of terms of the laws of Chess.
Starting from 1997 the FIDE Rules Commission (RC) makes changes of the Laws of Chess only every four years, coming into force on 1st July of the year following the decision.
Let us finish the history with the prefaces of the 1958 and 1974 Rules of Chess: 1958
“GENERAL OBSERVATIONS. The Laws of Chess cannot, and should not, regulate all possible situations that may arise during a game, nor they can regulate all questions of organization. In most cases not precisely regulate by an Article of the Laws, one should be able to reach a correct judgment by applying analogous stipulations for situations of a similar character. As to the arbiters’ tasks, in most cases one must presuppose that arbiters have the competence, sound of judgment, and absolute objectivity necessary. A regulation too detailed would deprive the arbiter of his freedom of judgment and might prevent him from finding the solution dictated by fairness and compatible with the circumstances of a particular case, since one cannot foresee every possibility.”1974
“FIDE INTERPRETATIONS. During recent years the Commission has been more or less overwhelmed by a steadily growing number of proposals and questions. That, of itself, is a good thing. However, there is a marked tendency in those many questions and proposals to bring more and more refinements and details into the Laws of Chess. Clearly the intention is to get more and more detailed instructions concerning “how to act in such and such case”. This may be profitable for a certain type of arbiter, but at the same time may be a severe handicap for another, generally the best, type of arbiter. The Commission in it’s entirely takes the firm position that the laws of Chess should be as short and as clear as possible. The Commission strongly believes that minor details should be left to the discretion of the arbiter. Each arbiter should have the opportunity, in case of a conflict, to take into account all the factors of the case and should be not bound by too detailed sub‐regulations which may be not applicable to the case in question. According to the Commission, the Laws of Chess must be short and clear and leave sufficient scope to the arbiter to deal with exceptional or unusual cases. The Commissions appeals to all chess federations to accept this view, which is in the interest of the hundreds of thousands of chess players, as well as of the arbiters, generally speaking. If any chess federation wants to introduce more detailed rules, it is perfectly free to do so, provided:
The Laws of Chess cannot cover all possible situations that may arise during a game, nor can they regulate all administrative questions. Where cases are not precisely regulated by an Article of the Laws, it should be possible to reach a correct decision by studying analogous situations which are regulated in the Laws. The Laws assume that arbiters have the necessary competence, sound judgement and absolute objectivity. Too detailed a rule might deprive the arbiter of his freedom of judgement and thus prevent him from finding a solution to a problem dictated by fairness, logic and special factors. FIDE appeals to all chess players and federations to accept this view. A necessary condition for a game to be rated by FIDE is that it shall be played according to the FIDE Laws of Chess. It is recommended that competitive games not rated by FIDE be played according to the FIDE Laws of Chess.
Member federations may ask FIDE to give a ruling on matters relating to the Laws of Chess.
The Preface of the Laws is one of the most important parts. Of course, the Laws of Chess cannot cover all possible situations arising during a game. Sometimes only a small part of a situation is changed and the Arbiter can make a decision based mainly on the Laws of Chess. For cases, however that are not covered completely by the Laws, the arbiter has to make a decision based on analogous situations that have happened in the past, as well as based on common logic, fairness and probably special factors. But it is always necessary for an arbiter to make his decisions and to solve all problems during the game. Therefore, an excellent knowledge of the Laws of Chess and his experience, obtained from his working in tournaments, are the most important qualifications of an Arbiter, together with objectivity at all times. 
Sometimes, neither white nor black can checkmate the opponent. In such a case, the game is drawn. The simplest example is when there are only the two kings on the chessboard. 
A white king usually indicated by the symbol K
A white queen usually indicated by the symbol Q
Two white rooks usually indicated by the symbol R
Two white bishops usually indicated by the symbol B
Two white knights usually indicated by the symbol N
Eight white pawns usually indicated by the symbol
A black king usually indicated by the symbol K
A black queen usually indicated by the symbol Q
Two black rooks usually indicated by the symbol R
Two black bishops usually indicated by the symbol B
Two black knights usually indicated by the symbol N
Eight black pawns usually indicated by the symbol
Staunton Pieces
p Q K B N R
A chessboard can be made of different materials, but the colour of the squares (dark = brown or black and light = white or cream) must be clearly different. It is useful that it is not shiny to avoid reflections and disturbance of players. The dimension of the chessboard must fit with the dimension of the pieces. (For more information see FIDE Handbook C.05 FIDE Tournament Rules). It is very important to check the orientation of the chessboard and the correct position of all the pieces before starting the game. By doing this, an arbiter can avoid a lot of possible claims about reversed Kings and Queens or Knights and Bishops. Sometimes, there is a disagreement between the players about how to place the knights. Each player has his own habits regarding this. Each player may place his own knights as he likes before the start of the game. He may only do so during the game when it is his move and after he has informed his opponent that he is going to adjust them (See Article 4: “J’adoube” – “I adjust”). 
Even if a piece is pinned against its own king, it attacks all the squares to which it would be able to move, if it were not pinned. 
Initially, each player has two bishops, one of which moves on light squares, the other one on dark squares. If a player has two or more bishops on squares of the same colour, it must be that the second bishop is the result of a promotion (See article 3.7.5.1), or an illegal move was played. 
When a player places an inverted (upside – down) Rook on the promotion square and continues the game, the piece is considered as a Rook, even if he names it as a “Queen” or any other piece. If he moves the upsidedown rook diagonally, it becomes an illegal move. In case of a promotion and if the player cannot find the required piece, he has the right to stop the game immediately and ask the Arbiter to bring him the piece he wants. Then the game continues after the promotion has been completed. 
Before white kingside castling / Before black queenside castling
After white kingside castling / After black queenside castling
Before white queenside castling / Before black kingside castling
After white queenside castling / After black kingside castling
4.1 Each move must be played with one hand only.
Article 4.2.1 may only be used to correct displaced pieces. Where the opponent is not present at the board, a player should inform the arbiter if there is an arbiter present – before he starts to adjust the pieces on the chess board. 
According to this rule, if a player has not said “I adjust” before touching a piece and touching the piece is not accidental, the touched piece must be moved. 
If an arbiter observes a violation of Article 4, he must always intervene immediately. He should not wait for a claim to be submitted by a player. 
A player may resign in a number of different ways:
All of these possibilities are capable of being misinterpreted. Therefore the situation has to be clarified. A player who does not wish to continue a game and leaves without resigning – or notifying the arbiter – is being discourteous. He may be penalized, at the discretion of the Chief Arbiter, for poor sportsmanship. 
This rule is applicable, only if Article 9.1.1 (not to agree for a draw before a specified number of moves by each player) is not enforced. The best way to conclude a game is to write down the result on the score sheet (if there is any, see Article 8) and for both players to sign it. This then forms a legal document, but. even then, things can go wrong. 
COMPETITION RULES
Article 6: The chessclock
6.1 Chessclock’ means a clock with two time displays, connected to each other in such a way that only one of them can run at one time.
‘Clock’ in the Laws of Chess means one of the two time displays.
Each time display has a ‘flag’.
‘Flagfall’ means the expiration of the allotted time for a player.
6.2.1 During the game each player, having made his move on the chessboard, shall stop his own clock and start his opponent’s clock (that is to say, he shall press his clock). This “completes” the move. A move is also completed if:
6.2.1.1 the move ends the game (see Articles 5.1.1, 5.2.1, 5.2.2, 9.6.1 and 9.6.2), or
6.2.1.2 the player has made his next move, when his previous move was not completed.
6.2.2 A player must be allowed to stop his clock after making his move, even after the opponent has made his next move. The time between making the move on the chessboard and pressing the clock is regarded as part of the time allotted to the player.
Some digital clocks show “ – “ instead of a flag.
Normally, when the player forgets to press his clock after making his move, the opponent has the following possibilities: (a) To wait for the player to press his clock. In this case there is a possibility to have a flag fall and the player to lose on time. Some may think that this is quite unfair, but the Arbiter cannot intervene and inform the player. (b) To remind the player to press his clock. In this case the game will continue normally. (c) To make his next move. In this case the player can also make his next move and press his clock. If the game is played with move counter, then one move has been missed by both players. In such a situation the Arbiter should, at an appropriate moment, (when he is sure it will not cause much distraction) intervene and press the lever of the clock once for each player (thus adding one move on the clock for each player). In this way the additional time of the next period (in case there is any) will be added properly (after move the first time control, usually move 40, has been completed).
6.2.3 A player must press his clock with the same hand with which he made his move. It is forbidden for a player to keep his finger on the clock or to ‘hover’ over it.
6.2.4 The players must handle the chessclock properly. It is forbidden to press it forcibly, to pick it up, to press the clock before moving or to knock it over. Improper clock handling shall be penalised in accordance with Article 12.9.
6.2.5 Only the player whose clock is running is allowed to adjust the pieces.
6.2.6 If a player is unable to use the clock, an assistant, who must be acceptable to the arbiter, may be provided by the player to perform this operation. His clock shall be adjusted by the arbiter in an equitable way. This adjustment of the clock shall not apply to the clock of a player with a disability.
6.3.1 When using a chessclock, each player must complete a minimum number of moves or all moves in an allotted period of time including any additional amount of time with each move. All these must be specified in advance.
The following situation may happen: A player makes a move, forgets to press the clock and leaves the table (for example to go to the toilet). After he returns he sees that his clock is running and believing that his opponent has completed his move he makes another move and presses the clock. In this situation the Arbiter must be summoned immediately to clarify the situation (did the opponent make a move or not?) and make the necessary corrections on the clock and the board.
A game may have more than one period. The requirements of the allotted number of moves and the additional amount of time with each move for each period must be specified in advance. These parameters should not change during a tournament. But, if there is a playoff, it is likely they will then change.
Sometimes the following happens: A player displaces some pieces. The opponent keeps his finger on the clock button to prevent the player pressing his clock. This is forbidden according to this Article.
If a player makes a move with one hand and presses the clock with the other, it is not considered as an illegal move, but it is penalized according to the article 12.9.
Where a player presses the clock without making a move, as mentioned in the article 6.2.4, it is considered as an illegal move and it is penalized according to the article 7.5.3.
It is clear that the player himself has to provide an assistant. He has to introduce this assistant in time to the arbiter, not just before the round. It is usual that 10 minutes are deducted from the time of the player who needs an assistant. No deduction should be made in the case of a disabled player. An example where a player is unable to use the clock is for religious reasons.
6.3.2 The time saved by a player during one period is added to his time available for the next period, where applicable.
In the timedelay mode both players receive an allotted ‘main thinking time’. Each player also receives a ‘fixed extra time’ with every move. The countdown of the main thinking time only commences after the fixed extra time has expired. Provided the player presses his clock before the expiration of the fixed extra time, the main thinking time does not change, irrespective of the proportion of the fixed extra time used.
6.4 Immediately after a flag falls, the requirements of Article 6.3.1 must be checked.
6.5 Before the start of the game the arbiter shall decide where the chessclock is placed.
1. Cumulative (Fischer) mode: Here each player has a main thinking time and receives a fixed extra time (increment) for each move. This increment for his first move is added before the game starts and then immediately after he has completed each of his following moves. If a player completes his move before the remaining time of this increment for the move expires, this remaining time will have been added to the main thinking time. 2. Bronstein mode: The main difference between Fisher mode and Bronstein mode is the handling of the extra time. If the player does not use the whole extra time in Bronstein mode, the remaining part is lost. 3. Time delay mode: Each player receives a main thinking time. When a player has the move the clock will not start counting for a fixed period. After this period expired the clock is counting down the main playing time.
This means that the arbiter and/or the players have to check if the minimum numbers of moves have been completed. Consider a game 90 minutes for 40 moves and 30 minutes for the rest of the game. It is normal to investigate whether 40 moves have been made by both players only after a flag has fallen. Unless a digital board is used this when the arbiter may determine the number of moves played from time.
If a move (push) counter is used in a digital clock, then it is possible to establish whether 40 moves have been made before a flag fall, as a ““ indication appears on the clock in case that the player does not complete the 40 moves before the allotted time. In the majority of the top tournaments the move counter is used.
Where electronic clocks are used and both clocks show 0.00, the Arbiter can usually establish which flag fell first, with the help of the ““ or any other flag indication. Where mechanical clocks are used then article III.3.1 of the Guidelines about games without increment including Quickplay Finishes are applied.
6.6 At the time determined for the start of the game White’s clock is started.
6.7.1 The regulations of an event shall specify a default time in advance. If the default time is not specified, then it is zero. Any player who arrives at the chessboard after the default time shall lose the game unless the arbiter decides otherwise.
6.7.2 If the regulations of an event specify that the default time is not zero and if neither player is present initially, White shall lose all the time that elapses until he arrives, unless the regulations of an event specify or the arbiter decides otherwise.
6.8 A flag is considered to have fallen when the arbiter observes the fact or when either player has made a valid claim to that effect.
In individual tournaments the chess clock is normally placed on the right side of the player who has the black pieces. The chess boards shall be placed so that the arbiter is able to check as many clocks as possible at the same time. In the case of a lefthanded player with black pieces, the arbiter might arrange for the players to sit on the other side of the board. In team competitions the members of the same team usually sit in a row. Then the pieces are set alternate black and white and the clocks all point the same way. Be careful! It happens quite often in team competitions that a player presses the clock of his neighbour.
In smaller tournaments the arbiters start all clocks. In tournaments with many players the arbiter announces the start of the round and states that White’s clock is started. The arbiter then goes round the room checking that White’s clock has been started on all boards. Where the push counter is used to add time after the first time control (often 40 moves), it is desirable for arbiters to start all White’s clocks.
The start of the session is the moment, when the arbiter announces it. If the default time is 0, the arbiter has to declare the game lost for the players who are not present at their chessboards. It is preferable to install a large digital countdown device in the playing hall. For FIDE events with fewer than 30 players an appropriate announcement must be made five minutes before the round is due to start and again one minute before start of the game. Alternatively, a clock should be on the wall inside the playing hall and provide the official time of the tournament.
If the default time is not 0, it is advisable that the arbiter publicly announces the time of the start of the round and that he writes down the starting time. If the default time is for example 30 minutes and the round was scheduled to start at 15.00, but actually started at 15.15, then any player who hasn’t arrived by 15.45 loses.
A flag is considered to have fallen when it is noticed or claimed, not when it physically happened.
6.9 Except where one of Articles 5.1.1, 5.1.2, 5.2.1, 5.2.2, 5.2.3 applies, if a player does not complete the prescribed number of moves in the allotted time, the game is lost by that player. However, the game is drawn if the position is such that the opponent cannot checkmate the player’s king by any possible series of legal moves.
6.10.1 Every indication given by the chessclock is considered to be conclusive in the absence of any evident defect. A chessclock with an evident defect shall be replaced by the arbiter, who shall use his best judgement when determining the times to be shown on the replacement chessclock.
6.10.2 If during a game it is found that the setting of either or both clocks is incorrect, either player or the arbiter shall stop the chessclock immediately. The arbiter shall install the correct setting and adjust the times and movecounter, if necessary. He shall use his best judgement when determining the clock settings.
6.11.1 If the game needs to be interrupted, the arbiter shall stop the chessclock.
6.11.2 A player may stop the chessclock only in order to seek the arbiter’s assistance, for example when promotion has taken place and the piece required is not available.
6.11.3 The arbiter shall decide when the game restarts.
6.11.4 If a player stops the chessclock in order to seek the arbiter’s assistance, the arbiter shall determine whether the player had any valid reason for doing so. If the player had no valid reason for stopping the chessclock, the player shall be penalised in accordance with Article 12.9.
It means that a simple flag fall might not lead the Arbiter to declare the game lost for the player whose flag has fallen. The Arbiter has to check the final position on the chessboard and only if the opponent can checkmate the player’s king by any possible series of legal moves, can he declare the game won by the opponent. In case there are forced moves that lead to a checkmate by the player or to a stalemate, then the result of the game is declared as a draw.
It is desirable to check the clocks during the round, for instance every 30 minutes, and to record the times and the number of moves made, by using a timecontrol sheet (see at the end of the Manual). This can be particularly valuable when an increment is used. If a chess clock must be replaced, it must be done as soon as possible and it is essential to mark it as defective and to separate it from the clocks that work correctly.
It is essential to write down all the known details of the two clocks before making any adjustment.
For example, if a fire alarm goes off.
6.12.1 Screens, monitors, or demonstration boards showing the current position on the chessboard, the moves and the number of moves made/completed, and clocks which also show the number of moves, are allowed in the playing hall.
6.12.2 The player may not make a claim relying only on information shown in this manner.
Article 7: Irregularities
7.1 If an irregularity occurs and the pieces have to be restored to a previous position, the arbiter shall use his best judgement to determine the times to be shown on the chessclock. This includes the right not to change the clock times. He shall also, if necessary, adjust the clock’s movecounter.
7.2.1 If during a game it is found that the initial position of the pieces was incorrect, the game shall be cancelled and a new game shall be played.
7.2.2 If during a game it is found that the chessboard has been placed contrary to Article 2.1, the game shall continue but the position reached must be transferred to a correctly placed chessboard.
7.3 If a game has started with colours reversed then, if less than 10 moves have been made by both players, it shall be discontinued and a new game played with the correct colours. After 10 moves or more, the game shall continue.
7.4.1 If a player displaces one or more pieces, he shall reestablish the correct position in his own time.
A player may stop the clocks if he feels disturbed by his opponent or by spectators or is unwell. Going to the toilet is not necessarily a valid reason for stopping the clocks. A disabled player must be treated with due respect.
An arbiter or player must realise that the information displayed may be incorrect.
Be aware that the error was found during and not after the game. It is not mentioned how the mistake was found or who found it. If a game is played on an electronic chessboard, it can happen that the computer stops recording the moves. In such cases the operator should inform the arbiter that something has gone wrong and it is the arbiter’s duty to check what has happened.
After 10 moves or more the game shall continue, otherwise, a new game shall be played with the correct colours. It doesn’t matter what the current position on the chessboard is and how many pieces or pawns have been captured by the ninth move.
If a game with reversed colours has ended by normal means (for example checkmate, resignation or draw agreement), in less than ten (10) moves, then the result stands.
7.4.2 If necessary, either the player or his opponent shall stop the chessclock and ask for the arbiter’s assistance.
7.4.3 The arbiter may penalise the player who displaced the pieces.
7.5.1 An illegal move is completed once the player has pressed his clock. If during a game it is found that an illegal move has been completed, the position immediately before the irregularity shall be reinstated. If the position immediately before the irregularity cannot be determined, the game shall continue from the last identifiable position prior to the irregularity. Articles 4.3 and 4.7 apply to the move replacing the illegal move. The game shall then continue from this reinstated position.
7.5.2 If the player has moved a pawn to the furthest distant rank, pressed the clock, but not replaced the pawn with a new piece, the move is illegal. The pawn shall be replaced by a queen of the same colour as the pawn.
7.5.3 If the player presses the clock without making a move, it shall be considered and penalized as if an illegal move.
7.5.4 If a player uses two hands to make a single move (for example in case of castling, capturing or promotion) and pressed the clock, it shall be considered and penalized as if an illegal move.
7.5.5 After the action taken under Article 7.5.1, 7.5.2, 7.5.3 or 7.5.4 for the first completed illegal move by a player, the arbiter shall give two minutes extra time to his opponent; for the second completed illegal move by the same player the arbiter shall declare the game lost by this player. However, the game is drawn if the position is such
The Arbiter must be very careful here. Suppose player A has the move and his clock is running. Then player B displaces one of his own pieces (by accident). It is not correct that player A starts player B’s clock. A should stop both clocks and summon the arbiter. This Article should be applied flexibly.
Most problems happen in Rapidplay or Blitz. The penalty should be according to Article 12.9. A player should not be forfeited immediately for accidentally displacing a piece. If he did it deliberately, perhaps in order to gain time, or does it several times, that is different.
It is very important that the irregularity must be discovered during the game. After the players have signed the scoresheets or it is clear in another way that the game is over, corrections are not possible. The result stands. Where the irregularity is discovered during the game, it is important, that the game continues by moving the piece with which the illegal move was played or that the piece which was taken is taken by another piece, if possible.
A move cannot be declared illegal until the player completed his move by stopping his clock. So, the player can correct his move without being penalized, even if he had already released the piece on the board, provided he hasn’t press the clock. Of course, he must comply with relevant parts of article 4. that the opponent cannot checkmate the player’s king by any possible series of legal moves.
7.6 If, during a game it is found that any piece has been displaced from its correct square, the position before the irregularity shall be reinstated. If the position immediately before the irregularity cannot be determined, the game shall continue from the last identifiable position prior to the irregularity. The game shall then continue from this reinstated position.
Article 8: The recording of the moves
8.1.1 In the course of play each player is required to record his own moves and those of his opponent in the correct manner, move after move, as clearly and legibly as possible, in the algebraic notation (Appendix C), on the ‘scoresheet’ prescribed for the competition.
8.1.2 It is forbidden to write the moves in advance, unless the player is claiming a draw according to Article 9.2, or 9.3 or adjourning a game according to Guidelines I.1.1
8.1.3 A player may reply to his opponent’s move before recording it, if he so wishes. He must record his previous move before making another.
8.1.4 The scoresheet shall be used only for recording the moves, the times of the clocks, offers of a draw, matters relating to a claim and other relevant data.
8.1.5 Both players must record the offer of a draw on the scoresheet with a symbol (=).
8.1.6 If a player is unable to keep score, an assistant, who must be acceptable to the arbiter, may be provided by the player to write the moves. His clock shall be adjusted by the arbiter in an equitable way. This adjustment of the clock shall not apply to a player with a disability.
It is advisable that the investigation to determine from which position the game shall be continued takes place by the two players and under the supervision of the arbiter.
The player is forfeited in case he completes two (2) of ANY of the above illegal moves. However when there are two (2) illegal moves in one move (for example illegal castling made by two hands, illegal promotion made by two hands and illegal capturing made by two hands), they count as one (1) illegal move and the player shall not be forfeited, unless it is the second such transgression.
Capturing of the opponent’s King is illegal and is penalized accordingly.
8.2 The scoresheet shall be visible to the arbiter throughout the game.
8.3 The scoresheets are the property of the organiser of the competition.
8.4 If a player has less than five minutes left on his clock at some stage in a period and does not have additional time of 30 seconds or more added with each move, then for the remainder of the period he is not obliged to meet the requirements of Article 8.1.1.
8.5.1 If neither player keeps score under Article 8.4, the arbiter or an assistant should try to be present and keep score. In this case, immediately after a flag has fallen the arbiter shall stop the chessclock. Then both players shall update their scoresheets, using the arbiter’s or the opponent’s scoresheet.
8.5.2 If only one player has not kept score under Article 8.4, he must, as soon as either flag has fallen, update his scoresheet completely before moving a piece on the chessboard. Provided it is that player’s move, he may use his opponent’s scoresheet, but must return it before making a move.
8.5.3 If no complete scoresheet is available, the players must reconstruct the game on a second chessboard under the control of the arbiter or an assistant. He shall first record the actual game position, clock times, whose clock was running and the number of
Notice that it is forbidden to record the move in advance. Only in case of a draw claim (Article 9.2. and 9.3) and adjourning is it allowed. It is permitted to record the moves as a pair (his opponent’s move and his own), but he must have recorded his previous own move before making his next move. Even if an opponent has only one legal response, this must not be recorded by the player in advance.
Nowadays there are generally no problems with this Article. The habit of concealing the move written on the score sheet moves with a pen does not violate this article. But still the arbiter has the right to remove the pen from the score sheet, whenever he wants to check the number of the moves played by the players, provided it does not distract the opponent.
A player is not allowed to keep the original score sheet. It belongs to the Organisers. The player has to deliver it to the arbiter when the game is finished and should keep a copy (if any).
It happens quite often that in this time trouble phase the player asks the arbiter how many moves are left until the time control. The arbiter shall never give any information about the number of the moves that have been made, even after a player or both players have completed the required number of moves. The arbiter should act, only after a flag fall. He stops both clocks and orders the players to update their scoresheets. The arbiter shall start the clock of the player who has the move, only after they have done this. A player rarely takes too long over this, sometimes contemplating his next move. Then he should be warned.
Notice that, in this situation, after a flag fall, the arbiter does not stop the clocks.moves made/completed, if this information is available, before reconstruction takes place.
8.6 If the scoresheets cannot be brought up to date showing that a player has overstepped the allotted time, the next move made shall be considered as the first of the following time period, unless there is evidence that more moves have been made or completed.
8.7 At the conclusion of the game both players shall sign both scoresheets, indicating the result of the game. Even if incorrect, this result shall stand, unless the arbiter decides otherwise.
Article 9: The drawn game
9.1.1 The regulations of an event may specify that players cannot offer or agree to a draw, whether in less than a specified number of moves or at all, without the consent of the arbiter.
9.1.2 However, if the regulations of an event allow a draw agreement the following shall apply:
The reconstruction should take place after both clocks have been stopped and should be done away from the other games, so as not to disturb them.
It is very important for the Arbiter to record the correct result of the games. At the moment the Arbiter sees that a game has been finished, he should go to that board and check if the players have recorded the result of the game and signed both scoresheets. The arbiter should immediately check that both score sheets show the same result.
This article allows the Chief Arbiter to overrule decisions made by other arbiters, even after the players have signed the scoresheets or the match protocols. It has also been known for both players to record the wrong result. This permits the arbiter to correct such errors.
If a competition applies this rule, then the mentioned number of moves or the no agreement at all, should be communicated with the players in the invitation to the tournament. It is advisable for the Arbiter to repeat the rule before the start of the tournament. It is clear that the rule applies only for draw agreements. Articles 9.2, 9.3 and 9.6 still apply during the whole game and give the possibility to the players to have a draw in less that the specified number of moves, which must be accepted by the Arbiter. For example, if two players make a draw by threefold repetition after 20 moves, in a tournament where there is a draw restriction rule before 30 moves have been completed by both players, then the Arbiter must allow the draw. If neither player claims a draw by threefold repetition, so that the repetition takes place five times, then the arbiter must step in and declare the game drawn, see 9.6.1.
9.1.2.1 A player wishing to offer a draw shall do so after having made a move on the chessboard and before pressing his clock. An offer at any other time during play is still valid but Article 11.5 must be considered. No conditions can be attached to the offer. In both cases the offer cannot be withdrawn and remains valid until the opponent accepts it, rejects it orally, rejects it by touching a piece with the intention of moving or capturing it, or the game is concluded in some other way.
9.1.2.2 The offer of a draw shall be noted by each player on his scoresheet with the symbol (=).
9.1.2.3 A claim of a draw under Article 9.2 or 9.3 shall be considered to be an offer of a draw.
9.2.1 The game is drawn, upon a correct claim by a player having the move, when the same position for at least the third time (not necessarily by a repetition of moves):
9.2.1.1 is about to appear, if he first writes his move, which cannot be changed, on his scoresheet and declares to the arbiter his intention to make this move, or
9.2.1.2 has just appeared, and the player claiming the draw has the move.
9.2.2 Positions are considered the same if and only if the same player has the move, pieces of the same kind and colour occupy the same squares and the possible moves of all the pieces of both players are the same. Thus positions are not the same if:
9.2.2.1 at the start of the sequence a pawn could have been captured en passant
This is a valuable rule for the arbiter and its use should be encouraged.
The correct sequence of a draw offer is clear: 1. making a move 2. offering of a draw 3. pressing the clock. If a player deviates from this order, the offer still stands though it has been offered in an incorrect manner. The arbiter in this case has to penalise the player, according to the Article 12.9. No conditions can be attached to a draw offer. Some examples: The player requires the opponent to accept the offer within 2 minutes. In a team competition: a draw is offered under the condition that another game in the match shall be resigned or shall be drawn as well. In both cases the offer of a draw is valid, but not the attached condition. Regarding 9.1.2.3: If a player claims a draw, the opponent has the possibility to agree immediately to the draw. In this case the arbiter does not need to check the correctness of the claim. But be careful. If there is a draw restriction (for example: no draw offers are allowed before 30 moves have been completed by both players) and the claim has been submitted before that move (perhaps after 28 moves), then the claim has to be checked by the Arbiter, even if the opponent would agree to the draw.
9.2.2.2 a king had castling rights with a rook that has not been moved, but forfeited these after moving. The castling rights are lost only after the king or rook is moved.
9.3 The game is drawn, upon a correct claim by a player having the move, if:
9.3.1 he writes his move, which cannot be changed, on his scoresheet and declares to the arbiter his intention to make this move which will result in the last 50 moves by each player having been made without the movement of any pawn and without any capture, or
9.3.2 the last 50 moves by each player have been completed without the movement of any pawn and without any capture.
9.4 If the player touches a piece as in Article 4.3, he loses the right to claim a draw under Article 9.2 or 9.3 on that move.
9.5.1 If a player claims a draw under Article 9.2 or 9.3, he or the arbiter shall stop the chessclock (see Article 6.12.1 or 6.12.2). He is not allowed to withdraw his claim.
9.5.2 If the claim is found to be correct, the game is immediately drawn.
9.5.3 If the claim is found to be incorrect, the arbiter shall add two minutes to the opponent’s remaining thinking time. Then the game shall continue. If the claim was based on an intended move, this move must be made in accordance with Articles 3 and 4.
9.6 If one or both of the following occur(s) then the game is drawn:
9.6.1 the same position has appeared, as in 9.2.2 at least five times.
9.6.2 any series of at least 75 moves have been made by each player without the movement of any pawn and without any capture. If the last move resulted in checkmate, that shall take precedence.
Only the player whose move it is, and whose clock is running, is allowed to claim a draw in this way.
The correctness of a claim must be checked in the presence of both players. It is also advisable to replay the game and not to decide by only using the score sheets. If electronic boards are used it is possible to check it on the computer.
See comment to article 9.2.
The player loses his right to claim a draw only on that move. He has always the possibility to make a new claim in the game later, provided the circumstances haven’t changed.
This claim is not treated as an illegal move.
It is mentioned that the intended move must be played, but if the intended move is illegal, another move with this piece must be made. All the other details of Article 4 are also valid.
Article 10: Points
10.1 Unless the regulations of an event specify otherwise, a player who wins his game, or wins by forfeit, scores one point (1), a player who loses his game, or forfeits, scores no points (0), and a player who draws his game scores a half point (½).
10.2 The total score of any game can never exceed the maximum score normally given for that game. Scores given to an individual player must be those normally associated with the game, for example a score of ¾  ¼ is not allowed.
Article 11: The conduct of the players
11.1 The players shall take no action that will bring the game of chess into disrepute.
11.2.1 The ‘playing venue’ is defined as the ‘playing area’, rest rooms, toilets, refreshment area, area set aside for smoking and other places as designated by the arbiter.
11.2.2 The playing area is defined as the place where the games of a competition are played.
11.2.3 Only with the permission of the arbiter can:
11.2.3.1 a player leave the playing venue,
11.2.3.2 the player having the move be allowed to leave the playing area.
11.2.3.3 a person who is neither a player nor arbiter be allowed access to the playing area.
In 9.6.1 case, the five times need not be consecutive. In both 9.6.1 and 9.6.2 cases the Arbiter must intervene and stop the game, declaring it as a draw.
Another scoring system from time to time used is for a win 3 points, for a draw 1 point and for a lost game 0 points. The idea is to encourage more positive play. Another is win 3 points, draw 2, loss 1 and forfeit 0. This is to discourage forfeits and may encourage children particularly as they gain a point despite losing. Yet another is win 2, draw 1, loss 0. This avoids ½ on the results sheet.
This is an Article which can be used for any infringements that are not specifically mentioned in the articles of the Laws of Chess.
If possible, spectators should not enter the playing area. It is advisable to have all other rooms (smoking areas, toilets, refreshment areas, and so on.) always to be under the control of the Arbiters or assistants.
11.2.4 The regulations of an event may specify that the opponent of the player having a move must report to the arbiter when he wishes to leave the playing area.
11.3.1 During play the players are forbidden to use any notes, sources of information or advice, or analyse any game on another chessboard.
11.3.2.1 During a game, a player is forbidden to have any electronic device not specifically approved by the arbiter in the playing venue.
However, the regulations of an event may allow such devices to be stored in a player’s bag, provided the device is completely switched off. This bag must be placed as agreed with the arbiter. Both players are forbidden to use this bag without permission of the arbiter.
11.3.2.2 If it is evident that a player has such a device on their person in the playing venue, the player shall lose the game. The opponent shall win. The regulations of an event may specify a different, less severe, penalty.
11.3.3 The arbiter may require the player to allow his clothes, bags, other items or body to be inspected, in private. The arbiter or person authorised by the arbiter shall inspect the player, and shall be of the same gender as the player. If a player refuses to cooperate with these obligations, the arbiter shall take measures in accordance with Article 12.9.
This article should not be confused with Articles 11.2.3.1 and 11.2.3.2. In 11.2.3.1 it is prohibited for any player to leave the playing venue without the permission of the arbiter and in 11.2.3.2 it is prohibited to leave the playing area for the player having the move. But in 11.2.4 it is possible to include, in the regulations, prohibition of the opponent leaving the playing area without permission of the arbiter.
The regulations about electronic devices are now very strict. No mobile phone is allowed in the playing venue and it makes no difference if it is switched on or off. If a mobile phone (even switched off) is found with a player, his game is immediately lost and the opponent shall win. The result shall be 10 or 01. It doesn’t matter if, when the mobile phone is found, the opponent cannot checkmate the offending player by any series of legal moves: he wins the game. The opponent may have cheated earlier. It is different if the game has not yet started. Suppose the following situation occurs: There is no zero‐tolerance. Player A is in the playing hall at the start of the round. His opponent, Player B is absent. Immediately after player A made his first move his mobile rings. The arbiter declares the game lost for Player A. Some minutes later, but still on time, Player B arrives. The score is “‐/+”, it is not a “played” game and it cannot be rated.
11.3.4 Smoking, including ecigarettes, is permitted only in the section of the venue designated by the arbiter.
11.4 Players who have finished their games shall be considered to be spectators.
11.5 It is forbidden to distract or annoy the opponent in any manner whatsoever. This includes unreasonable claims, unreasonable offers of a draw or the introduction of a source of noise into the playing area.
11.6 Infraction of any part of Articles 11.1 – 11.5 shall lead to penalties in accordance with Article 12.9.
11.7 Persistent refusal by a player to comply with the Laws of Chess shall be penalised by loss of the game. The arbiter shall decide the score of the opponent.
11.8 If both players are found guilty according to Article 11.7, the game shall be declared lost by both players.
11.9 A player shall have the right to request from the arbiter an explanation of particular points in the Laws of Chess.
However, there is the possibility for an arbiter or an organizer to specify in advance (in the regulations of the event) a less severe penalty for a violation of this article (perhaps a fine) They can also include in the regulations of the event the possibility of bringing such a device to the tournament provided that certain conditions are fulfilled: that it is completely switched off and stored in a separate bag, so that it is not in contact with the player and the player does not have access to the bag during the game, without the arbiter's permission (and he cannot take the bag with him to the toilet, and so on.).
If possible, this smoking area should be close to the playing area and supervised by an Arbiter or an Assistant.
It means that the players, who finished their games, have to leave the playing area. Nevertheless, give them a few minutes to watch the other boards, making sure they do not disturb players still in play.
Even if the draw offers or claims are quite reasonable, repeating them too often can annoy the opponent. The Arbiter must always intervene when the opponent is disturbed or distracted.
It is very difficult to give a general guideline for the application of this Article, but if an arbiter has to warn the player for the third or fourth time, this is a good reason to declare the game lost. It is necessary to inform the player that Article 11.7 shall be applied at the next infringement.
11.10 Unless the regulations of an event specify otherwise, a player may appeal against any decision of the arbiter, even if the player has signed the scoresheet (see Article 8.7).
11.11 Both players must assist the arbiter in any situation requiring reconstruction of the game, including draw claims.
11.12 Checking three times occurrence of the position or 50 moves claim is a duty of a the players, under supervision of the arbiter.
Article 12: The role of the Arbiter (see Preface)
12.1 The arbiter shall see that the Laws of Chess are observed.
12.2 The arbiter shall:
12.2.1 ensure fair play,
12.2.2 act in the best interest of the competition,
12.2.3 ensure that a good playing environment is maintained,
12.2.4 ensure that the players are not disturbed,
12.2.5 supervise the progress of the competition,
12.2.6 take special measures in the interests of disabled players and those who need medical attention,
The Arbiter must take care to avoid any kind of cheating by the players. 
8.5 when at least one flag has fallen. The arbiter shall refrain from informing a player that his opponent has completed a move or that the player has not pressed his clock.
This Article includes also the calling of a flag fall. 
Nobody is allowed to use their mobiles in the playing hall during the games. 
Article 12.9.9. is also be applied in cooperation with the organiser of the event. 
Example 1: According to the Tournament Regulations of an event, the time control is 30 minutes for the whole game and 30 seconds increment for each move. That is: for 60 moves we would get 30'+ (30"x 60) = 30' +30' = 60'. So as according to the Article A1 "A Rapidplay" is a game where all moves must be completed in less than 60 minutes for each player, then such a game is considered to be standard chess. Example 2: According to the Tournament Regulations of an event, the time control is 10 minutes for the whole game and 5 seconds increment for each move. That is: for 60 moves we would get 10'+ (5” x 60) = 10' +5' = 15'. So as according to the Article A.1 such a game is considered to be Rapidplay chess 
Players are allowed to record the moves, but they may stop recording any time they wish. 
In case that a player asks the Arbiter to show him the score sheet, the clock should not be stopped. 
In case of incorrect king placement, castling is not allowed. In case of incorrect rook placement, castling with this rook is not allowed.
It means that the player does not lose the game with the first illegal move, but only with the second, as it is in standard chess. The penalty is also the same as in standard chess. 
In case that both clocks show the indication 0.00, no claim for win on time can be submitted by the players, but the Arbiter shall decide the result of the game according the indication of ““ that is shown on one of the clocks. The player whose clock shows this indication loses the game. 
If the player completes a move by giving a check and the opponent completes his next move by also giving a check (creating a position where both Kings are in check), and the player, instead of claiming the opponent’s illegal move, completes his next move with his King, avoiding the check, then the game shall be continued by the next opponent’s move, as the new position is not illegal any more. No illegal move can be claimed by the opponent. 
It is an obligation of the Arbiter to call the flag fall, if he sees it. 
According to the Tournament Regulations of an event the time control is 5 minutes for the whole game and 5 seconds increment for each move. That is: for 60 moves we would get 5'+ (5'x60) = 5'+5' = 10'. According to Art. B.1 we have a Blitz game. 
In Blitz games the Arbiter SHALL CALL the flag fall, if he observes it, as in Rapid play 
In Blitz games, if the player asks from the Arbiter to see the score sheet, the clock should not be stopped, as in Rapidplay. 
FIDE recognises for its own tournaments and matches only one system of notation, the Algebraic System, and recommends the use of this uniform chess notation also for chess literature and periodicals. Score sheets using a notation system other than algebraic may not be used as evidence in cases where normally the score sheet of a player is used for that purpose. An arbiter who observes that a player is using a notation system other than the algebraic should warn the player of this requirement.
A longer form containing the square of departure is acceptable. Examples: Bb2e5, Ng1f3, Ra1d1, e7e5, d2d4, a6a5.
00 = castling with rook h1 or rook h8 (kingside castling)
000 = castling with rook a1 or rook a8 (queenside castling) x = captures
+ = check
++ or # = checkmate
e.p. = captures ‘en passant’ The last four are optional.
Sample game: 1.e4 e5 2. Nf3 Nf6 3. d4 exd4 4. e5 Ne4 5. Qxd4 d5 6. exd6 e.p. Nxd6 7. Bg5 Nc6 8. Qe3+ Be7 9. Nbd2 00 10. 000 Re8 11. Kb1 (=)
Or: 1. e4 e5 2. Nf3 Nf6 3. d4 ed4 4. e5 Ne4 5. Qd4 d5 6. ed6 Nd6 7. Bg5 Nc6 8. Qe3
Be7 9 Nbd2 00 10. 000 Re8 11. Kb1 (=)
Or: 1. e2e4 e7e5 2.Ng1f3 Ng8f6 3. d2d4 e5xd4 4. e4e5 Nf6e4 5. Qd1xd4 d7d5 6.
e5xd6 e.p. Ne4xd6 7. Bc1g5 Nb8c6 8. Qd4d3 Bf8e7 9. Nb1d2 00 10. 000 Rf8e8
11. Kb1 (=)
A  Anna B  Bella C  Cesar D  David E  Eva
F  Felix G  Gustav H  Hector
Unless the arbiter decides otherwise, ranks from White to Black shall be given the German numbers
Castling is announced “Lange Rochade” (German for long castling) and “Kurze Rochade” (German for short castling).
The pieces bear the names: Koenig, Dame, Turm, Laeufer, Springer, Bauer.
D.2.6.1 A specially constructed chessclock for the visually disabled shall be admissible. It should be able to announce the time and number of moves to the visually disabled player.
D.6.2.2 Alternatively an analogue clock with the following features may be considered:
The starting position for Chess960 must meet certain rules. White pawns are placed on the second rank as in regular chess. All remaining white pieces are placed randomly on the first rank, but with the following restrictions:
The starting position can be generated before the game either by a computer program or using dice, coin, cards, etc.
Thus, after cside castling (notated as 000 and known as queenside castling in orthodox chess), the king is on the csquare (c1 for white and c8 for black) and the rook is on the dsquare (d1 for white and d8 for black). After gside castling (notated as 00 and known as kingside castling in orthodox chess), the king is on the gsquare (g1 for white and g8 for black) and the rook is on the fsquare (f1 for white and f8 for black).
III.1 A ‘quickplay finish’ is the phase of a game when all the remaining moves must be completed in a finite time.
Example 1: According to the Tournament Regulations of an event, the time control is 2 hours for 40 moves and then 1 hour for the end of the game. The last 1 hour will be played according to the rules of the Quickplay finish. Example 2: According to the Tournament Regulations of an event, the time control is 2 hours for the whole game. It means that the whole game will be played according to the rules of the Quick play finish. 
It means that the Arbiter may make a decision to declare a game as a draw, even after a flag fall has occurred. As it may sometimes cause many conflicts between Arbiters and players, the games played according to Quickplay finish should be avoided. This requires use of digital clocks. Time controls WITH an increment is a much better way of concluding a game. 
In III.6.1.1 the player must write down the final position and his opponent must verify it.
In III.6.1.2 the player must write down the final position and submit an uptodate scoresheet. The opponent shall verify both the scoresheet and the final position.
The number after the term refers to the first time it appears in the Laws.
adjourn: 8.1. Instead of playing the game in one session it is temporarily halted and then continued at a later time.
algebraic notation: 8.1. Recording the moves using ah and 18 on the 8x8 board.
analyse: 11.3. Where one or more players make moves on a board to try to determine what is the best continuation.
appeal: 11.10. Normally a player has the right to appeal against a decision of the arbiter or organiser.
arbiter: Preface. The person(s) responsible for ensuring that the rules of a competition are followed.
arbiter’s discretion: There are approximately 39 instances in the Laws where the arbiter must use his judgement.
assistant: 8.1. A person who may help the smooth running of the competition in various ways.
attack: 3.1.A piece is said to attack an opponent’s piece if the player’s piece can make a capture on that square.
black: 2.1. 1. There are 16 darkcoloured pieces and 32 squares called black. Or 2. When capitalised, this also refers to the player of the black pieces.
blitz: B. A game where each player’s thinking time is 10 minutes or less.
board: 2.4.Short for chessboard.
Bronstein mode: 6.3.2 See delay mode.
capture: 3.1. Where a piece is moved from its square to a square occupied by an opponent’s piece, the latter is removed from the board. See also 3.7.4.1 i 3.4.7.2.In notation x.
castling: 3.8.2 A move of the king towards a rook. See the article. In notation 00 kingside castling, 000 queenside castling.
cellphone: See mobile phone.
check: 3.9. Where a king is attacked by one or more of the opponent’s pieces. In notation +.
checkmate: 1.2. Where the king is attacked and cannot parry the threat. In notation ++ or #.
chessboard: 1.1. The 8x8 grid as in 2.1.
chessclock: 6.1. A clock with two time displays connected to each other.
chess set: The 32 pieces on the chessboard.
Chess960: A variant of chess where the backrow pieces are set up in one of the 960 distinguishable possible positions
claim: 6.8. The player may make a claim to the arbiter under various circumstances.
clock: 6.1. One of the two time displays.
completed move: 6.2.1 Where a player has made his move and then pressed his clock.
contiguous area: 12.8. An area touching but not actually part of the playing venue. For example, the area set aside for spectators.
cumulative (Fischer) mode: Where a player receives an extra amount of time (often 30 seconds) prior to each move.
dead position: 5.2.2 Where neither player can mate the opponent’s king with any series of legal moves.
default time: 6.7. The specified time a player may be late without being forfeited.
delay (Bronstein) mode: 6.3.2Both players receive an allotted ‘main thinking time’. Each player also receives a ‘fixed extra time’ with every move. The countdown of the main thinking time only commences after the fixed extra time has expired. Provided the player presses his clock before the expiration of the fixed extra time, the main thinking time does not change, irrespective of the proportion of the fixed extra time used.
demonstration board: 6.13. A display of the position on the board where the pieces are moved by hand.
diagonal: 2.4.A straightline of squares of the same colour, running from one edge of the board to an adjacent edge.
disability: 6.2.6 A condition, such as a physical or mental handicap, that results in partial or complete loss of a person's ability to perform certain chess activities.
displaced: 7.4.1 to put or take pieces from their usual place. For example, a pawn from a2 to a4.5; a rook partway between d1 and e1; a piece lying on its side; a piece knocked onto the floor.
draw: 5.2. Where the game is concluded with neither side winning.
draw offer: 9.1.2 Where a player may offer a draw to the opponent. This is indicated on the scoresheet with the symbol (=).
ecigarette: device containing a liquid that is vaporised and inhaled orally to simulate the act of smoking tobacco.
en passant: 3.7.4.1See that article for an explanation. In notation e.p.
exchange: 1. 3.7.5.3 Where a pawn is promoted. Or 2.Where a player captures a piece of the same value as his own and this piece is recaptured. Or 3. Where one player has lost a rook and the other has lost a bishop or knight.
explanation: 11.9. A player is entitled to have a Law explained.
fair play: 12.2.1 Whether justice has been done has sometimes to be considered when an arbiter finds that the Laws are inadequate.
file: 2.4. A vertical column of eight squares on the chessboard.
Fischer mode: See cumulative mode.
flag: 6.1. The device that displays when a time period has expired.
flagfall: 6.1. Where the allotted time of a player has expired.
forfeit: 4.8.1. To lose the right to make a claim or move. Or 2. To lose a game because of an infringement of the Laws.
handicap: See disability.
I adjust: See j’adoube.
illegal: 3.10.1.A position or move that is impossible because of the Laws of Chess.
impairment: See disability.
increment: 6.1. An amount of time (from 2 to 60 seconds) added from the start before each move for the player. This can be in either delay or cumulative mode.
intervene: 12.7. To involve oneself in something that is happening in order to affect the outcome.
j’adoube: 4.2. Giving notice that the player wishes to adjust a piece, but does not necessarily intend to move it.
kingside: 3.8.1.The vertical half of the board on which the king stands at the start of the game.
legal move: See Article 3.10a.
made: 1.1. A move is said to have been ‘made’ when the piece has been moved to its new square, the hand has quit the piece, and the captured piece, if any, has been removed from the board.
mate: Abbreviation of checkmate. minor piece. Bishop or knight. mobile phone: 11.3.2. Cellphone.
monitor: 6.13. An electronic display of the position on the board.
move: 1.1. 1. 40 moves in 90 minutes, refers to 40 moves by each player. Or 2. having the move refers to the player’s right to play next. Or 3. White’s best move refers to the single move by White.
movecounter: 6.10.2. A device on a chessclock which may be used to record the number of times the clock has been pressed by each player.
normal means: G.5. Playing in a positive manner to try to win; or, having a position such that there is a realistic chance of winning the game other than just flagfall.
organiser. 8.3. The person responsible for the venue, dates, prize money, invitations, format of the competition and so on.
overtheboard: Introduction. The Laws cover only this type of chess, not internet, nor correspondence, and so on.
penalties: 12.3. The arbiter may apply penalties as listed in 12.9 in ascending order of severity.
piece: 2. 1. One of the 32 figurines on the board. Or 2. A queen, rook, bishop or knight. playing area: 11.2. The place where the games of a competition are played.
playing venue: 11.2. The only place to which the players have access during play.
points: 10. Normally a player scores 1 point for a win, ½ point for a draw, 0 for a loss. An alternative is 3 for a win, 1 for a draw, 0 for a loss.
press the clock: 6.2.1 The act of pushing the button or lever on a chess clock which stops the player’s clock and starts that of his opponent.
promotion: 3.7.5.3 Where a pawn reaches the eighth rank and is replaced by a new queen, rook, bishop or knight of the same colour.
queen: As inqueen a pawn, meaning to promote a pawn to a queen.
queenside: 3.8.1. The vertical half of the board on which the queen stands at the start of the game.
quickplay finish: G. The last part of a game where a player must complete an unlimited number of moves in a finite time.
rank: 2.4. A horizontal row of eight squares on the chessboard.
rapid chess: A. A game where each player’s thinking time is more than 10 minutes, but less than 60.
repetition: 5.3.1. 1. A player may claim a draw if the same position occurs three times.
2. A game is drawn if the same position occurs five times.
resigns: 5.1.2 Where a player gives up, rather than play on until mated.
rest rooms: 11.2. Toilets, also the room set aside in World Championships where the players can relax.
result: 8.7. Usually the result is 10, 01 or ½½. In exceptional circumstances both players may lose (Article 11.8), or one score ½ and the other 0. For unplayed games the scores are indicated by +/ (White wins by forfeit), /+ (Black wins by forfeit), / (Both players lose by forfeit).
regulations of an event: 6.7.1 At various points in the Laws there are options. The regulations of an event must state which have been chosen.
sealed move: E. Where a game is adjourned the player seals his next move in an envelope.
scoresheet: 8.1. A paper sheet with spaces for writing the moves. This can also be electronic.
screen: 6.13. An electronic display of the position on the board.
spectators: 11.4. People other than arbiters or players viewing the games. This includes players after their games have been concluded.
standard chess: G3. A game where each player’s thinking time is at least 60 minutes.
stalemate: 5.2.1Where the player has no legal move and his king is not in check.
square of promotion: 3.7.5.1 The squarea pawn lands on when it reached the eighth rank.
supervise: 12.2.5Inspect or control.
time control: 1. The regulation about the time the player is allotted. For example, 40 moves in 90 minutes, all the moves in 30 minutes, plus 30 seconds cumulatively from move 1. Or2. A player is said ‘to have reached the time control’, if, for example he has completed the 40 moves in less than 90 minutes.
time period: 8.6.A part of the game where the players must complete a number of moves or all the moves in a certain time.
touch move: 4.3. If a player touches a piece with the intention of moving it, he is obliged to move it.
vertical: 2.4. The 8th rank is often thought as the highest area on a chessboard. Thus each file is referred to as ‘vertical’.
white: 2.2. 1. There are 16 lightcoloured pieces and 32 squares called white. Or 2. When capitalised, this also refers to the player of the white pieces.
zero tolerance: 6.7.1. Where a player must arrive at the chessboard before the start of the session.
50move rule: 5.3.2 A player may claim a draw if the last 50 moves have been completed by each player without the movement of any pawn and without any capture.
75move rule: 9.6.2The game is drawn if the last 75 moves have been completed by each player without the movement of any pawn and without any capture.
Evolution of the FIDE Laws of Chess has given more freedom to the organisers about the regulations of a specific event. The Competition Rules enable organisers to choose options which are the best, in their opinion, for a given tournament. But greater freedom means greater responsibility.
The FIDE Laws of Chess regulate many of the specific rules, but not always. For example, in Rapid chess and Blitz, the regulations of an event shall specify if the entire event shall be played according to the Competition Rules or with some exceptions. Apart of that, is good to remind the player of such important things as the default time and the conditions when a draw can be agreed.
If the organisers forget to make these matters clear in advance, it will not be any use making an announcement at the start of a round. Players may not be present and, anyway, do not listen.
To avoid such situations, the FIDE Rules Commission has decided to prepare Guidelines for the Organisers. These are divided in three parts: what must be specified in the regulations of the event; what the RC recommends be specified; and optional rules. The RC strongly recommends to the organisers that their choice should always be exercised in conjunction with the Chief Arbiter.
According to the article 6.7.1 of the FIDE Laws of Chess, the regulations of an event shall specify a default time. If the default time is not specified, then it is zero.
According to the article A.5/B.5 of the FIDE Laws of Chess, the regulations of an event shall specify if the entire event shall be played according to the Competition Rules (all articles from 6 to the 12 of the FIDE Laws of Chess) or with some exceptions as described in the article A.4/B.4.
The regulation of the event shell specify if the game is played according to Guidelines III (Quickplay Finishes), as described in the article III.2.1. If yes, than the regulations of an event shall specify the procedure for the player having the move and less than two minutes left on his clock for a draw claim. There are two options: according to the article III.4 of the FIDE Laws of Chess, an increment extra five seconds be introduced for both players or according to the article III.5 of the FIDE Laws of Chess, a draw claim procedure shall follow. If these matters are not specified, then, for example, king and knight v king and knight can be played on until one flag falls.
According to the article 9.1.1 of the FIDE Laws of Chess, the regulations of an event may specify that players cannot offer or agree to a draw, whether in less than a specified number of moves or at all, without the consent of the arbiter. If the draw condition is not specified then, according to the article 5.3.2 of the FIDE Laws of Chess, players can offer or agree to a draw when both have made at least one move.
According to the article 11.3.2.1 of the FIDE Laws of Chess, the regulations of an event may allow to the player to have an electronic device not specifically approved by the arbiter in the playing venue, provided that this device is stored in a player’s bag and the device is completely switched off. This bag must be placed as agreed with the arbiter. Neither player is allowed are to use this bag without permission of the arbiter. If the above permission is not specified, then is forbidden to have any electronic device in the playing venue.
According to the article 10.1 of the FIDE Laws of Chess, the regulations of an event may specify a different scoring system. For example a player who wins his game, or wins by forfeit, scores three points (3), a player who draws his game scores a two points (2), a player who loses his game scores one point (1), a player who loses by default scores zero points (0). If not specified, normal scoring is used (1, ½, 0).
According to Article 11.2.4 of the FIDE Laws of Chess, the regulations of an event may specify that the opponent of the player having a move must report to the arbiter when he wishes to leave the playing area. If this is not specified, there is no obligation for the opponent to communicate his intention to leave.
According to the article 11.10 of the FIDE Laws of Chess, the regulations of an event may specify that a player cannot appeal against any decision of the arbiter, if he has signed the scoresheet. If not specified, the player may appeal even after signing the scoresheet. It is strongly recommended that an Appeal Committee should be set up in advance.
According to the article I.11 of the FIDE Laws of Chess, the regulations of an event may specify the procedure regarding elapsed time before arrival of the players. If not specified, than the player who has to reply to the sealed move shall lose all the time that elapses until he arrives, even if both players are not present initially.
Approved by the 1986 General Assembly, 2007 PB
Amended by the 1989, 1992, 1993, 1994, 1998, 2006, 2010, 2014 General Assemblies and 2011 Executive Board.
All chess competitions shall be played according to the FIDE Laws of Chess (E.I.01A). The FIDE Competition Rules shall be used in conjunction with the Laws of Chess and shall apply to all official FIDE competitions. These Rules shall also be applied to all FIDErated competitions, amended where appropriate. The organisers, competitors and arbiters involved in any competition are expected to be acquainted with these Rules before the start of the competition. In these Rules the words ‘he’, ‘him’ and ‘his’ shall be considered to include ‘she’ and ‘her’.
National Laws take precedence over FIDE Rules.
L1: Official FIDE events as defined by the FIDE Events Commission (D.IV.01.1) or FIDE World Championship and Olympiad Commission (D.I, D.II)
L2: Competitions where FIDE titles and title norms can be earned L3: FIDE Rated Competitions
L4: All other competitions
Rules that apply to specific types of competitions shall have the competition level indicated. Otherwise the rules shall apply to all levels of competitions.
Other rules hereunder may apply also to the role of the CO. He and the Chief Arbiter (see 3) must work closely together in order to ensure the smooth running of an event.
Refer to the Technical Commission Rules
Refer to the Technical Commission Rules
But, if electronic boards are used, an illegal move shall be made, before placing the kings in the centre.
“Player” in 8.1 8.3.3, includes a “team” where appropriate.
Such permission might not be granted to a player who receives conditions, or who has been given a free entry to the tournament. It is not permitted in the last round of a tournament.
Normally such ½ point byes may be given to players who cannot be present in the first and in the second round of the tournament. It is advisable not to give them in later rounds, especially in the last round, because they can affect the final standings and thus the prize distribution. 
A player is not permitted to complain directly to his opponent (E.I.01A.11.5)
of Chess or the Competition Rule. Then the CA (in consultation with the CO) shall have discretionary power to impose penalties. He should seek to maintain discipline and offer other solutions which may placate the offended parties.
The Competition Rules may include other rules due to the peculiarities of the event. The authorised photographers may take photographs without flash during the rest of the round in the playing area, only with the permission of the CA
A team competition is one where the results of individual games contribute equally to the final score of a defined group of players.
each round, to communicate to his players the pairings, to sign the protocol indicating the results in the match at the end of play.
In the regulations of a Team Tournament, details about the Team Compositions should be included. Normally the following may be applied: A fixed board order (it might be according to the ELO rating of the players; the highest rating gets no 1. Some events allow the captain full discretion, others do not permit a player to play on a board higher than a team mate who is 100 Rating points higher. The team list must be submitted before the first round at a time stipulated by the regulations. The order shall not be changed during the whole tournament. If a team has reserve(s): then for every round, each team must submit its composition (for example, if the team consists of 4 players and one reserve, the team composition may be: 1, 2, 3, 4, or 1, 2, 4, 5, or 1, 3, 4, 5, or 1, 2, 3, 5 or 2, 3, 4, 5,), provided a given deadline before the start of the round. Where the captain does not submit any composition by the deadline, its composition shall be: 1, 2, 3, 4. No player with higher number in the fixed board order is allowed to play above a player with lower number. The reserve player must always play board 4. Other permutations are not allowed. Where players play on the wrong boards, the result of the game counts for the rating, but not for the final score in the match. They will be forfeited (+/ or /+). 
A team 
 
B team 

1A 
10 
2B: correct board 

2A 
01 
4B: wrong board (it should be 3B) 
+  : corrected result 
3A 
½ 
3B: wrong board (it should be 4B) 
+  : corrected result 
5A 
01 
5B: correct board 

Example: 
Initial Match Result: 1.52.5 (valid only for ratings) 
Corrected result of the 
Match: 3.01.0 (valid for standings and future pairings).
Each Team Captain is responsible for the submission of the compositions of their team to the appropriate officer.
The Arbiters of a Team Tournament, in cooperation with the Team Captains, MUST CHECK the Teams Compositions for every round, in order to avoid incorrect board order. 
Berger Tables for RoundRobin Tournaments
Where there is an odd number of players, the highest number counts as a bye.
Rd 1: 14, 23.
Rd 2: 43, 12.
Rd 3: 24, 31.
Rd 1: 16, 25, 34.
Rd 2: 64, 53, 12.
Rd 3: 26, 31, 45.
Rd 4: 65, 14, 23.
Rd 5: 36, 42, 51.
Rd 1: 18, 27, 36, 45.
Rd 2: 85, 64, 73, 12.
Rd 3: 28, 31, 47, 56.
Rd 4: 86, 75, 14, 23.
Rd 5: 38, 42, 51, 67.
Rd 6: 87, 16, 25, 34.
Rd 7: 48, 53, 62, 71.
Rd 1: 110, 29, 38, 47, 56.
Rd 2: 106, 75, 84, 93, 12.
Rd 3: 210, 31, 49, 58, 67.
Rd 4: 107, 86, 95, 14, 23.
Rd 5: 310, 42, 51, 69, 78.
Rd 6: 108, 97, 16, 25, 34.
Rd 7: 410, 53, 62, 71, 89.
Rd 8: 109, 18, 27, 36, 45.
Rd 9: 510, 64, 73, 82, 91.
Rd 1: 
112, 211, 310, 49, 58, 67. 
Rd 2: 
127, 86, 95, 104, 113, 12. 
Rd 3: 
212, 31, 411, 510, 69, 78. 
Rd 4: 
128, 97, 106, 115, 14, 23. 
Rd 5: 
312, 42, 51, 611, 710, 89. 
Rd 6: 
129, 108, 117, 16, 25, 34. 
Rd 7: 
412, 53, 62, 71, 811, 910. 
Rd 8: 
1210, 119, 18, 27, 36, 45. 
Rd 9: 
512, 64, 73, 82, 91, 1011. 
Rd 10: 1211, 110, 29, 38, 47, 56.
Rd 11: 612, 75, 84, 93, 102, 111.
Rd 1: 
114, 213, 312, 411, 510, 69, 78. 
Rd 2: 
148, 97, 106, 115, 124, 133, 12. 
Rd 3: 
214, 31, 413, 512, 611, 710, 89. 
Rd 4: 
149, 108, 117, 126, 135, 14, 23. 
Rd 5: 
314, 42, 51, 613, 712, 811, 910. 
Rd 6: 
1410, 119, 128, 137, 16, 25, 34. 
Rd 7: 
414, 53, 62, 71, 813, 912, 1011. 
Rd 8: 
1411, 1210, 139, 18, 27, 36, 45. 
Rd 9: 
514, 64, 73, 82, 91, 1013, 1112. 
Rd 10: 1412, 1311, 110, 29. 38, 47, 56.
Rd 11: 614, 75, 84, 93, 102, 111, 1213.
Rd 12: 1413, 112, 211, 310, 49, 58, 67.
Rd 13: 714, 86, 95, 104, 113, 122, 131.
Rd 1: 
116, 215, 314, 413, 512, 611, 710, 89. 
Rd 2: 
169, 108, 117, 126, 135, 144, 153, 12. 
Rd 3: 
216, 31, 415, 514, 613, 712, 811, 910. 
Rd 4: 
1610, 119, 128, 137, 146, 155, 14, 23. 
Rd 5: 
316, 42, 51, 615, 714, 813, 912, 1011. 
Rd 6: 
1611, 1210, 139, 148, 157, 16, 25, 34. 
Rd 7: 
416, 53, 62, 71, 815, 914, 1013, 1112. 
Rd 8: 
1612, 1311, 1410, 159, 18, 27, 36, 45. 
Rd 9: 
516, 64, 73, 82, 91, 1015, 1114, 1213. 
Rd 10: 1613, 1412, 1511, 110, 29, 38, 47, 56.
Rd 11: 616, 75, 84, 93, 102, 111, 1215, 1314.
Rd 12: 1614, 1513, 112, 211, 310, 49, 58, 67.
Rd 13: 716, 86, 95, 104, 113, 122, 131, 1415.
Rd 14: 1615, 114, 213, 312, 411, 510, 69, 78.
Rd 15: 816, 97, 106, 115, 124, 133, 142, 151.
For a doubleround tournament it is recommended to reverse the order of the last two rounds be reversed of the first cycle. This is to avoid three consecutive games with the same colour
Directions for “restricted” drawing of tournament numbers:
The choice of the tiebreak system to be used in a tournament shall be decided in advance and shall be announced prior to the start of the tournament. If all tiebreaks fail, the tie shall be broken by drawing of lots. A playoff is the best system, but it is not always appropriate. For example, there may not be adequate time.
Fundamentally this is the fairest way to decide the final ranking of players having equal scores at the end of a tournament. The only problem is that there will not be enough time to organise tie‐break matches with similar playing time as to the main tournament. Therefore, tie‐break matches with very short playing times, mainly rapid or blitz are organised, and this is a different type of tournament. That’s one of the reasons why some players are not happy with playoffs. 
There shall be twogame elimination matches at the rate as in (1) (a).
In all systems the players shall be ranked in descending order of the respective system. The following list is simply in alphabetical order.
The Average Rating of Opponents (ARO) is the sum of the ratings of the opponents of a player, divided by the number of games played.
(a1) The Average Rating of Opponents Cut (AROC) is the Average Rating of Opponents, excluding one or more of the ratings of the opponents, starting from the lowestrated opponent.
The Buchholz System is the sum of the scores of each of the opponents of a player.
(b1) The Median Buchholz is the Buchholz reduced by the highest and the lowest scores of the opponents.
(b2) The Median Buchholz 2 is the Buchholz score reduced by the two highest and the two lowest scores of the opponents.
(b3) The Buchholz Cut 1 is the Buchholz score reduced by the lowest score of the opponents.
(b4) The Buchholz Cut 2 is the Buchholz score reduced by the two lowest scores of the opponents.
If all the tied players have met each other, the sum of points from these encounters is used. The player with the highest score is ranked number 1 and so on. If some but not all have played each other, the player with a score that could not be equalled by any other player (if all such games had been played) is ranked number 1 and so on.
This is the number of points achieved against all opponents who have achieved 50 % or more.
(d1) The Koya System Extended
The Koya system may be extended, step by step, to include score groups with less than 50 %, or reduced, step by step, to exclude players who scored 50 % and then higher scores.
The greater number of games played with the black pieces (unplayed games shall be counted as played with the white pieces).
(f1) SonnebornBerger for Individual Tournaments is the sum of the scores of the opponents a player has defeated and half the scores of the players with whom he has drawn.
(f2) SonnebornBerger for Team Tournaments is the sum of the products of the scores made by each opposing team and the score made against that team.
(g1) Match points in team competitions that are decided by game points. For example:
2 points for a won match where a team has scored more points than the opposing team.
1 point for a drawn match. 0 points for a lost match.
(g2) Game points in team competitions that are decided by match points. The tie is broken by determining the total number of game points scored.
The tie is broken by determining the total number of game points scored.
Note: all these scores are determined in each case after the application of the rule concerning unplayed games. 
(a5) Sum of Buchholz: the sum of the Buchholz scores of the opponents
(b1) SonnebornBerger for Individual Tournaments
(b2) SonnebornBerger for Team Tournaments A: the sum of the products of the match points made by each opposing team and the match points made against that team, or
(b3) SonnebornBerger for Team Tournaments B: the sum of the products of the match points made by each opposing team and the game points made against that team, or
(b4) SonnebornBerger for Team Tournaments C: the sum of the products of the game points made by each opposing team and the match points made against that team, or
(b5) SonnebornBerger for Team Tournaments D: the sum of the products of the game points made by each opposing team and the game points made against that team
(b6) SonnebornBerger for Team Tournaments Cut 1 A: the sum of the products of the match points made by each opposing team and the match points made against that team, excluding the opposing team who scored the lowest number of match points, or
(b7) SonnebornBerger for Team Tournaments Cut 1 B: the sum of the products of the match points made by each opposing team and the game points made against that team, excluding the opposing team who scored the lowest number of match points, or
(b8) SonnebornBerger for Team Tournaments Cut 1 C: the sum of the products of the game points made by each opposing team and the match points made against that team, excluding the opposing team who scored the lowest number of game points, or
(b9) SonnebornBerger for Team Tournaments Cut 1 D: the sum of the products of the game points made by each opposing team and the game points made against that team, excluding the opposing team who scored the lowest number of game points.
When a player has elected not to play more than two games in a tournament, his ARO or AROC shall be considered to be lower than that of any player who has completed more of the schedule.
For tiebreak purposes all unplayed games in which players are indirectly involved (results by forfeit of opponents) are considered to have been drawn.
For tiebreak purposes a player who has no opponent will be considered as having played against a virtual opponent who has the same number of points at the beginning of the round and who draws in all the following rounds. For the round itself the result by forfeit will be considered as a normal result.
This gives the formula:
Svon = SPR + (1 – SfPR) + 0.5 * (n – R)
where for player P who did not play in round R: n = number of completed rounds
Svon = score of virtual opponent after round n SPR = score of P before round R
SfPR = forfeit score of P in round R
Example 1: in Round 3 of a nineround tournament Player P did not show up. Player P’s score after 2 rounds is 1.5. The score of his virtual opponent is Svon = 1.5 + (1 – 0) + 0.5 * (3 – 3) = 2.5 after round 3
Svon = 1.5 + (1 – 0) + 0.5 * (9 – 3) = 5.5 at the end of the tournament
Example 2: in Round 6 of a nineround tournament player P’s opponent does not show up.
Player P’s score after 5 rounds is 3.5. The score of his virtual opponent is: Svon = 3.5 + (1 – 1) + 0.5 * (6 – 6) = 3.5 after round 6
Svon = 3.5 + (1 – 1) + 0.5 * (9 – 6) = 5.0 at the end of the tournament
For different types of tournaments the TieBreak Rules are as listed below and are recommended to be applied in the listed order.
Direct encounter
The greater number of wins SonnebornBerger
Koya System
Remark: Don’t use Buchholz systems for Round Robin tournaments 
Match points (if ranking is decided by game points), or Game points (if ranking is decided by match points) Direct encounter
SonnebornBerger
Direct encounter
The greater number of wins
The greater number of games with Black (unplayed games shall be counted as played with White)
Buchholz Cut 1 Buchholz SonnebornBerger
Direct encounter
The greater number of wins
The greater number of games with Black (unplayed games shall be counted as played with White)
AROC
Buchholz Cut 1 Buchholz SonnebornBerger
Match points (if ranking is decided by game points), or Game points (if ranking is decided by match points) Direct encounter
Buchholz Cut 1 Buchholz SonnebornBerger

player "A" 
virtual opponent 
points before the round 
2 
2 
result of the round 
1 
0 
points after the round 
3 
2 
points for the subsequent rounds 
? 
3 
points at the end of the tournament 
? 
5 

player "A" 
virtual opponent 
points before the round 
2 
2 
result of the round 
1 
0 
points after the round 
3 
2 
points for the subsequent rounds 
? 
3 
points at the end of the tournament 
? 
5 

player "A" 
virtual opponent 
points before the round 
2 
2 
result of the round 
0 
1 
points after the round 
2 
3 
points for the subsequent rounds 
? 
3 
points at the end of the tournament 
? 
6 
If the tie‐break system is not fixed in the existing tournament regulations, it is up to the organiser to decide the system. The type of the tournament has to be taken in account (Round Robin, Swiss System, Team Tournament, and so on) and the structure of the expected participants (such as juniors, seniors, rated or unrated players).
Whatever system used, it must be announced by the organiser in advance or by the Chief Arbiter before the start of the first round. 
If two or more players finish a tournament with equal points the organizers have three possibilities to award money prizes:
In Hort system 50% of the prize money is given according the tie‐break ranking. The second half of the prize money of all the players, having the same number of points at the end, is added together and shared equally.

system a) 
system b) 
system c) 


A ‐ 
5.000 € 
10.000 € 
5.000 + 
2.500 = 
7.500 € 
B ‐ 
5.000 € 
5.000 € 
2.500 + 
2.500 = 
5.000 € 
C ‐ 
5.000 € 
3.000 € 
1.500 + 
2.500 = 
4.000 € 
D ‐ 
5.000 € 
2.000 € 
1.000 + 
2.500 = 
3.500 € 
Example: The prizes in the tournament are: 1st place 10.000 € 2nd place 5.000 € 3rd place 3.000€ 4th place 2.000€ Players A, B, C and D finish a tournament with 8 points each. The Buchholz points are: A has 58 Buchholz points B has 57 Buchholz points C has 56 Buchholz points D has 54 Buchholz points. The money prizes for A, B, C and D ‐ depending on the system used ‐ will be: 
The total is €20,000€ whatever system is used. 
Organizers have to decide in advance and to inform the players before the start of the tournament which system will be used for calculation of money prizes.
Additionally, in systems a) and c) the organizers have to decide and to inform the participants how many players will have the right to be awarded with money prizes in case of equal points after the last round.
If it is announced to give 10 money prizes and the final ranking is: players ranked 1 to 4 have 8 points players ranked 5 to 9 have 7.5 points players ranked 10 to 20 have 7 points. In such a case it is not wise to share the money for rank 10 between 11 players. To avoid such a problem it should be announced in advance that money prizes are equally shared or given by Hort system to the players ranked on 1 to 10. 
The distribution of the prize money is better done by two people working independently. This might be the Treasurer and Chief Arbiter. 
name  rtg  1  2  3  4  5  6  7  8  9  10  11  12  points  SB  Koya  Rp 
Alexander  2269  *  1  1  ½  0  1  1  ½  1  ½  ½  1  8  42  3½  2414 
Joseph  2171  0  *  ½  1  1  0  ½  1  1  1  1  0  7  36,75  2½  2350 
Robert  2276  0  ½  *  ½  0  1  0  1  1  1  ½  1  6½  31,75  2  2304 
Walter  2290  ½  0  ½  *  1  1  ½  ½  ½  ½  ½  ½  6  32,5  3  2273 
Peter  2273  1  0  1  0  *  ½  ½  0  1  ½  ½  1  6  32  2½  2275 
Olaf  2299  0  1  0  0  ½  *  1  1  0  1  1  ½  6  30,25  1½  2273 
Mark  2281  0  ½  1  ½  ½  0  *  ½  ½  0  ½  1  5  25,75  2½  2202 
Ivan  2333  ½  0  0  ½  1  0  ½  *  ½  0  1  1  5  25  2  2198 
Sandor  2233  0  0  0  ½  0  1  ½  ½  *  ½  1  1  5  23,25  1½  2207 
Martin  2227  ½  0  0  ½  ½  0  1  1  ½  *  0  ½  4½  23,75  1½  2178 
Frederik  2340  ½  0  ½  ½  ½  0  ½  0  0  1  *  1  4½  22,75  2  2168 
Valery  1910  0  1  0  ½  0  ½  0  0  0  ½  0  *  2½  15,25  2  2061 
Comparison of several tie‐break criteria in a swiss tournament
using the final results of the European Individual Championship 2011 in Aix‐les‐ Bains, France:
Rk  title  name  rtg  fed  pt  Rp‐2  Rp  BH  m BH  SB 
1  GM  Potkin Vladimir  2653  RUS  8½  2849  2822  78  63½  59,25 
2  GM  Wojtaszek Radoslaw  2711  POL  8½  2826  2812  77  63  58,5 
3  GM  Polgar Judit  2686  HUN  8½  2799  2781  77  63½  58,25 
4  GM  Moiseenko Alexander  2673  UKR  8½  2755  2790  74½  62  56,5 
5  GM  Vallejo Pons Francisco  2707  ESP  8  2819  2764  80  66½  57,75 
6  GM  Ragger Markus  2614  AUT  8  2783  2768  76  62½  54,25 
7  GM  Feller Sebastien  2657  FRA  8  2766  2763  70½  58½  49 
8  GM  Svidler Peter  2730  RUS  8  2751  2757  76½  62½  54,75 
9  GM  Mamedov Rauf  2667  AZE  8  2751  2754  74  61  52,25 
10  GM  Vitiugov Nikita  2720  RUS  8  2741  2744  76½  63  54,5 
11  GM  Zhigalko Sergei  2680  BLR  8  2732  2731  72  59½  50 
12  GM  Jakovenko Dmitry  2718  RUS  8  2719  2704  72½  60  53 
13  GM  Korobov Anton  2647  UKR  8  2697  2740  75  61½  53,5 
14  GM  Inarkiev Ernesto  2674  RUS  8  2695  2735  72½  60  51,5 
15  GM  Postny Evgeny  2585  ISR  8  2633  2676  64  52  44,75 
16  GM  Azarov Sergei  2615  BLR  7½  2776  2723  75  62½  47,5 
17  GM  Khairullin Ildar  2634  RUS  7½  2771  2720  74½  61½  49 
18  GM  Kobalia Mikhail  2672  RUS  7½  2754  2716  70½  57  45,5 
19  GM  Zherebukh Yaroslav  2560  UKR  7½  2739  2712  71½  59  45,5 
20  GM  Guliyev Namig  2522  AZE  7½  2739  2652  71  59½  45,5 
21  GM  Riazantsev Alexander  2679  RUS  7½  2728  2687  72½  60  48,75 
22  GM  Iordachescu Viorel  2626  MDA  7½  2725  2716  76  62  50,25 
23  GM  Lupulescu Constantin  2626  ROU  7½  2722  2677  71  58  46 
24  GM  Mcshane Luke J  2683  ENG  7½  2718  2684  72½  59  47 
25  GM  Fridman Daniel  2661  GER  7½  2717  2684  69  56½  45,25 
26  GM  Motylev Alexander  2677  RUS  7½  2716  2710  71  59  47,5 
27  GM  Ivanisevic Ivan  2617  SRB  7½  2712  2704  71  58½  47 
28  GM  Jobava Baadur  2707  GEO  7½  2711  2656  71½  58  47,5 
29  GM  Parligras Mircea‐ Emilian  2598  ROU  7½  2709  2735  78½  65  50,75 
30  GM  Romanov Evgeny  2624  RUS  7½  2709  2668  68½  55½  43,75 
31  GM  Esen Baris  2528  TUR  7½  2707  2669  73  61  47,25 
32  GM  Nielsen Peter Heine  2670  DEN  7½  2703  2707  67½  55  45,5 
33  GM  Cheparinov Ivan  2664  BUL  7½  2698  2693  75  62  49,75 
34  GM  Gustafsson Jan  2647  GER  7½  2687  2687  67  55  45 
35  GM  Kulaots Kaido  2601  EST  7½  2669  2633  67½  54½  44 
36  GM  Smirin Ilia  2658  ISR  7½  2668  2675  69  56½  47,25 
37  GM  Saric Ivan  2626  CRO  7½  2651  2692  72½  58½  47 
38  GM  Pashikian Arman  2642  ARM  7½  2649  2640  68  55½  46 
39  GM  Edouard Romain  2600  FRA  7½  2634  2602  66  52½  42,5 
40  GM  Bologan Viktor  2671  MDA  7½  2629  2673  68½  56  45,75 
Rating prizes are another type of prize. Perhaps the best achievement by a player rated 23002399 and another 22002299. Do not fall into the trap of writing U2400, U2300. In that case the U2300 player might get both prizes. A good way of awarding these is not just on score, or Tournament Performance Rating. Best Improvement in Rating as measured by WWe. W is the score achieved, We is the expected score against the average strength of the opponents. This has the advantage that ties are almost unknown. It is possible, in a Swiss, for a player with a lower score to get the greater achievement. 
Chess pieces should be made of wood, plastic or an imitation of these materials.
The size of the pieces should be proportionate to their height and form; other elements such as stability, aesthetic considerations etc., may also be taken into account. The weight of the pieces should be suitable for comfortable moving and stability.
Recommended height of the pieces is as follows: King  9.5 cm, Queen  8.5 cm, Bishop
 7 cm, Knight  6 cm, Rook  5.5 cm and Pawn  5 cm. The diameter of the piece's base should measure 40¬50% of its height. These dimensions may differ up to 10% from the above recommendation, but the order (e.g. King is higher than Queen etc.) must be kept.
Recommended for use in FIDE competitions are pieces of Staunton style. The pieces should be shaped so as to be clearly distinguishable from one another. In particular the top of the King should distinctly differ from that of the Queen. The top of the Bishop may bear a notch or be of a special colour clearly distinguishing it from that of the Pawn.
Examples of chess pieces:
Original Staunton chess pieces, left to right: pawn, rook, knight, bishop, queen, and king
World Chess set approved by FIDE for the 2013 Candidate Tournament in London
The "black" pieces should be brown or black, or of other dark shades of these colours. The "white" pieces may be white or cream, or of other light colours. The natural colour of wood (walnut, maple, etc.) may also be used for this purpose. The pieces should not be shiny and should be pleasing to the eye.
The initial position of the pieces – see FIDE Laws of Chess art.2
For the World or Continental top level competitions wooden boards should be used. For other FIDE registered tournaments boards made of wood, plastic or card are recommended. In all cases boards should be rigid. The board may also be of stone or marble with appropriate light and dark colours, provided the Chess Organiser and Chief Arbiter find it acceptable. Natural wood with sufficient contrast, such as birch, maple or European ash against walnut, teak, beech, etc., may also be used for boards, which must have a dull or neutral finish, never shiny. Combination of colours such as brown, green, or very light tan and white, cream, offwhite ivory, buff, etc., may be used for the chess squares in addition to natural colours.
The side of the square should measure 5 to 6 cm. Referring to 2.2 the side of a square should be at least twice the diameter of a pawn's base (that means four pawns on one square). A comfortable table of suitable height may be fitted in with a chessboard. If the table and the board are separate from one another, the latter must be fastened and thus prevented from moving during play.
For all official FIDE tournaments, the length of the table is 110 cm (with 15% tolerance). The width is 85 cm (for each player at least 15 cm). The height of the table is 74 cm. The chairs should be comfortable for the players. Special dispensation should be given for children's events. Any noise when moving the chairs must be avoided.
For the FIDE World or Continental Championships and Olympiads electronic chess clocks must be used. For other FIDE registered tournaments organisers are recommended to use also electronic chess clocks.
If mechanical chess clocks are used, they should have a device (a "flag") signalling precisely when the hour hand indicates the end of each hour. The flag must be arranged so that its fall can be clearly seen, helping the arbiters and players to check time. The clock should not be reflective, as that may make it difficult to see. It should run as silently as possible in order not to disturb the players during play.
The same type of clock should be used throughout the tournament.
For a highlevel tournaments The organizer should have the possibility (the device) to adjust the light in the hall  quality of lighting covering a larger area to the same level of flux requires a greater number of lumens.
Some definitions and recommendations regarding sizes L : Length of the table.
L = 110 cm, tolerances: +20 cm, 10 cm W : Width of the table.
W = 85 cm, tolerances: +5 cm, 5 cm. S : Horizontal space between table rows.
S = 3m, tolerances: +1.5 m, 0.5 m. R : Vertical space between table rows. R = 3m, tolerances: +1.5 m, 0.5 m.
Diagram B
Basic tournament hall placement styles
Dual Row For large events (open tournaments, youth champ. etc) (an arbiter may check two table in a same time) 
Single Row Preferable style for individual competitions 
Multi Row For team competitions (should be avoided for individual events as much as possible) 
These guidelines may also be useful indications for ordinary school chess which is often described as "noncompetitive" (games are usually played without clocks and not usually notated) in cases where the organizer is trying to introduce players to the world of "competitive" chess.
Note: Some children run to their parents very fast and forget to report the result. Sometimes they give false results when coming to the arbiters place or they change the colour. After that the arbiter has less time to intervene or check who won the game.
To establish the pairings for a chess tournament the following systems may be used:
In a Round Robin Tournament all the players play each other. Therefore the number of rounds is the number of participants minus one, in case of an even number of players. If there is an odd number of participants, the number of rounds is equal to the number of players.
Usually the Berger Tables are used to establish the pairings and the colours of each round.
If the number of players is odd, then the player who was supposed to play against the last player has a free day in every round.
The best system for players is a Double Round Robin Tournament, because in such a system all players have to play two games against each opponent, one with white pieces and another one with black pieces. But mainly there is not time enough for it and other systems have to be used.
An example of a cross table of the final ranking of a Round Robin Tournament:
2009 China (Nanjing) Pearl Spring Chess Tournament 

Final Ranking crosstable after 10 Rounds  
Rk. 

Name  Rtg  FED  1  2  3  4  5  6  Pts.  TB1  TB2  TB3 
1  GM  CARLSEN MAGNUS  2772  NOR  ***  1 ½  ½ 1  1 1  1 ½  1 ½  8  6  0  35 
2  GM  TOPALOV VESELIN  2813  BUL  0 ½  ***  ½ ½  ½ 1  ½ ½  ½ 1  6  2  0  24,5 
3  GM  WANG YUE  2736  CHN  ½ 0  ½ ½  ***  ½ ½  ½ ½  ½ ½  5  0  0  21,5 
4  GM  JAKOVENKO DMITRY  2742  RUS  0 0  ½ 0  ½ ½  ***  ½ 1  ½ ½  4  1  0  17,3 
5  GM  RADJABOV TEIMOUR  2757  AZE  0 ½  ½ ½  ½ ½  ½ 0  ***  ½ ½  4  0  1  20 
6  GM  LEKO PETER  2762  HUN  0 ½  ½ 0  ½ ½  ½ ½  ½ ½  ***  4  0  1  19,3 
In FIDE, there are five different Swiss systems to be used for pairings:
It is the usual Swiss system for open tournaments well known by players and organizers, and will be described in detail later (see paragraph 8: “Annotated rules for the Dutch Swiss System”);
The pairings are made from to top score group down before the middle group, then from the bottom score group to the middle group and finally the middle score group;
The objective of this system is to equalize the rating average (ARO) of all players. Therefore, in a score group, the white‐seeking players are sorted according to their ARO, the black‐seeking players according to their rating. Then, the white‐seeking player with the highest ARO is paired against the black‐seeking player with the lowest rating;
The players in a score group are sorted according to their Sonneborn‐Berger points (then Buchholz, then Median) and then the top ranked player is paired against the last ranked player, the second ranked player against the last but one, and so on, with floaters coming from the middle.
It was used to pair teams in the Olympiad before 2006;
This system is similar to the Lim system for individual tournaments with only small amendments (reduced requirements for colour preference and floating) for team pairings.
An example of a cross table of the final ranking of a Swiss Tournament:
Final ranking 








































Rank  SNo. 

Name  IRtg  FED  1.Rd.  2.Rd.  3.Rd.  4.Rd.  5.Rd.  6.Rd.  7.Rd.  8.Rd.  9.Rd.  Pts  Res.  BH.  BH.  BL  Vict  Rtg+/  Ra  Rp  
1  6  IM  J. Thybo  2466  DEN  30  b  1  6  w  0  54  b  1  18  w  1  8  b  1  24  w  1  4  b  ½  7  b  ½  13  w  1  7  0  46½  50½  5  6  13,6  2371  2591 
2  9  IM  J. Plenca  2440  CRO  33  w  1  37  b  ½  27  w  ½  59  b  1  15  w  ½  25  b  ½  53  w  1  16  b  1  7  w  1  7  0  40½  44  4  5  13,0  2336  2556 
3  2  GM  B. Deac  2559  ROU  11  b  ½  38  w  1  22  b  1  16  b  1  9  w  ½  12  b  1  7  w  ½  5  w  ½  4  b  ½  6½  0  48  52½  5  4  1,2  2387  2551 
4  7  IM  L. Livaic  2461  CRO  66  w  1  32  b  1  21  w  1  14  b  ½  7  w  0  29  b  1  1  w  ½  24  b  1  3  w  ½  6½  0  45  48½  4  5  14,4  2415  2581 
5  3  IM  M. Santos Ruiz  2505  ESP  46  w  1  24  b  ½  17  w  ½  37  b  1  10  w  1  16  b  ½  14  w  ½  3  b  ½  20  w  1  6½  0  44  48½  4  4  3,2  2357  2523 
6  33 

J. Radovic  2330  SRB  68  w  1  1  b  1  14  w  0  53  b  1  29  w  ½  26  b  ½  12  w  1  10  b  ½  15  w  1  6½  0  44  47½  4  5  30,1  2377  2543 
7  1  IM  H. Martirosyan  2570  ARM  25  w  0  60  b  1  46  w  1  48  b  1  4  b  1  9  w  1  3  b  ½  1  w  ½  2  b  0  6  0  47  51  5  5  5,1  2390  2510 
8  39 

I. Akhvlediani  2303  GEO  77  w  1  12  b  1  9  w  ½  15  b  ½  1  w  0  23  b  ½  48  w  1  31  b  1  10  w  ½  6  0  45½  48  4  4  25,3  2343  2464 
9  8  IM  N. Morozov  2461  MDA  62  b  1  56  w  1  8  b  ½  20  w  1  3  b  ½  7  b  0  15  w  ½  18  w  ½  25  b  1  6  0  44  48  5  4  5,9  2385  2510 
10  17  FM  B. Haldorsen  2397  NOR  57  w  1  55  b  ½  25  w  ½  27  b  1  5  b  0  38  w  1  11  b  1  6  w  ½  8  b  ½  6  0  43½  47½  5  4  4,1  2307  2432 
11  45  FM  M. Askerov  2281  RUS  3  w  ½  17  b  0  73  w  1  43  b  1  21  w  1  19  b  ½  10  w  0  53  b  1  26  w  1  6  0  42½  45½  4  5  32,2  2373  2498 
12  12  FM  I. Janik  2418  POL  79  b  1  8  w  0  40  b  1  41  w  1  28  b  1  3  w  0  6  b  0  54  w  1  24  w  1  6  0  42  44½  4  6  3,9  2326  2451 
13  11  IM  M. Costachi  2418  ROU  72  w  ½  40  b  ½  66  w  1  21  b  ½  37  w  1  15  b  ½  17  w  1  14  b  1  1  b  0  6  0  41½  44½  5  4  6,8  2350  2475 
14  4  FM  A. Sorokin  2486  RUS  67  b  1  44  w  1  6  b  1  4  w  ½  24  b  0  20  w  1  5  b  ½  13  w  0  17  b  ½  5½  0  46  49½  5  4  3,2  2375  2455 
15  18  S. Tica  2389  CRO  64  b  1  54  w  1  28  b  ½  8  w  ½  2  b  ½  13  w  ½  9  b  ½  39  w  1  6  b  0  5½  0  45  49  5  3  1,7  2318  2398  
16  63 

K. Yayloyan  2142  ARM  53  w  1  59  b  1  41  b  1  3  w  0  19  b  1  5  w  ½  24  b  ½  2  w  0  18  b  ½  5½  0  44½  48  5  4  61,6  2410  2492 
17  32  FM  D. Tokranovs  2334  LAT  49  b  ½  11  w  1  5  b  ½  19  w  0  79  b  1  28  w  1  13  b  0  38  w  1  14  w  ½  5½  0  43½  46  4  4  10,1  2321  2401 
18  23  FM  J. Haug  2379  NOR  60  w  ½  45  b  ½  39  w  1  1  b  0  66  w  1  33  b  1  26  w  ½  9  b  ½  16  w  ½  5½  0  41½  45  4  3  0,9  2283  2363 
19  15  FM  M. Warmerdam  2399  NED  39  w  ½  72  b  ½  23  w  1  17  b  1  16  w  0  11  w  ½  56  b  1  25  w  ½  22  b  ½  5½  0  41½  44½  4  3  4,8  2275  2355 
20  21  IM  A. Sousa  2386  POR  63  w  1  25  b  ½  55  w  1  9  b  0  57  w  1  14  b  0  27  w  1  34  w  1  5  b  0  5½  0  41  45  4  5  6,2  2351  2431 
21  30  FM  R. Lagunow  2357  GER  75  b  1  76  w  1  4  b  0  13  w  ½  11  b  0  30  w  ½  62  b  1  28  w  ½  39  b  1  5½  0  39½  42  5  4  2,2  2247  2327 
22  28  IM  A. Perez Garcia  2361  ESP  78  b  1  58  w  ½  3  w  0  57  b  0  63  w  1  45  b  1  23  w  1  26  b  ½  19  w  ½  5½  0  39  42  4  4  1,1  2263  2343 
23  56  FM  M. Jogstad  2259  SWE  31  b  0  80  w  1  19  b  0  76  w  1  35  b  1  8  w  ½  22  b  0  47  w  1  42  b  1  5½  0  38½  40½  5  5  15,1  2262  2329 
24  36 

G. Kouskoutis  2314  GRE  70  b  1  5  w  ½  34  b  1  26  b  1  14  w  1  1  b  0  16  w  ½  4  w  0  12  b  0  5  0  47  50½  5  4  14,7  2368  2409 
25  44  FM  T. Lazov  2289  MKD  7  b  1  20  w  ½  10  b  ½  42  w  ½  44  b  1  2  w  ½  31  w  ½  19  b  ½  9  w  0  5  0  45½  50  4  2  25,7  2423  2466 
26  10  FM  J. Vykouk  2440  CZE  38  b  ½  49  w  1  47  b  1  24  w  0  56  b  1  6  w  ½  18  b  ½  22  w  ½  11  b  0  5  0  42  46  5  3  12,2  2287  2330 
27  40  FM  I. Lopez Mulet  2302  ESP  80  b  1  31  w  ½  2  b  ½  10  w  0  36  b  1  42  w  ½  20  b  0  68  w  1  34  b  ½  5  0  41½  43½  5  3  6,3  2294  2319 
28  57  FM  V. Sevgi  2240  TUR  50  w  1  43  b  1  15  w  ½  29  b  ½  12  w  0  17  b  0  46  w  1  21  b  ½  32  w  ½  5  0  41  45½  4  3  26,6  2362  2405 
29  16  FM  R. Haria  2398  ENG  36  b  1  47  w  ½  58  b  1  28  w  ½  6  b  ½  4  w  0  39  b  0  40  w  ½  54  b  1  5  0  40½  4½  5  3  9,4  2276  2319 
30  49  FM  C. Meunier  2270  FRA  1  w  0  68  b  1  50  w  1  31  b  0  48  w  ½  21  b  ½  41  b  1  42  w  35  b  ½  5  0  40½  44  5  3  18,9  2349  2392  
31  13 

S. Drygalov  2415  RUS  23  w  1  27  b  ½  37  w  0  30  w  1  55  b  ½  62  w  1  25  b  ½  8  w  0  33  b  ½  5  0  40  44  4  3  10,8  2285  2328 
32  37 

K. Nowak  2314  POL  86    +  4  w  0  81  b  ½  36  w  ½  39  b  ½  58  w  1  42  b  ½  35  w  ½  28  b  ½  5  0  38½  40  4  2  8,4  2244  2192 
33  52  FM  K. Karayev  2266  AZE  2  b  0  74  w  ½  49  b  ½  75  w  1  41  b  1  18  w  0  47  b  ½  56  w  1  31  w  ½  5  0  37½  40  4  3  4,8  2244  2287 
34  5  IM  V. Dragnev  2483  AUT  40  w  ½  61  b  1  24  w  0  55  b  0  72  w  1  37  b  1  44  w  1  20  b  0  27  w  ½  5  0  37  40  4  4  17,0  2287  2330 
35  29 

V. Lukiyanchuk  2358  UKR  61  w  0  52  b  1  57  w  ½  45  b  ½  23  w  0  60  b  1  66  w  1  32  b  ½  30  w  ½  5  0  36  39½  4  3  14,7  2197  2240 
36  59  FM  C. Patrascu  2227  ROU  29  w  0  83  b  1  43  w  ½  32  b  ½  27  w  0  59  b  1  50  w  ½  44  b  ½  53  w  1  5  0  35½  37  4  3  20,1  2302  2329 
37  38  FM  K. Koziol  2313  POL  74  b  1  2  w  ½  31  b  1  5  w  0  13  b  0  34  w  0  58  b  ½  62  w  1  41  b  ½  4½  0  42  44½  5  3  3,0  2326  2321 
38  53 

M. Friedland  2264  ISR  26  w  ½  3  b  0  69  w  1  51  w  1  42  b  ½  10  b  0  55  w  1  17  b  0  43  w  ½  4½  0  40½  44  4  3  14,2  2357  2357 
39  58 

J. Thorgeirsson  2232  ISL  19  b  ½  42  w  ½  18  b  0  82  w  1  32  w  ½  51  b  1  29  w  1  15  b  0  21  w  0  4½  0  40½  42  4  3  14,1  2314  2285 
40  48  FM  S. Tifferet  2273  ISR  34  b  ½  13  w  ½  12  w  0  73  b  1  53  w  0  61  b  1  43  w  ½  29  b  ½  48  w  ½  4½  0  39  42  4  2  10,3  2340  2340 
The Scheveningen system is mainly used for teams. In such a team competition, each player of one team meets each player of the opposing team. The number of rounds therefore is equal to the number of players in a team. Scheveningen system, the players of first half of one team meet all players of the first half of the opposing team and players of the second half of one team play against players of the second half of the other team. Example: Team A and B have eight players each. A1, A2, A3 and A4 play versus B1, B2, B3 and B4. At the same time A5, A6, A7 and A8 play versus B5, B6, B7 and B8. Finally four rounds are necessary
In a Semi‐
Standard Tables
Match on 2 Boards
Round 1 A1B1 A2B2
Round 2 B2A1 B1A2
Match on 3 Boards
Round 1 A1B1 A2B2 B3A3
Round 2 B2A1 A2B3 B1A3
Round 3 A1B3 B1A2 A3B2
Match on 4 Boards
Round 1 A1B1 A2B2 B3A3 B4A4
Round 2 B2A1 B1A2 A3B4 A4B3
Round 3 A1B3 A2B4 B1A3 B2A4
Round 4 B4A1 B3A2 A3B2 A4B1
Match on 5 Boards
Round 1 A1B1 A2B2 A3B3 B4A4 B5A5
Round 2 B2A1 B3A2 B4A3 A4B5 A5B1
Round 3 A1B3 A2B4 B5A3 B1A4 A5B2
Round 4 B4A1 B5A2 A3B1 A4B2 B3A5
Round 5 A1B5 B1A2 B2A3 A4B3 A5B4
Match on 6 Boards
Round 1 B1A1 B5A2 A3B4 A4B2 A5B3 B6A6
Round 2 B2A1 A2B1 B3A3 B4A4 A5B6 A6B5
Round 3 A1B3 A2B2 B1A3 B6A4 B5A5 A6B4
Round 4 A1B4 B6A2 A3B5 A4B1 B2A5 B3A6
Round 5 B5A1 B4A2 A3B6 B3A4 A5B1 A6B2
Round 6 A1B6 A2B3 B2A3 A4B5 B4A5 B1A6
Match on 7 Boards
Round 1 A1B1 A2B2 A3B3 A4B4 B5A5 B6A6 B7A7
Round 2 B2A1 B3A2 B4A3 A4B5 A5B6 A6B7 B1A7
Round 3 A1B3 A2B4 A3B5 B6A4 B7A5 B1A6 A7B2
Round 4 B4A1 B5A2 A3B6 A4B7 A5B1 B2A6 B3A7
Round 5 A1B5 A2B6 B7A3 B1A4 B2A5 A6B3 A7B4
Round 6 B6A1 A2B7 A3B1 A4B2 B3A5 B4A6 B5A7
Round 7 A1B7 B1A2 B2A3 B3A4 A5B4 A6B5 A7B6
Match on 8 Boards
Round 1 A1B1 A2B2 A3B3 A4B4 B5A5 B6A6 B7A7 B8A8
Round 2 B2A1 B3A2 B4A3 B1A4 A5B6 A6B7 A7B8 A8B5
Round 3 A1B3 A2B4 A3B1 A4B2 B7A5 B8A6 B5A7 B6A8
Round 4 B4A1 B1A2 B2A3 B3A4 A5B8 A6B5 A7B6 A8B7
Round 5 A1B5 A2B6 A3B7 A4B8 B1A5 B2A6 B3A7 B4A8
Round 6 B6A1 B7A2 B8A3 B5A4 A5B2 A6B3 A7B4 A8B1
Round 7 A1B7 A2B8 A3B5 A4B6 B3A5 B4A6 B1A7 B2A8
Round 8 B8A1 B5A2 B6A3 B7A4 A5B4 A6B1 A7B2 A8B3
Match on 9 Boards
Round 1 A1B1 A2B2 A3B3 A4B4 A5B5 B6A6 B7A7 B8A8 B9A9
Round 2 B2A1 B3A2 B4A3 B5A4 A5B6 A6B7 A7B8 A8B9 B1A9
Round 3 A1B3 A2B4 A3B5 A4B6 B7A5 B8A6 B9A7 B1A8 A9B2
Round 4 B4A1 B5A2 B6A3 A4B7 A5B8 A6B9 A7B1 B2A8 B3A9
Round 5 A1B5 A2B6 A3B7 B8A4 B9A5 B1A6 B2A7 A8B3 A9B4
Round 6 B6A1 B7A2 A3B8 A4B9 A5B1 A6B2 B3A7 B4A8 B5A9
Round 7 A1B7 A2B8 B9A3 B1A4 B2A5 B3A6 A7B4 A8B5 A9B6
Round 8 B8A1 A2B9 A3B1 A4B2 A5B3 B4A6 B5A7 B6A8 B7A9
Round 9 A1B9 B1A2 B2A3 B3A4 B4A5 A6B5 A7B6 A8B7 A9B8
Match on 10 Boards
Round 1 A1B1 A2B2 A3B8 B9A4 B5A5 A6B3 A7B4 B6A8 B7A9 B10A10
Round 2 B2A1 B1A2 B4A3 A4B7 A5B10 B8A6 B3A7 A8B5 A9B6 A10B9
Round 3 A1B3 A2B8 A3B1 B2A4 B6A5 A6B4 A7B10 B7A8 B9A9 B5A10
Round 4 B4A1 B3A2 A3B9 B1A4 A5B7 B10A6 A7B6 B8A8 A9B5 A10B2
Round 5 A1B5 A2B4 B2A3 A4B3 B1A5 B9A6 B7A7 A8B10 B8A9 A10B6
Round 6 B6A1 A2B7 B5A3 B4A4 A5B8 A6B1 A7B9 A8B2 B10A9 B3A10
Round 7 A1B7 B5A2 A3B10 A4B6 B4A5 B2A6 B1A7 B9A8 A9B3 A10B8
Round 8 B8A1 B6A2 B3A3 B10A4 A5B9 A6B5 A7B2 A8B1 A9B4 B7A10
Round 9 A1B9 A2B10 A3B6 A4B8 B2A5 A6B7 B5A7 B3A8 B1A9 B4A10
Round 10 B10A1 B9A2 B7A3 A4B5 A5B3 B6A6 B8A7 A8B4 A9B2 A10B1
When using a Round Robin system for three teams it is necessary to organize three rounds and in each round one team is without an opponent. Skalitzka system gives a possibility to find a ranking for three teams by playing only two rounds and to avoid that a team has no opponent. Each team has to be composed of an even number of players, all of them ranked in a fixed board order. Before the pairing is made one team is marked by capital letters, then second one by small letters and the third one by figures. Then the pairings are:
round 1  round 2 
A ‐ a  1 ‐ A 
b ‐ 1  a ‐ 2 
2 ‐ B  B ‐ b 
C ‐ c  3 ‐ C 
d ‐ 3  c ‐ 4 
4 ‐ D  D ‐ d 
E ‐ e  5 ‐ E 
f ‐ 5  e ‐ 6 
6 ‐ F  F ‐ f 
Most matches between two players are played over a restricted number of games. Matches may be rated by FIDE if they are registered in advance with FIDE and if both players are rated before the match. After one player has won the match all subsequent games are not rated.
The main advantage of a knock‐out system is to create a big final match. The whole schedule is known in advance. Mostly a knock‐out match consists of two games. As it is necessary to have a clear winner of each round another day for the tie‐break games has to be foreseen. Such tie‐ break games usually are organized with two rapid games followed by two or four blitz games. If still the tie is unbroken, one final “sudden death match” shall be played. The playing time should be 5 minutes for White and 4 minutes for Black, or a similar playing time. White has to win the game, for Black a draw is sufficient to win the match. See chapter “Tie‐break Systems”.
Approved by the 1987 General Assembly
In certain cases, regulations state that the drawing of lots should be carried out in such a way that players of the same federation do not meet in the last three rounds, if possible.
This may be done by using the Varma tables, reproduced below, which can be modified for tournaments of from 10 to 24 players.
D) as indicated below, will not meet in the last three rounds: (6, 7, 8, 9, 15, 16, 17, 18)
(1, 2, 3, 11, 12, 13, 14)
(5, 10, 19)
(4, 20)
The arbiter shall prepare beforehand, unmarked envelopes each containing one of the above numbers. The envelopes containing a group of numbers are then placed in unmarked larger envelopes.
1. (3, 4, 8);
2. (5, 7, 9);
3. (1, 6);
4. (2, 10)
11/12 players
1. (4, 5, 9, 10);
2. (2, 6, 7);
3. (1, 8, 12);
4. (3, 11)
13/14 players
1. (4, 5, 6, 11, 12);
2. (1, 2, 8, 9);
3. (7, 10, 13);
4. (3, 14)
15/16 players
1. (5, 6, 7, 12, 13, 14);
2. (1, 2, 3, 9, 10);
3. (8, 11, 15);
4. (4, 16)
17/18 players
1. (5, 6, 7, 8, 14, 15, 16);
2. (1, 2, 3, 10, 11, 12);
3. (9, 13, 17);
4. (4, 18)
19/20 players
1. (6, 7, 8, 9, 15, 16, 17, 18);
2. (1, 2, 3, 11, 12, 13, 14);
3. (5, 10, 19);
4. 4, 20)
21/22 players
1. (6, 7, 8, 9, 10, 17, 18, 19, 20);
2. (1, 2, 3, 4, 12, 13, 14, 15);
3. (11, 16, 21);
4. (5, 22)
23/24 players
1. (6, 7, 8, 9, 10, 11, 19, 20, 21, 22);
2. (1, 2, 3, 4, 13, 14, 15, 16, 17);
3. (12, 18, 23);
4. (5, 24)
The following rules are valid for each Swiss system unless explicitly stated otherwise.
Each system may have exceptions to this rule in the last round of a tournament.
Each system may have exceptions to this rule in the last round of a tournament.
2. If colours are already balanced, then, in general, the player is given the colour that alternates from the last one with which he played.
i. The pairing rules must be such transparent that the person who is in charge for the pairing can explain them.
Accelerated methods are acceptable if they were announced in advance by the organizer and are published in section C.04.5.
In any tournament where such systems are used, different arbiters, or different endorsed software programs, must be able to arrive at identical pairings.
Where it can be shown that modifications of the original pairings were made to help a player achieve a norm or a direct title, a report may be submitted to the QC to initiate disciplinary measures through the Ethics Commission.
It is advisable to check all ratings supplied by players. If no reliable rating is known for a player, the arbiters should make an estimation of it as accurately as possible.
No modification of a pairing number is allowed after the fourth round has been paired.
An exception may be made in the case of a registered participant who has given written notice in advance that he will be unavoidably late.
and a player communicates this to the arbiter within a given deadline after publication of results, the new information shall be used for the standings and the pairings of the next round. The deadline shall be fixed in advance according to the timetable of the tournament.
If the error notification is made after the pairing but before the end of the next round, it will affect the next pairing to be done.
If the error notification is made after the end of the next round, the correction will be made after the tournament for submission to rating evaluation only.
The sorting criteria are (with descending priority)
Version approved at the 87th FIDE Congress in Baku 2016
See C.04.2.B (General Handling Rules  Initial order)
For pairings purposes only, the players are ranked in order of, respectively
A scoregroup is normally composed of (all) the players with the same score. The only exception is the special "collapsed" scoregroup defined in A.9.
A (pairing) bracket is a group of players to be paired. It is composed of players coming from one same scoregroup (called resident players) and of players who remained unpaired after the pairing of the previous bracket.
A (pairing) bracket is homogeneous if all the players have the same score; otherwise it is heterogeneous.
A remainder (pairing bracket) is a subbracket of a heterogeneous bracket, containing some of its resident players (see B.3 for further details).
A player who, for whatever reason, does not play in a round, also receives a downfloat.
See C.04.1.c (Should the number of players to be paired be odd, one player is unpaired. This player receives a pairingallocated bye: no opponent, no colour and as many points as are rewarded for a win, unless the regulations of the tournament state otherwise).
The colour difference of a player is the number of games played with white minus the number of games played with black by this player.
The colour preference is the colour that a player should ideally receive for the next game. It can be determined for each player who has played at least one game.
Topscorers are players who have a score of over 50% of the maximum possible score when pairing the final round of the tournament.
The pairing of a bracket is composed of pairs and downfloaters.
Its Pairing Score Difference is a list of scoredifferences (SD, see below), sorted from the highest to the lowest.
For each pair in a pairing, the SD is defined as the absolute value of the difference between the scores of the two players who constitute the pair.
For each downfloater, the SD is defined as the difference between the score of the downfloater, and an artificial value that is one point less than the score of the lowest ranked player of the current bracket (even when this yields a negative value).
Note: The artificial value defined above was chosen in order to be strictly less than the lowest score of the bracket, and generic enough to work with different scoringpoint systems and in presence of nonexistent, empty or sparsely populated brackets that may follow the current one.
PSD(s) are compared lexicographically (i.e. their respective SD(s) are compared one by one from first to last  in the first corresponding SD(s) that are different, the smallest one defines the lower PSD).
The pairing of a round (called roundpairing) is complete if all the players (except at most one, who receives the pairingallocated bye) have been paired and the absolute criteria C1C3 have been complied with.
If it is impossible to complete a roundpairing, the arbiter shall decide what to do. Otherwise, the pairing process starts with the top scoregroup, and continues bracket by bracket until all the scoregroups, in descending order, have been used and the round pairing is complete.
However, if, during this process, the downfloaters (possibly none) produced by the bracket just paired, together with all the remaining players, do not allow the completion of the roundpairing, a different processing route is followed. The last paired bracket is called Penultimate Pairing Bracket (PPB). The score of its resident players is called the "collapsing" score. All the players with a score lower than the collapsing score constitute the special "collapsed" scoregroup mentioned in A.3.
The pairing process resumes with the repairing of the PPB. Its downfloaters, together with the players of the collapsed scoregroup, constitute the Collapsed Last Bracket (CLB), the pairing of which will complete the roundpairing.
Note: Independently from the route followed, the assignment of the pairingallocated bye (see C.2) is part of the pairing of the last bracket.
Section B describes the pairing process of a single bracket.
Section C describes all the criteria that the pairing of a bracket has to satisfy.
Section E describes the colour allocation rules that determine which players will play with white.
Note: MaxPairs is usually equal to the number of players divided by two and rounded downwards. However, if, for instance, M0 is greater than the number of resident players, MaxPairs is at most equal to the number of resident players.
Note: M1 is usually equal to the number of MDPs coming from the previous bracket, which may be zero. However, if, for instance, M0 is greater than the number of resident players, M1 is at most equal to the number of resident players.
Of course, M1 can never be greater than MaxPairs.
To make the pairing, each bracket will be usually divided into two subgroups, called S1 and S2.
S1 initially contains the highest N1 players (sorted according to A.2), where N1 is either M1 (in a heterogeneous bracket) or MaxPairs (otherwise).
S2 initially contains all the remaining resident players.
When M1 is less than M0, some MDPs are not included in S1. The excluded MDPs (in number of M0  M1), who are neither in S1 nor in S2, are said to be in a Limbo.
Note: the players in the Limbo cannot be paired in the bracket, and are thus bound to doublefloat.
S1 players are tentatively paired with S2 players, the first one from S1 with the first one from S2, the second one from S1 with the second one from S2 and so on.
In a homogeneous bracket: the pairs formed as explained above and all the players who remain unpaired (bound to be downfloaters) constitute a candidate (pairing).
In a heterogeneous bracket: the pairs formed as explained above match M1 MDPs from S1 with M1 resident players from S2. This is called a MDPPairing. The remaining resident players (if any) give rise to the remainder (see A.3), which is then paired with the same rules used for a homogeneous bracket.
Note: M1 may sometimes be zero. In this case, S1 will be empty and the MDP(s) will all be in the Limbo. Hence, the pairing of the heterogeneous bracket will proceed directly to the remainder.
A candidate (pairing) for a heterogeneous bracket is composed by a MDPPairing and a candidate for the ensuing remainder. All players in the Limbo are bound to be downfloaters.
If the candidate built as shown in B.3 complies with all the absolute and completion criteria (from C.1 to C.4), and all the quality criteria from C.5 to C.19 are fulfilled, the candidate is called "perfect" and is (immediately) accepted. Otherwise, apply B.5 in order to find a perfect candidate; or, if no such candidate exists, apply B.8.
The composition of S1, Limbo and S2 has to be altered in such a way that a different candidate can be produced.
The articles B.6 (for homogeneous brackets and remainders) and B.7 (for heterogeneous brackets) define the precise sequence in which the alterations must be applied.
After each alteration, a new candidate shall be built (see B.3) and evaluated (see B.4).
Alter the order of the players in S2 with a transposition (see D.1). If no more transpositions of S2 are available for the current S1, alter the original S1 and S2 (see B.2) applying an exchange of resident players between S1 and S2 (see D.2) and reordering the newly formed S1 and S2 according to A.2.
Operate on the remainder with the same rules used for homogeneous brackets (see B.6).
Note: The original subgroups of the remainder, which will be used throughout all the remainder pairing process, are the ones formed right after the MDPPairing. They are called S1R and S2R (to avoid any
confusion with the subgroups S1 and S2 of the complete heterogeneous bracket).
If no more transpositions and exchanges are available for S1R and S2R, alter the order of the players in S2 with a transposition (see D.1), forming a new MDPPairing and possibly a new remainder (to be processed as written above).
If no more transpositions are available for the current S1, alter, if possible (i.e. if there is a Limbo), the original S1 and Limbo (see B.2), applying an exchange of MDPs between S1 and the Limbo (see D.3), reordering the newly formed S1 according to A.2 and restoring S2 to its original composition.
Choose the best available candidate. In order to do so, consider that a candidate is better than another if it better satisfies a quality criterion (C5C19) of higher priority; or, all quality criteria being equally satisfied, it is generated earlier than the other one in the sequence of the candidates (see B.6 or B.7).
Absolute Criteria
No pairing shall violate the following absolute criteria:
To obtain the best possible pairing for a bracket, comply as much as possible with the following criteria, given in descending priority:
Before any transposition or exchange take place, all players in the bracket shall be tagged with consecutive inbracket sequencenumbers (BSN for short) representing their respective ranking order (according to A.2) in the bracket (i.e. 1, 2, 3, 4,...).
A transposition is a change in the order of the BSNs (all representing resident players) in S2.
All the possible transpositions are sorted depending on the lexicographic value of their first N1 BSN(s), where N1 is the number of BSN(s) in S1 (the remaining BSN(s) of S2 are ignored in this context, because they represent players bound to constitute the remainder in case of a heterogeneous bracket; or bound to downfloat in case of a homogeneous bracket  e.g. in a 11player homogeneous bracket, it is 678910, 67 8911, 6781011,..., 6111098, 768910,..., 1110987 (720 transpositions); if
the bracket is heterogeneous with two MDPs, it is: 34, 35, 36,..., 311, 43, 45,...,
1110 (72 transpositions)).
An exchange in a homogeneous brackets (also called a residentexchange) is a swap of two equally sized groups of BSN(s) (all representing resident players) between the original S1 and the original S2.
In order to sort all the possible residentexchanges, apply the following comparison rules between two residentexchanges in the specified order (i.e. if a rule does not discriminate between two exchanges, move to the next one).
The priority goes to the exchange having:
An exchange in a heterogeneous bracket (also called a MDPexchange) is a swap of two equally sized groups of BSN(s) (all representing MDP(s)) between the original S1 and the original Limbo.
In order to sort all the possible MDPexchanges, apply the following comparison rules between two MDPexchanges in the specified order (i.e. if a rule does not discriminate between two exchanges, move to the next one) to the players that are in the new S1 after the exchange.
The priority goes to the exchange that yields a S1 having:
Any time a sorting has been established, any application of the corresponding D.1, D.2 or D.3 rule, will pick the next element in the sorting order.
Initialcolour
It is the colour determined by drawing of lots before the pairing of the first round. For each pair apply (with descending priority):
Note: Always consider sections C.04.2.B/C (Initial Order/Late Entries) for the proper management of the pairing numbers.
Use of these systems is deprecated unless for a system there is a FIDE endorsed program (see, in Appendix C.04.A, the Annex3 "List of FIDE Endorsed Programs") with a free pairingchecker (see A.5 in the same appendix) able to verify tournaments run with this system.
In Swiss tournaments with a wide range of (mostly reliable) playing strengths, the results of the first round(s) are usually quite predictable. In the first round, only a few percent of the games have a result other than "win to the stronger part". The same may happen again in round two. It can be shown that, in title tournaments, this can prevent players from achieving norms.
An accelerated pairing is a variation of Swiss pairings in which the first rounds are modified in such a way as to overcome the aforementioned weaknesses of the Swiss system, without compromising the reliability of the final rankings.
It is not appropriate to design an entirely new pairing system for acceleration, but rather design a system that works together with existing FIDEdefined pairing systems. This result is normally achieved by rearranging score brackets in some way that is not only dependent on the points that the players have scored. For instance, one of the possible methods is to add socalled "virtual points" to the score of some higher rated players (who are supposedly stronger) and henceforth build the score brackets based on the total score (real score + virtual points).
The following chapters will describe the methods that were statistically proven to accomplish the aforementioned goals. The Baku Acceleration Method is presented first, because it was the first that, through statistical analysis, was proven to be good and stable (and is also easy to explain).
Other accelerated methods may be added, as long as they can be proven, through statistical analysis, to get better results than already described methods or, if their effectiveness is comparable, to be simpler.
Unless explicitly specified otherwise, each described acceleration method is applicable to any Swiss Pairing System.
In its current presentation, the Baku Acceleration Method is applicable for tournaments that last nine rounds or more, and in which the standard scoring point system (one point for a win, half point for a draw) is used.
Before the first round, the list of players to be paired (properly sorted) shall be split in two groups, GA and GB.
The first group (GA) shall contain the first half of the players, rounded up to the nearest even number. The second group (GB) shall contain all the remaining players.
Note: for instance, if there are 161 players in the tournament, the nearest even number that comprises the first half of the players (i.e. 80.5) is 82. The formula 2 * Q (2 times Q), where Q is the number of players
divided by 4 and rounded upwards, may be helpful in computing such number  that, besides being the number of GAplayers, is also the pairing number of the last GAplayer.
If there are entries after the first round, those players shall be accommodated in the pairing list according to C.04.2.B/C (Initial Order/Late Entries).
The last GAplayer shall be the same as in the previous round.
Note 1: In such circumstances, the pairing number of the last GAplayer may be different by the one set accordingly to Rule 2.
Note 2: After the first round, GA may contain an odd number of players.
Before pairing the first three rounds, all the players in GA are assigned a number of points (called virtual points) equal to 1.
Such virtual points are reduced to 0.5 before pairing the fourth and the fifth round.
Note: Consequently, no virtual points are given to players in GB or to any player after the fifth round has been played.
The pairing score of a player (i.e. the value used to define the scoregroups and internally sort them) is given by the sum of his standings points and the virtual points assigned to him.
with Baku 2016 Swiss Rules
In the Baku 2016 FIDE Congress, the new Rules for the FIDE (Dutch) Swiss system were approved. The Rules were thoroughly rewritten, and are now far easier to understand and use. It is now time to begin to put them into practice – let us therefore peer together into some examples of pairings – we will start with a very simple one and proceed to situations that are a bit more difficult. Before reading the examples, however, it is strongly advisable to carefully read the new Rules, which can be found (together with more interesting material about pairings) in the Systems of Pairing and Programs Commission (SPP) webpage, http://pairings.fide.com.
All the examples here come from one same tournament, whose crosstable is this:
ID NAME Rtg Pts  1 2 3 4 5 6 7 8 9

1 Alice 2600 6.5  +W7 +B9 =W4 +B5 =W12 =W2 B3 +B10 +W6
2 Bob 2550 6.0  =B8 W12 +W11 +B6 +W4 =B1 =B5 +W9 =B3
3 Charline 2500 5.0  W9 B11 +W10 B12 +W6 +B8 +W1 =B4 =W2
4 David 2450 6.0  +B10 =W5 =B1 +W8 B2 +W9 +B12 =W3 =B7
5 Eleanor 2400 4.5  +W11 =B4 +W12 W1 B9 =B7 =W2 =B8 =W10
6 Frank 2350 4.0  =B12 =W8 =B9 W2 B3 +B10 =W7 +W11 B1
7 Gale 2300 4.0  B1 +W10 =B8 =W9 B11 =W5 =B6 =W12 =W4
8 Hans 2250 4.0  =W2 =B6 =W7 B4 W10 W3 +B11 =W5 +B12
9 Isabelle 2200 4.0  +B3 W1 =W6 =B7 +W5 B4 =W10 B2 =B11
10 Jack 2150 3.0  W4 B7 B3 +W11 +B8 W6 =B9 W1 =B5
11 Karima 2100 3.0  B5 +W3 B2 B10 +W7 =B12 W8 B6 =W9
12 Leonard 2050 4.0  =W6 +B2 B5 +W3 =B1 =W11 W4 =B7 W8

In the second round, the first scoregroup is:
Player Score Col. hist. Opp. hist.
1  1,0  B  7 
4  1,0  W  10 
5  1,0  B  11 
9  1,0  W  3 
To make the pairing, we must first determine the colour preferences of the players and divide the bracket into subgroups: S1=[1W, 4B] and S2=[5W, 9B] (B,W are strong colour preferences). Now we pair the first player of S1 to the first of S2, the second to the second, and so on, as usual, obtaining the candidate pairing 15, 49, which we must now evaluate against all the pairing criteria.
This candidate is legal (which means that it complies with all the relevant absolute criteria), and therefore we do not discard it at once; but it is nonetheless a definitely poor pairing, as we are disregarding two colour preferences, while we could – and therefore should – disregard zero. In the pairing jargon, we say that the candidate has a failure value equal to two for criterion C.10 (which tells us to minimise the number of disregarded colour preferences). Hence, we temporarily store the candidate (it might still be used, should we find nothing better), but proceed to look for a worthier one.
To look for a better pairing, the first alteration to try is always a transposition in S2; in this bracket, there is only one possible (the rules to generate transpositions, as well as exchanges, are in Section D of the FIDE Dutch rules), which lends S2=[9B, 5W]. Now, the candidate pairing is 19, 54 and, evaluating it against the pairing criteria, we promptly realize that it complies with them all – it is therefore a perfect candidate, which we immediately chose, thus terminating the pairing process for the bracket.
To complete it, however, yet another step is required: we must check that the rest of the players can actually be paired – this is informally called the “Requirement Zero”, and the test is a “completion test”. A failure in this test would mean that the pairing of the current bracket does not allow the pairing for the whole round to come to fruition. In such a case, we discard the candidate and change the pairing rules – but we do not need to worry about this just now: in early rounds, every participant may be legally paired with almost everyone else, so the probability of such an event is virtually zero – and, in fact, the completion test is passed. We will go back to the matter later on.
In the fourth round, player #11 got the second downfloat running… we want to show him that his pairing is correct! Here is the prepairing situation:
Player Score Col. hist. Float hist. Opp. hist.
…
12  1,5  WBB  6 2 5  
3  1,0  WBW  9 11 10  
11  1,0  BWB  ↓  5 3 2 
10  0,0  WBB  4 7 3 
Player #12 here is a “Moveddown player” (MDP) – that is, a player that downfloated from the previous bracket – so this bracket is heterogeneous. From now on, we will also need the float history of the players (which we only seldom need in early rounds). Here, we can see that player #11 just downfloated. By the way, this bracket shall of course give a downfloater, because the number of players is odd.
The pairing of heterogeneous brackets is built in two phases. In the first phase, we compose the MDPPairing, which is a partial pairing containing the MDP(s) (or, at least, as many of them as possible). In the second phase, we complete the candidate with the pairs containing only resident players (viz., those players that belong to the scoregroup); and then, only after the candidate is complete, we evaluate it as a whole.
The first phase is actually very simple: we start as usual by preparing the subgroups, annotating the colour preferences and the float status for each player: S1=[12W*], S2=[3B, 11W↓] (B*, W* are absolute colour preferences), then pair it in the usual way. The only difference is that in subgroup S1 we will put not half the players, as we would do in a homogeneous bracket, but all the MDPs that we are going to pair.
The first possible MDPPairing contains the pair 123. The remaining players (here, we have just one!) constitute the remainder {11W↓}, which we are going to pair in the second phase. With three players, we can build just one pair; hence, the remainder will yield no pairs at all and the candidate (which is built putting together the pairs from the
MDPPairing, the pairs from the pairing of the remainder and the possible downfloaters) is 123, 11 to float.
We must now evaluate this candidate – which means that we must verify its compliance with each of the relevant paring criteria. We readily verify that this candidate complies with all the pairing criteria but C.12 (Minimize the number of players who receive the same downfloat as the previous round). Hence, the candidate is legal but not perfect – let us call it the “temporarybest” (or “champ”). We therefore store it and proceed to look for something better – but, if nothing better is available, we will retrieve it and use it as “the” pairing.
To get the next candidate, we apply a transposition (there is only one), obtaining S2=[11W↓, 3B], which yields the candidate 1211, 3 to float. This candidate is legal and therefore we evaluate it – but only to learn that it does not comply with criterion
C.10 (Minimize the number of players who do not get their colour preference). We must now compare the current temporarybest with this new candidate: the latter fails on C.10, while the former fails on C.12. Since C.10 is more important than C.12, we discard the new candidate and keep the old temporarybest.
Now there are no more transpositions, but we must continue to look for a perfect candidate – or use up all the possible ones. The next step would be to try an exchange between S1 and the Limbo (which means, to try and change the MDPs to be paired), but there are no more MDPs.
The last attempt is to reduce the number of MDPs to pair – this means, in practice, to make player #12 join the Limbo, from which it cannot help but float again, and pair players 113 between themselves. In a homogeneous bracket, a similar exchange would have been possible – but here player #12 is a MDP! This means that its score is higher than that of the residents – making it float causes a worse PSD, and thus a failure on criterion C.6 (Minimize the PSD), which is more important than both C.10 and C.12. Thus, also this candidate is discarded.
Well, we have now exhausted all the possible pairings, but found no perfect candidates: hence, we must choose the “less imperfect” one, which is of course the current temporarybest – and contains player #11 as a floater.
The third example is rather typical, and still it is not a difficult one – but it definitely requires some patience. In round four, the twopoint scoregroup contains only one player, who shall downfloat to the next bracket:
Player Score Col. hist. Float hist. Opp. hist.
4  2,0  BWB  ↑  10 5 1 
2  1,5  BWW  ↑  8 12 11 
6  1,5  BWB  12 8 9  
7  1,5  BWB  1 10 8  
8  1,5  WBW  2 6 7  
9  1,5  BWW 

3 1 6 
12  1,5  WBB  6 2 5  
3  1,0  WBW  9 11 10  
11  1,0  BWB  ↓  5 3 2 
Player #4 is a MDP. Player #2 just upfloated (by the way, the same holds true for player #4 – but we may ignore it, because that player is certainly not going to upfloat now), and we foresee a downfloater (again, because the number of players is odd). Four players expect White and three expect Black, so we should be able to form three pairs, with no disregarded colour preferences (x=0).
As always, we first compose the MDPPairing, then we complete the candidate with the pairs containing only resident players, and finally we evaluate it.
First, we prepare the subgroups, annotating the colour preferences and the float status for each player: S1=[4W], S2=[2B*↑, 6W, 7W, 8B, 9B*, 12W*]. The first possible MDPPairing contains the pair 42. The remaining players constitute the remainder
{6W, 7W, 8B, 9B*, 12W*}, which we are going to pair in the second phase. Once again, we divide the players (of the remainder) in two subgroups: S1R=[6W, 7W], S2R=[8B, 9B*, 12W*], thus obtaining the remaining pairs of the candidate as 68, 7 9, 12 to float. Now we must evaluate the candidate, which is not legal, because players #6 and #8 already played each other in a previous round.
We discard the candidate at once, and proceed with a transposition in the remainder – that is, in the subgroup S2R. We might observe, however, that player #6 already met all the players in S2R, and therefore no transposition at all can yield a legal pairing. We may therefore be “smart” and jump ahead to try a resident exchange, viz. an exchange of players between S1R and S2R.
The first such exchange is between players #7 (the lower in S1R) and #8 (the higher in S2R), and gives the new subgroups S1R=[6W, 8B] and S2R=[7W, 9B*, 12W*]. This gives a candidate 42, 67, 89, 12 to float which is legal, so that we may at least evaluate it. The first thing to verify is that the chosen downfloater (#12) can be paired in the next bracket (criterion C.7), and we find no problems here. Two colour preferences are disregarded, so we have a double failure for criterion C.10. Moreover, player #2 upfloats for the second time running, so we have one failure for criterion C.13 too. In conclusion, the candidate is legal but not perfect – as usual, we store it as temporarybest and proceed to look for something better.
To get the next candidate, we apply a transposition, obtaining S2R=[7W, 12W*, 9B*]. The resulting candidate (42, 67, 128, 9 to float) still has one failure for C.10 and one for C.13 – however, it is better than the previous one, so we store this one and discard the latter, but we keep looking for a perfect one.
Any other transposition yields illegal candidates (let’s remember that 69 and 612 are forbidden), so we should now try the following resident exchange, which gives S1R=[6W, 9B*], S2R=[7W, 8B, 12W*]. The only legal candidate that we did not already try is (after a transposition) 42, 67, 129, 8 to float – but it is easy to verify that this is not better than the current champ, so we discard it and keep the latter.
We should then proceed examining all exchanges and transpositions, one by one, but we might already suspect that this can be long and tedious work… luckily, we may avoid some useless effort and save some precious time by looking into the remainder
{6W, 7W, 8B, 9B*, 12W*} in a “smarter” way. The key lies in player #6: since it has already played with #8, #9, and #12, the only legal candidates containing #6 must have the pair 67 in them, which gives a failure for C.10. However, if we make #6 downfloat, we may build the pairs 79, 128, thus disregarding no colour preferences at all (please note that 78, 129 would give an illegal candidate, because #7 and #8 already met eachother). This yields the candidate 42, 79, 128, 6 to float, which contains only one failure for criterion C.13, and therefore becomes the new temporarybest.
This is the best we may do with the current MDPPairing; yet, it is not a perfect candidate, so our quest is not over… The next step is looking for a better MDPPairing; let us therefore start again with the original subgroups: S1=[4W], S2=[2B*↑, 6W, 7W, 8B, 9B*, 12W*].
We might observe that we have one C.13 failure because #4 has been paired with #2, who was an upfloater in the previous round. To get a better pairing, we need to pair player #4 with someone else, and this requires one or more transpositions in S2 – which, in practice, means to try to pair #4 with each member of S2 in turn. However, the pairings 46 and 47, although legal, introduce (at least) one failure for C.10 – therefore, no candidate originating from those MDPPairings can be better than the current temporarybest, and there is no point in examining them. The first potentially interesting pairing is 48, which is legal and yields the remainder {2B*↑, 6W, 7W, 9B*, 12W*}.
We are now interested only in (legal) candidates better than the temporarybest (if any)
– that is, with perfect colour matching – and this is only possible after exchanging #6 for #7: S1R=[2B*↑, 7W], S2R=[6W, 9B*, 12W*]. This yields the candidate 48, 62, 79, 12 to float. Its evaluation shows that all criteria are complied with (in particular, the downfloater optimises the pairing in the next bracket) and therefore the candidate is, at long last, perfect! We immediately choose it (discarding the previous temporary best, now useless) and proceed to perform the completion test for Requirement Zero – which is (luckily) successful.
In this last example taken from round 5 (after that round, the pairings become quite challenging), player #7 got a “doubledownfloat” – let us see why. Actually, this pairing involves the last three scoregroups, and is a bit (but not very much) more difficult than the previous ones.
Player Score Col. hist. Float hist. Opp. hist.
5  2,5  WBWW  11 4 12 1  
7  2,0  BWBW  1 10 8 9  
9  2,0  BWWB 

3 1 6 7 
6  1,5  BWBW 

12 8 9 2 
8  1,5  WBWB  ↑  2 6 7 4 
3  1,0  WBWB  ↑  9 11 10 12 
10  1,0  WBBW  ↑  4 7 3 11 
11  1,0  BWBB  ↓↓  5 3 2 10 
The first bracket is heterogeneous, with player #5 as a MDP: [5B*][7b, 9w] (w, b= mild colour preferences). This bracket is easily paired: after a transposition, player #5, who has an absolute preference for Black, is paired with the only Whiteseeker, which is #9, while player #7 downfloats. In the destination bracket, player #7 is compatible, so criterion C.7 is satisfied – actually, all the relevant criteria are complied with, so the candidate is perfect.
However, before proceeding to the next scoregroup, we must prove that the remaining players, together with the floater, allow at least one legal pairing (in early rounds we may usually omit this check, as it is practically always satisfied – but not now!). It is important to appreciate that, in performing this test, we are not looking for “the pairing”. All we want is to show that at least one pairing exists for the remaining players and the quality of this pairing is not important in the least: its mere existence is proof that the roundpairing can be completed, and therefore we will not need to go back to this bracket again. Actually, for example we might pair 711, 610, 83.
Now we proceed to the next bracket, which is [7b][6b, 8w↑]. It is easy to proof that this bracket gives no perfect candidate; in the end, we settle for the (final) temporarybest: 67, 8 to float. However, here comes the twist: the remaining players, who are now {8, 3, 10, 11}, cannot be paired, and therefore we have a Requirement Zero failure.
We must restart from the current bracket [7b][6b, 8w↑], which becomes the Penultimate Pairing Bracket (PPB). This bracket is now subject to a new and special rule: it must provide the downfloaters needed to pair the rest of the players. All those remaining players are put together in a big melting pot, which is called the Special Collapsed Scoregroup (SCS). This special scoregroup may therefore contain players with different scores (and indeed this is usually the case, although this does not happen here) and, together with the downfloaters from the PPB, forms the Collapsed Last Bracket (CLB).
Actually, players #3, #10, and #11 are all incompatible with each other – and therefore, to complete the pairing, we apparently need three MDPs, one for each of them. Well, the PPB contains just three players – hence, we obtain zero pairs from the PPB (which is, sometimes, a perfectly legitimate pairing) and send the three required floaters in the CLB.
We must now pair the CLB. It contains players with different scores, so, in general, we must pay attention to the Pairing Score Difference (PSD). Because of this, in the bracket we want to annotate also the scores of the players: {(2.0)7b, (1.5)6b, (1.5)8w↑, (1.0)3w↑, (1.0)10b↑, (1.0)11W*↓↓}.
For this bracket, we must find the best possible pairing. This may seem a tough task, but actually most possible candidates are illegal, because many players already played with each other (of course, this is more or less usual in a CLB). The legal pairings are therefore comparatively few, and usually the best strategy is simply to evaluate them all and apply the “Sieve pairing method”. In the present case, player #7 has already played with #8 and #10. Moreover, it cannot be paired with #6 because the rest would not be paired. Hence, we have only four legal candidates complying with criteria C.1
C.5. The first criterion to check is now C.6 to minimise the PSD (see C.04.3.A.8 in the FIDE Swiss Rules):
73, 610, 811 PSD={1.0, 0.5, 0.5}
73, 611, 810 PSD={1.0, 0.5, 0.5}
711, 610, 83 PSD={1.0, 0.5, 0.5}
711, 63, 810 PSD={1.0, 0.5, 0.5}
Actually, all the candidates have the same PSD – and of course, this is no coincidence: since the SCS contained no different scores, all pairings with its resident must end up with the same score differences. Now, criteria C.7, C.8 and C.9 do not apply, so we proceed to C.10 to check for colour matching.
1. 7b3w, 6b10b, 8w11W*
2. 7b3w, 6b11W*, 8w10b
3. 7b11W*, 6b10b, 8w3w
4. 7b11W*, 6b3w, 8w10b
The first and third candidates contain each two disregarded colour preferences, so we discard them and keep the second and fourth, which are perfectly colour matched.
1. 7b3w↑, 6b11W*, 8w↑10b↑
2. 7b11W*, 6b3w↑, 8w↑10b↑
We have no strong colour preferences, so criterion C.11 does not apply. Since we have no prospective downfloaters, also criterion C.12 does not apply. We proceed to criterion C.13, regarding the repetition of upfloats: the first candidate fails once on pair 37, whilst the second one fails once on pair 36. As the failure values are equal, the candidates are still “tied” and we must proceed. We skip criteria C.14, C15, and C.16 (they are not relevant) and reach criterion C.17 (Minimize the score differences of players who receive the same upfloat as the previous round). Here, at last, we have a difference: in the first candidate, the score difference of the player who receives the second upfloat in a row, which is #3, is 1.0; in the second candidate, the score difference of the player who receives the second upfloat in a row, which incidentally is again #3, is 0.5. Hence, the failure value for this criterion is greater for the first candidate. The latter is therefore discarded too, and we finally remain with the only surviving candidate, which is also the required pairing: 117, 36, 810.
I hope that these examples may be of help in studying the new FIDE (Dutch) Swiss Rules. I strongly suggest to all those interested in this matter to pay a visit the FIDE Systems of Pairings and Programs Commission (SPP) webpage (http://pairings.fide.com).
An example of a sixround tournament (With Baku 2016 FIDE C.04 Swiss Rules) By Mario Held (Member of FIDE Systems of Pairings and Programs Commission)
ANNOTATED RULES FOR THE FIDE (DUTCH) SWISS SYSTEM
Hereafter, we present the general rules for Swiss Systems (FIDE Handbook C.04.1 and C.04.2) and the Rules for the FIDE (Dutch) System (FIDE Handbook C.04.3), together with some notes to explain them.
The first part contains rules that define the technical requirements any Swiss pairing system must obey, whilst the second part targets a set of various aspects relating to the handling of tournaments, from the fairness of the systems to the management of late entrants, and several rules that are common to all the FIDE approved systems.
The third part contains the Rules for the FIDE (Dutch) Swiss System, which in its turn is comprised of the following sections:
With reference to previous versions, the FIDE (Dutch) rules have been almost completely reworded, in order to make them simpler and more intuitive. The algorithm, which used to occupy the whole section C, has now been completely evicted from the rules, together with the whole old section B. Instead of the latter, a new section C contains a revised list of the pairing quality criteria, which is both more detailed and clearer than the previous one.
For all this rewording, the real changes in the pairings address only a few cases, while a vast majority of the pairings remain just the same as they were with the previous rules 1.
We would like to suggest you to carefully study the Rules until you feel you master their principles and meanings, before starting to study the tournament example.
The following rules are valid for each Swiss system unless explicitly stated otherwise.
After the start of the tournament, we are not allowed to change the number of rounds (however, this may become inevitable by force of circumstances).
This is the only principle of Swiss Systems we cannot dispense with (unless doing differently is absolutely inevitable...)!
Please note that this rule allows event organizers to establish a different value for byes (e.g. half a point) instead of the usual whole point.
However, and whatever its value is, a pairing allocated bye (“PAB”) cannot be assigned to any player who has already received a previous one, or a forfeit win. The allocation of a PAB, though, is not prevented by a previous bye “on request”1 (when such a provision is permitted by the tournament rules).
The location of this principle before colour balancing rules highlights its greater importance with respect to the latter. It is because of this rule that we cannot make players float to suit colour
1 In the previous versions of the Swiss Rules, any number of points got without playing, like e.g. a requested “Half Point Bye”, did prevent the allocation of a PAB.
Each system may have exceptions to this rule in the last round of a tournament.
Each system may have exceptions to this rule in the last round of a tournament.
preferences that are not absolute (see C.04.3:A.6.a).
We should emphasize that the exceptions to rules f and g for the last round are possible, but not compulsory. The FIDE (Dutch) system adopts them, tough in practice only when there are very good reasons to do so. Other systems do not do the same  e.g., the Dubov Swiss System definitely refuses to make such exceptions, which seem not to be consistent with the basic principles of that system.
This rule warrants the good colour balancing typical of all FIDE approved Swiss Systems.
Sometimes, players ask the Arbiter to justify, or explain, the pairings, which, nowadays, are most usually prepared with the help of a software program (which should be a FIDE endorsed one, if only possible). However, we want to remember that, even if the pairings are made by means of a computer, it is always the arbiter who takes responsibility for the pairing, not the software.
All the rules in this section tend to the same aim: to prevent any possible tampering with the pairings in favour of one or more participants (such as helping a player to obtain a norm). To this effect, the pairing rules must be well specified, transparent, and unambiguous in the first place.
pairing rules should arrive at identical pairings.
Where it can be shown that modifications of the original pairings were made to help a player achieve a norm or a direct title, a report may be submitted to the Qualification Commission to initiate disciplinary measures through the Ethics Commission.
 CM  WFM  WCM  no title)
If, for any reason, the data used to determine the rankings were not correct, they can be adjusted at any time. The pairing numbers may
The fundamental principle of all Swiss systems is to pair tied players (i.e. players with the same number of points) so that, in the top echelon, the number of ties is halved at every round. Thus, in a tournament with T rounds, if the number N of players is less than 2^{T} [i.e. T ≥ log2 (N)], we should (theoretically) have no ties for the first place.
However, practice shows that, to reach this goal in a real environment (which includes draws and unexpected results), a precise evaluation of the strength of players is essential.
When no better information is available, the estimated rating of an unknown player can be determined based on a national rating (if available) using the appropriate conversion formulas; or other rating lists, tranches, tournament results and so on may be used, if reliable. In conclusion, the Arbiter shall have to use sound judgment and reasoning, to obtain the best possible evaluation with what data is available.
FIDE titles are ordered by descending nominal rating; when ratings are equal, titles obtained through norms take precedence with respect to automatic ones.
Alphabetical sorting is unessential, its only rationale being that of ensuring an unambiguous order. Thus, this criterion can be substituted for by any other sorting method capable of giving an unambiguous order, provided this method has been previously declared in the tournament regulations.
Please notice that a lower numeric value corresponds to a higher ranking; this choice may not seem “natural”, but it is deeply rooted in common language by now.
Pairing numbers are used by all Swiss pairing systems except Dubov. Thus, a change in be reassigned accordingly to the corrections, but only for the first three rounds. No modification of a pairing number is allowed after the fourth round.
An exception may be made in the case of a registered participant who has given written notice in advance that he will be unavoidably late.
pairing numbers changes the pairings too. We would expect this to happen, if at all, in the first round of a tournament  in some (rare) instances even in the second or in the third round  and, when such changes happen, they make the checking of the pairings rather difficult. Hence, in order to make it easier to perform such checks on advanced stages of a tournament, the rule prohibits late changes of the pairing numbers.
As correct ratings, titles and so on are needed to correctly rate the tournament, such data may always be corrected, even in late rounds (and even after the tournament is finished!), but without changing the pairing numbers.
It seems appropriate to point out that the declaration of delay must be given in advance, in writing, and stating reasons for it. Verbal communications (telephone, etc.) do not suffice. Since exceptions may be made, it is the Arbiter’s responsibility to grant or decline such requests.
We want to take notice that the admission of a latecomer is a choice of the Chief Arbiter, who takes the final decision – and must take the responsibility too, especially if during the round there are empty seats... Thus, before accepting a latecomer and making the actual pairing, we want to be very sure that the player will actually be there in time to play. If we are not that sure, it is probably better to let the player enter the tournament, and be paired, only for a subsequent (second, third) round.
Entering a late player in the tournament causes the pairing numbers to change according to the new ranking list; some of the players will thus play the following rounds with a different pairing number, and this may cause some perplexity among the players. For example, consider a player, correctly registered from the beginning, but entering a tournament (say, with 100 players) on the second round, as #31. In the first round that player had no pairing number – hence, the
=BWBW. WB=WB will count as =WBWB, BWW=B=W as = =BWWBW and so on.
players who (now) have numbers 33, 35, 37 and so on, in the first round had even pairing numbers and thus the colour opposite to that of player #1.
By the way, we should also observe that the limit imposed in C.04.2.B.4 on the regeneration of pairing numbers does not extend to the case of a newly added late player.
Basically, we look only at actually played games, skipping “holes”, which float to the top of the list. Thus, for example, in the comparison between the colours histories of two players, the sequence
== WB is equivalent to =W=B and WB== (and the latter two are equivalent to each other!).
The application of this rule and the next requires us to set (and post!) a timetable for the publication of pairings. Above all, these rules put
and a player communicates this to the arbiter within a given deadline after publication of results, the new information shall be used for the standings and the pairings of the next round. The deadline shall be fixed in advance according to the timetable of the tournament.
If the error notification is made after the pairing but before the end of the next round, it will affect the next pairing to be done.
If the error notification is made after the end of the next round, the correction will be made after the tournament for submission to rating evaluation only.
a constraint on the possible revision of the pairings: if an error is not reported within the specified deadline, all subsequent pairings, as well as the final standings, shall be prepared making use of the wrong result as if it were correct.
The sorting criteria are (with descending priority):
C.04.1.b (Two players shall not play against each other more than once).
Even when using a pairing software program, it is mostly advisable to check boards order before publishing the pairing, because many players interpret even an incorrect board order as a “pairing error”.
Version approved at the 87th FIDE Congress in Baku 2016.
See C.04.2.B (General Handling Rules  Initial order)
For pairings purposes only, the players are ranked in order of, respectively:
A scoregroup is normally composed of (all) the players with the same score. The only exception is the special “collapsed” scoregroup defined in A.9.
A (pairing) bracket is a group of players to be paired. It is composed of players coming from one same scoregroup (called resident players) and of players who remained unpaired after the pairing of the previous bracket.
A (pairing) bracket is homogeneous if all the players have the same score; otherwise it is heterogeneous.
A remainder (pairing bracket) is a subbracket of a heterogeneous bracket, containing some of its resident players (see B.3 for further details).
In the destination bracket, such players are called “moveddown players” (MDPs for short)
Players are ordered in such a way that their presumable strengths are likely to decrease from top to bottom of the list (see also C.04.2:B).
Please notice that when we include a late entry, the list should be sorted again, thus assigning new pairing numbers to the players (C.04.2:C.2,3). The same may be done when some wrongly entered rating had to be corrected. When this happens, some participants may play subsequent rounds with new, different numbers; and, of course, this change may, if not adequately advertised, muddle players who, in reading the pairings, still look for their old numbers.
This definition solves any ambiguity between scoregroups and pairing brackets, stating that the scoregroup is the “backbone” of a pairing bracket, which is made of a scoregroup together with the players remaining from the pairing of the previous bracket. The players from the scoregroup are called “resident”, and usually have all the same score, which is called resident score and is the “nominal score” of the bracket. Only when the scoregroup is the “Special collapsed” one, the resident players may have different scores.
The difference is that in a homogeneous bracket there are no score differences between players to be taken care of (to be homogeneous, a bracket must be made of just a (normal) scoregroup and nothing more).
Article B.3 illustrates how to build a candidate pairing for a bracket and explains how and when a remainder is built and used.
A player may become a downfloater because of several reasons; first, the bracket may contain an odd number of players, so that one shall unavoidably remain unpaired. Then, the player may have no possible opponent (and hence no legal pairing) in the bracket. Sometimes, two or more players share between them a number of possible opponents in such a way that no player is incompatible, but we cannot pair all of them (e.g., two players with only one possible opponent, three players with only two possible opponents, and so on)2. Last, but not least, in some instances the player may have to float down, in order to allow the pairing of the following bracket.
2 This situation is sometimes (unofficially) called semiincompatibility or island(in)compatibility.
A player who, for whatever reason, does not play in a round, also receives a downfloat.
See C.04.1.c (Should the number of players to be paired be odd, one player is unpaired. This player receives a pairingallocated bye: no opponent, no colour and as many points as are rewarded for a win, unless the regulations of the tournament state otherwise).
The colour difference of a player is the number of games played with white minus the number of games played with black by this player.
The colour preference is the colour that a player should ideally receive for the next game. It can be determined for each player who has played at least one game.
In analogy to “downfloater”, we will use the term “upfloater” to indicate a player paired to another one having a higher score3 (usually, the opponent of a downfloater).
Downfloats and upfloats are a sort of markers, used to record previous unequal pairings of the player. The reason to keep track of such pairings is that, in general, we want to minimise, and, as far as possible, avoid, their occurrence for the same players. Actually, a pairing between floaters constitutes a disturbance to the general principle of Swiss systems that the players in a pair should have the same score, and therefore the rule try to limit the repetition of such events4.
We want to notice that any player who did not play a round receives a downfloat. This is important because it affects the following two pairings for that player. For example, it becomes unlikely that such a player may receive a downfloat or get the PAB [A.5] in the next round5.
In other Swiss systems (e.g. Dubov) the player, whom the PAB will be assigned to, is selected before starting the pairing for the round.
In the FIDE (Dutch) system, on the contrary, the roundpairing (see A.9) ends up with an unpaired player, who will receive the pairingallocated bye (PAB).
During pairing, we will try to accommodate (as much as possible) the colour preferences of the players – and this is the reason for the good balance of colours of Swiss modern systems.
Participants, who have not played any games yet, just have no preference, and shall therefore accept any colour (see A.6.d).
In general, the colour difference should not become greater than 2 or less than 2 – with the possible exception of high ranked players in the last round, which can receive, if necessary, the third colour in a row or a colour three times more than the opposite (but this is still a relatively rare event).
3 Please notice that in other Swiss pairing systems (e.g. Dubov), the same term “upfloater” may indicate a player transferred to a higher bracket.
4 We may also note that the FIDE (Dutch) system uses a “local” approach to this problem, which looks only to the last two rounds. On the contrary, the Dubov system adopts also a “global” approach, putting also a limit on the total number of floats in the whole tournament (three floats for tournaments up to nine rounds, four for longer tournaments).
5 On the contrary, the previous rules did not assign a downfloat to a player who forfeited a game, so such players had no protection against getting a PAB or a downfloat in the following round. Because of this, a weak player absent in the first round could get a PAB in the second round.
than 1 or when the last two games were played with black. The preference is black when the colour difference is greater than +1, or when the last two games were played with white.
Topscorers are players who have a score of over 50% of the maximum possible score when pairing the final round of the tournament.
The pairing of a bracket is composed of pairs and downfloaters.
To determine an absolute colour preference, we examine only the actually played rounds, skipping any unplayed games6 (whatever the reason may be) in compliance with [C.04.2:D.5] (e.g., the sequence WBBW=W gives an absolute colour preference).
Notice that any disregarded colour preference, be it strong or mild, will give origin to an absolute colour preference on the subsequent round.
If neither player has a colour preference (as is normal when pairing the first round, but may sometimes happen also in subsequent rounds), we resort to the colour allocation rules in section E. There, by means of the initialcolour (decided by drawing of lots before the pairing of the first round) and of rule E.5, we will be able to assign the correct colour to both players.
Such highscoring players are especially important in the determination of the winner and of the top ranking7. Hence, we may apply some special treatment criteria to their pairings  e.g., a player may receive a same colour three times more than the other one, or three times in a row, if this is needed to make it meet an opponent better suited to the strength the player demonstrated.
This is an important idea: the pairing of a bracket is not made only of pairs: the downfloaters are part of it too – and a very important part, at that! In fact, as we shall see, the choice of the downfloaters may determine if it will be possible to pair the remaining players – and therefore if the pairing is a valid one.
6 Please note the difference with floats, for which we look at the last two rounds of the tournament schedule (but remember that an unplayed game gives a downfloat).
7 Not all the “topscorers” are really competing for top ranking places; nonetheless, they are more likely to be of importance in the formation of the top standings than lowranked players, in several collateral ways – e.g. they may be opponents to prospective prize winners, or their score may give a determinant contribute in tiebreak calculations, and so on.
Its Pairing Score Difference is a list of score differences (SD, see below), sorted from the highest to the lowest.
For each pair in a pairing, the SD is defined as the absolute value of the difference between the scores of the two players who constitute the pair.
For each downfloater, the SD is defined as the difference between the score of the downfloater, and an artificial value that is one point less than the score of the lowest ranked player of the current bracket (even when this yields a negative value).
Note: The artificial value defined above was chosen in order to be strictly less than the lowest score of the bracket, and generic enough to work with different scoringpoint systems and in presence of nonexistent, empty or sparsely populated brackets that may follow the current one.
PSD(s) are compared lexicographically (i.e. their respective SD(s) are compared one by one from first to last  in the first corresponding SD(s) that are different, the smallest one defines the lower PSD).
The Pairing Score Difference allows the best management of the overall difference in scores between the paired players. In practice, it is a list of the score differences, built as follows: we calculate the score differences (SD) in each pair and for each downfloater, then sort them from higher to lower, thus obtaining a string of numbers. Each single difference is taken in absolute value (so that it is always positive), because it’s irrelevant which one of the players have a higher score.
While the meaning of the SD is obvious for pairs, it is far less obvious for downfloaters, who have no opponent yet. Nonetheless, we need to account, somehow, for the perspective score difference relative to the player when it will finally be paired  in such a way that giving a float, or a PAB, to a higher scored player should be worse than giving it to a lower scored one. So we go for a “presumptive” score difference, establishing a hypothetical score for the residents of the (yet undefined!) next bracket.
In order to be sure that we can accommodate a wide variety of possible next brackets, we choose a value lower enough than that of the current bracket, namely one point less than the minimum score of its (resident) players. In the last two brackets, this may yield a negative value – e.g., in the 0.5 points bracket this value is 0.5 points. This is not a problem, as we will simply take the difference between a positive value and this one, so the result will always be positive.
Please note that in the last bracket the only possible downfloater is the player who is going to get the PAB. Thus, this calculation provides an easy and uniform way to minimise the score of the players who get the PAB.
PSDs are compared following the lexicographical order (the “order of the dictionary”). We start by comparing the first number of the first PSD with the first number of the second PSD: if one of those two is smaller than the other one, the PSD it belongs to is the “smaller”. If they are equal, we proceed to the second element of each PSD, and repeat the comparison. Then, if needed, we go on to the third, the fourth, and so on  until we reach the end of the strings8.
An alternative (but fully equivalent) method of comparison is the following: substitute a letter for each number of each PSD, following the correspondence A=0, B=0.5, C=1, D=1.5, E=2 and so on. Doing so, we transform the PSDs in
8 Of course, this method only has significance if the two PSD have the same length; but this is always the case, because the PSD comparison is used only when pairings with the same number of pairs are involved. Were the number of pairs different, we would never get to a PSD comparison.
alphabetical words, which can be compared using the simple alphabetical order. The word that comes first (alphabetically) corresponds to the “smaller” PSD.
This article is essentially a guideline giving a panoramic vision of the pairing process, both in the more common case in which the pairing can be completed by normal means, and in the special case in which this is not possible. This is a very important thing to do, as the new Rules do not any more contain an algorithm to dictate a stepbystep procedure.
The pairing of a round (called roundpairing) is complete if all the players (except at most one, who receives the pairingallocated bye) have been paired and the absolute criteria C1C3 have been complied with.
If it is impossible to complete a roundpairing, the arbiter shall decide what to do.
Otherwise, the pairing process starts with the top scoregroup, and continues bracket by bracket until all the scoregroups, in descending order, have been used and the roundpairing is complete.
However, if, during this process, the downfloaters (possibly none) produced by the bracket just paired, together with all the remaining players, do not allow the completion of the roundpairing, a different processing route is followed.
We want to notice that that this definition refers not to a bracket but to the complete round. Thus, we cannot accept unpaired players (apart from a possible PAB)  all players must be paired. On the other hand, the constraints for such a pairing are very loose, not to say minimal – we are only asking for it to comply with the absolute criteria. This does not mean that we may feel free to make a poor pairing: in general, several complete pairings will be possible for each round, and “the” pairing – the correct one  shall simply be the one among them that best satisfies all the pairing criteria.
This is something really brandnew: for the first time ever, the case in which no pairing at all can be done is referred to by the rules. From a practical point of view, this is not a very helpful rule  in fact, in these (luckily rare) cases, the arbiters must act according to their best judgment
– but, at least, the possibility has been accounted for.
The pairing process starts with the topmost scoregroup; with it, we build the first bracket and try to pair it. This pairing may possibly leave some downfloaters that, together with the next scoregroup, will form the next bracket, and so forth – until all players have been paired.
Before starting the pairing of a bracket, we must verify that at least one legal pairing (i.e. a pairing that complies with the absolute criteria) exists for all the players as yet unpaired, together with the downfloaters (of course, possibly none) left from the bracket just paired9. This requirement is informally called the “Requirement Zero”, and its check is called a “Completion test”.
If this check fails before pairing the first bracket, there is no way at all to complete the round pairing, so we have an impossible pairing  which is bad news.
When, on the contrary, this happens after the pairing of the first bracket, we already know that at least one legal pairing exists for the entire round (we checked this before pairing the first
9 Of course, this check is far simpler than the actual complete pairing, because (for the moment) we are not interested in finding the best (correct) pairing, but only in showing that at least a legal one exists.
The last paired bracket is called Penultimate Pairing Bracket (PPB). The score of its resident players is called the “collapsing” score. All the players with a score lower than the collapsing score constitute the special “collapsed” scoregroup mentioned in A.3.
The pairing process resumes with the repairing of the PPB. Its downfloaters, together with the players of the collapsed scoregroup, constitute the Collapsed Last Bracket (CLB), the pairing of which will complete the roundpairing.
Note: Independently from the route followed, the assignment of the pairingallocated bye (see C.2) is part of the pairing of the last bracket.
bracket!). Nevertheless, if the set formed by the downfloaters together with all of the remaining players cannot be paired, it means that, given those downfloaters, we cannot complete the pairing without infringing the absolute criteria.
In this situation, the pairing produced by the last (in fact, still current!) paired bracket is not adequate, and we need to modify it before proceeding. We must restart with this same bracket, while changing the pairing conditions, in order to be able to find the pairing (which, as we already know, must undoubtedly exist). This change of conditions may have two effects: the first, and less invasive, is a different choice of downfloaters10, while the second is an increase in the number itself of downfloaters. (The latter is of course the only option available when the original pairing did not produce any floater.)
First, we pool together all the players, whose score is lower than the collapsing score. Then, with those players, we build the “special collapsed scoregroup” (SCS)  whose players are all resident, regardless of their score.
The bracket just tentatively paired, and which we are now going to pair again, is now called PPB11.
The primary goal in pairing the PPB is to have it produce a set of downfloaters that allows a complete pairing of the SCS [C.4]. With those downfloaters, together with the SCS, we build the CLB, which is by definition the last bracket. The pairing of those two brackets requires some special attentions12.
By stating that the assignment of the PAB is always part of the pairing of the last bracket, this note is telling us that criterion C.2, which regulates the assignment of the PAB, is only significant when the last bracket is in some way involved in the pairing – that is to say:
10 Please note that we check (and, if necessary, change) the selected downfloaters in two completely different situations: the first is when we try to optimise the number of pairings and PSD in the next bracket (see C.7). The second is when the rest of the players cannot be paired and the PPB must give the correct floaters to allow a complete pairing. Here we are referring to the latter situation.
11 Actually, the name for this bracket comes from the previous version of the Swiss rules, which included a bracket with a similar function.
12 For further details, see [B.7].
Section B describes the pairing process of a single bracket.
Section C describes all the criteria that the pairing of a bracket has to satisfy.
Section E describes the colour allocation rules that determine which players will play with white.
Without this note, we might think the allocation of the PAB to be something to be done after having paired the last bracket – in fact, just as if that bracket had produced a floater  to be paired with a fictitious player in a virtual afterthelast bracket. Hence, if that player could not receive the PAB, we would have to consider the last bracket as the PPB, and subsequently restart the pairing process from this point of view... This note is specifically meant to avoid any possible ambiguity, explicitly excluding such an interpretation.
Moreover, the note also states that, even when it is readily apparent that from the current bracket a downfloater will result, who is bound to get the PAB (e.g., in the next bracket(s) there is no player who can get it), the choice of the floater shall not keep in mind the allocation of the PAB.
We should also notice that pairs are made based also on expected colours, but actual colour assignment is only done at the end of the pairing.
For those who knew the old version of the Dutch rules, it may be useful to spend some words about the new structure. The new Sections B and C contain all the rules that were previously detailed in the algorithmic section with the support from (previous) Section B. Nonetheless, in the previous version of the Rules, the pairing route was different. When the pairing of a bracket was completed, it was accepted (for the moment), and the pairing went forward to the next bracket. If the next bracket was satisfactorily paired (and, sometimes, not even satisfactorily, since a downfloater could create a situation in which a resident player of the new bracket was made incompatible), the pairing for the previous one became (almost) final. If, on the contrary, a better pairing was possible for the next bracket (i.e., one that produced more pairs, or a smaller PSD), we went back to the previous bracket (backtracking) to pair it again, looking for better downfloaters. This is of course equivalent to verify that the floaters produced are the best possible choice before starting the pairing of the next bracket.
For the last bracket, where an unsatisfactory pairing means the impossibility to complete the pairing, the backtracking could be more complicate. First, when pairing the last bracket, a simple backtracking to the previous one was not always enough. Sometimes we had to join (“collapse”) those two brackets, in order to be able to gain access to the preceding bracket and change its floaters
 and sometimes this process had to be repeated until an acceptable pairing was found.
It is readily evident that this backwards course had to go up, starting from the last bracket, until the point was reached, in which the produced downfloaters did actually allow the pairing of the rest of the players. Hence, the backtracking did necessarily extend until it reached the bracket that, with the lookahead methodology, is at once defined as the PPB  thus bringing us back to the same conditions. The new lookahead method is then equivalent to the backtracking  with the advantage of a fairly simpler logic.
Anyway, the new wording of the Rules does not specify any particular method to enforce compliance with the pairing criteria. Hence, both the arbiter and the programmer enjoy complete freedom in choosing their preferred method to implement the system (lookahead, backtracking, weighted matching or other), as long as the rules are fully complied with.
This section’s goal, from the Rules standpoint, is to univocally define the sequence of generation for the candidate pairings  and, to this aim, it precisely defines the constraints inside which the pairing must be built. From the arbiter’s point of view, however, this section may also be used as a roadmap to actually build the pairing and evaluate its quality. In fact, it can be readily adopted as a guideline to make  or, far more often, prove  a pairing.
Note: MaxPairs is usually equal to the number of players divided by two and rounded downwards. However, if, for instance, M0 is greater than the number of resident players, MaxPairs is at most equal to the number of resident players.
Note: M1 is usually equal to the number of MDPs coming from the previous bracket, which may be zero. However, if, for instance, M0 is greater than the number of resident players, M1 is at most equal to the number of resident players. Of course, M1 can never be greater than MaxPairs
To make the pairing, each bracket will be usually divided into two subgroups, called S1 and S2.
S1 initially contains the highest N1 players (sorted according to A.2), where N1 is either M1 (in a heterogeneous bracket) or MaxPairs (otherwise).
S2 initially contains all the remaining resident players.
When M1 is less than M0, some MDPs are not included in S1. The excluded MDPs (in number of M0  M1), who are neither in S1 nor in S2, are said to be in a Limbo.
Note: the players in the Limbo cannot be paired in the bracket, and are thus bound to double float.
In a given bracket we have a given number M0 of MDPs13 (possibly none), but we have no certainty that all those MDPs can be paired14.
Thus, we define a second parameter M1, representing the number of MDPs that can actually be paired  where, of course, M1 is less than or equal to M0. In summary, the bracket will contain MaxPairs pairs, at most M1 of which contain a downfloater.
We want also to observe that, while M0 is a well known constant, we usually do not know precisely how many players, and especially MDPs, can be paired, until the actual pairing is made – actually, we need to “divine” M1 and MaxPairs out of sound reasoning, assuming a tentative value, which might initially be wrong. Nonetheless, those numbers, however identified, are considered constants  and that is why there is no rule to change them.
The composition of the original subgroups is different when we have MDPs, because those players, having already floated, need now some “special protection”.
In setting the number of pairs to be done to M1 for heterogeneous brackets, we focus only on MDPs, who (or, at least, the maximum possible number of them) actually are to be paired first15. On the contrary, setting the number of pairs to MaxPairs says that we are trying to pair the entire bracket all at once (so it must be homogeneous).
After M1 moveddown players have been selected for pairing, the remaining MDPs, in number M0 M1, cannot be paired in the bracket16. Those players form a subgroup called “Limbo”. During the pairing proceedings, it may happen that some players need to be swapped between S1 and the Limbo  but, at the end of the pairing, the players still in the Limbo will be bound to float again.
13 We want to remember that the “Moveddown players” (MDPs) are the downfloaters of the previous bracket.
14 For example, the number of MDPs may be greater than MaxPairs; or some among them may be incompatible; or we may have a semiincompatibility, in which a group of players ‘compete’ for too few possible opponents, just like the situation described in the comment to A.4.
15 To avoid any misunderstanding, please take notice that this is only a procedural indication that has nothing to do with the order of generation of candidates. In fact, independent of the method and algorithm used to generate them, each candidate is regarded as a whole; and, when we choose the ‘earlier’ candidate from a pool of equivalent ones, we only consider the order of generation of the complete candidates.
16 Those players are not necessarily incompatible in the bracket – there may just be no place to pair them. E.g., if two MDPs share the same one possible opponent, neither of the two is incompatible  but nonetheless one of the two MDPs cannot be paired!
S1 players are tentatively paired with S2 players, the first one from S1 with the first one from S2, the second one from S1 with the second one from S2 and so on.
In a homogeneous bracket: the pairs formed as explained above and all the players who remain unpaired (bound to be downfloaters) constitute a candidate (pairing).
In a heterogeneous bracket: the pairs formed as explained above match M1 MDPs from S1 with M1 resident players from S2. This is called a MDPPairing. The remaining resident players (if any) give rise to the remainder (see A.3), which is then paired with the same rules used for a homogeneous bracket.
Note: M1 may sometimes be zero. In this case, S1 will be empty and the MDP(s) will all be in the Limbo. Hence, the pairing of the heterogeneous bracket will proceed directly to the remainder.
A candidate (pairing) for a heterogeneous bracket is composed by an MDPPairing and a candidate for the ensuing remainder. All players in the Limbo are bound to be downfloaters.
If the candidate built as shown in B.3 complies with all the absolute and completion criteria (from C.1 to C.4), and all the quality criteria from
C.5 to C.19 are fulfilled, the candidate is called “perfect” and is (immediately) accepted. Otherwise, apply B.5 in order to find a perfect candidate; or, if no such candidate exists, apply B.8.
Here is where we build the candidate pairing. In the most general case, this is done in two steps:
Of course, if the bracket is homogeneous, or if none of the MDPs is pairable (i.e. if M1 is zero), the first step is omitted.
Thus, in general, the candidate comprises three parts:
Having prepared a candidate, we must evaluate its quality; that is, we must check the compliance of the candidate with the pairing criteria given in Section C.
If we are very lucky, it may be “perfect”: in this case, we accept it straight away.
Otherwise, we must apply some changes to try and make it perfect (B.5). If this proves impossible, the last resource is accepting a candidate that, although it is not perfect, is nonetheless the best we can have (B.8). Of course, a candidate that does not comply with the absolute criteria is not even acceptable.
After the pairing is made, and before accepting it and proceeding to the next bracket, we will have to perform a completion test, to check that all the remaining players, including the downfloaters from the bracket just paired, allow the round pairing to be completed (see A.9). If this completion test fails, we define the Collapsed Last Bracket and proceed as explained in A.9.
The composition of S1, Limbo and S2 has to be altered in such a way that a different candidate can be produced.
The articles B.6 (for homogeneous brackets and remainders) and B.7 (for heterogeneous brackets) define the precise sequence in which the alterations must be applied.
After each alteration, a new candidate shall be built (see B.3) and evaluated (see B.4).
The process of pairing is an iterative one: if the pairing is not perfect, we try (one by one) a precise sequence of alterations in the subgroups S1, Limbo, and S2, and each time we repeat the preparation and evaluation of the candidate. There are, in fact, two different sequences:
The first perfect candidate found in this process is the required pairing. If there is no perfect candidate, we shall have to use the best available one; since we are scrutinizing all candidates, we can find this best candidate as we proceed. To do that, when we find the first legal (but not perfect) candidate, we mark it as a “provisionalbest”. Each time we find another legal candidate, we shall compare17 it with the current provisional best candidate. If the former is better than the latter, we store it as the new provisionalbest; otherwise, we keep the old one. In the end, all candidates have been examined; hence, the surviving provisionalbest is actually the best possible (although imperfect) candidate, which will be accepted as pairing, because of rule B.8.
The main guideline to carry out this task is the “minimum disturbance”: every alteration must be the minimum possible, so that the resulting pairing can be as similar as possible to a “perfect” one.
For more detail about the iterative pairing process, see B.6 and B.7.
Alter the order of the players in S2 with a transposition (see D.1). If no more transpositions of S2 are available for the current S1, alter the original S1 and S2 (see B.2) applying an exchange of resident players between S1 and S2
Since we are now managing only homogeneous brackets, we do not need to worry about pairing MDPs.
The possible actions to be tried here are:
17 Two candidates are compared based on the compliance with the pairing criteria, which are defined in order of priority in section C. The first check is on the priority of the higher infringed criterion: the higher it is, the lower is the quality of the candidate. Then the second check is on a “failure value” which is peculiar to that criterion – this will often be the number of times the criterion is infringed (e.g., the numbers of disregarded colour preferences) but it may also be of a completely different nature (e.g., the PSDs of two candidates to be compared). Then we go to the second higher infringed criterion; then to the latter’s failure value  and so on until we find a difference. When there is no difference at all, the first generated candidate takes precedence.
(see D.2) and reordering the newly formed S1 and S2 according to A.2.
After we made transpositions in a bracket, alterations in the order are desired; hence, players in the S2 subgroup should not be sorted again (while S1 does not need to be sorted, as it has not been changed).
On the contrary, after exchanges, which swap one or more players between subgroups S1 and S2, we must sort both subgroups S1 and S2 according to A.2, to reestablish a correct order before beginning a new sequence of pairing attempts. If the first attempt of the new exchange fails to give a valid result, we will try transpositions too, thus changing the natural order in the modified S2.
Both transpositions and exchanges should not be applied at random: to comply with the general principle of minimal disturbance of the pairing, section D dictates a precise sequence of possible transpositions and exchanges. This sequence begins with alterations that give only mild disturbances to the pairing (with respect to the “natural” one), moving gradually towards those changes that cause definitely important effects.
The order of actions is as follows: first, we try, one by one, all the possible transpositions (see D.1). If we find one that allows a perfect pairing, the process is completed. Otherwise, we try the first exchange (see D.2): with this, we proceed
Operate on the remainder with the same rules used for homogeneous brackets (see B.6).
Note: The original subgroups of the remainder, which will be used throughout all the remainder pairing process, are the ones formed right after the MDPPairing. They are called S1R and S2R (to avoid any confusion with the subgroups S1 and S2 of the complete heterogeneous bracket).
If no more transpositions and exchanges are available for S1R and S2R, alter the order of the players in S2 with a transposition (see D.1), forming a new MDPPairing and possibly a new remainder (to be processed as written above).
again to try every possible transposition18, until we succeed  or use them up. In the latter case, we try the second exchange, once again with all the possible transpositions, and so on.
If we get to the point in which we have used up all the possible transpositions and exchanges, then a perfect pairing simply does not exist. In that case, we apply B.8, thus accepting a less than perfect result.
This article, a companion to the previous one, addresses the case of heterogeneous brackets. This kind of bracket is paired in two logical steps19:
The rules to operate on the remainder are just the same that apply for a homogeneous bracket. The difference shows only when we reach the point in which all of the possible transpositions and exchanges in the remainder have been unsuccessfully tried.
In a homogeneous bracket, this is the moment when we lower our expectations, settling for a less than perfect pairing (see B.6). In a heterogeneous bracket, however, we are not yet ready to surrender: before laying down arms, we can try to change the composition of the remainder.
To do that, we try a new, different MDPpairing by applying a transposition to the original subgroup S2 (viz. the subgroup S2 of the complete
18 Suppose we exchanged player A from S1 with player B from S2. After the exchange, player B, now in S1, has a rank that is lower than that of player A, now in S2. As transpositions proceed, we will get to a point in which the candidate puts together players B and A – and then of course some other pairs of players. Now, before making the exchange, we tried all transpositions in S2, and thus also the one which contains the pair AB and all the same other pairs as well – in summary, this candidate has already been evaluated! Reasoning along the same lines, we reach the conclusion that the same holds true also for exchanges involving more players. We can thus deduce that every time a pair contains a player from S1 with a lower rank (higher BSN) than its opponent from S2, this pair belongs to a candidate that has already been evaluated, and therefore we do not need to evaluate it again.
19 Of course, a practical implementation need not necessarily compose the pairing in two steps, as long as the final effect is the same as specified by the rules. bracket, not that of the remainder!). This may leave us with a new, different remainder, which we process (just as described above) trying to find a complete pairing – and, if we have no success, we try transposition after transposition until we succeed, or exhaust them all20
If no more transpositions are available for the current S1, alter, if possible (i.e. if there is a Limbo), the original S1 and Limbo (see B.2), applying an exchange of MDPs between S1 and the Limbo (see D.3), reordering the newly formed
As we hinted above, the PPB and the CLB are subject to slightly different pairing rules: the downfloaters of the PPB are no longer required to optimise the pairing in the next bracket (as it would be for normal brackets, see C.7), but just to allow it (see C.4). With those downfloaters, together with the SCS, we build the CLB, which is (by definition) the last bracket.
This is a rather unusual bracket: it is by definition heterogeneous21, and its residents often have different scores (because they come from the SCS). Its pairing is different from that of the usual heterogeneous bracket in that we have a remainder that must be paired just like if it were homogeneous, but without disregarding the needs of players with different scores.
Thus, we must enforce some criteria that usually are not important in remainders. The main goal in pairing the CLB is to get the lowest possible PSD (because, basically, the number of pairs is determined by the number of PPB floaters). To find this minimum PSD, we have to look not only at the MDP(s) and at their opponents (as usual), but also at the pairs that can be made inside the remainder (i.e. between SCS residents).
When several candidates have the lowest possible PSD, we must also enforce some criteria for the remainders, which are not usually required. If in a pair there are players with different scores, to such players we must apply all those criteria that limit the repetition of floats [C.12 to C.15] and the score difference of the protected players whose protection has already failed once or more [C.16 to C.19].
If all the possible transpositions have been used up, we have a resource yet: trying to change the MDPs to be paired. Of course, this is only possible if there is a Limbo in the bracket. In this case, we can exchange one or more of the MDPs with a same number of players from the Limbo. This is called an MDPexchange (see D.3).
20 Actually, we do not need to try all of the transpositions, because not all of them are meaningful: in fact, we only have to try those transpositions that actually change the players, or their order, in the first part of the subgroup S2 – i.e. those players, who are going to be paired with the MDPs from S1. All the other players in S2 do not take part in this phase of the pairing and are thus irrelevant (at least for the moment).
21 Remember that the CLB is born from a failure in a completion test. This means that the “rest of the players”, with the current downfloaters (possibly none!) from the just (unsuccessfully!) paired bracket, cannot be paired  it therefore requires some adequate MDPs.
S1 according to A.2 and restoring S2 to its original composition.
Choose the best available candidate. In order to do so, consider that a candidate is better than another if it better satisfies a quality criterion (C5 C19) of higher priority; or, all quality criteria being equally satisfied, it is generated earlier than the other one in the sequence of the candidates (see B.6 or B.7).
After any MDPexchange, we are actually pairing an altogether different bracket; hence, we need to reorder S1 and restore S2 to its original composition, in fact starting the pairing process anew. As it was for the homogeneous case, the MDPexchanges must be tried in the correct sequence, one by one; and, for each one of them, we shall try all the possible transpositions in S2, thus generating a different remainder  that will of course have to undergo all the usual pairing attempts as described above.
This is where we must make ourselves content with what best we can: if we arrive here, we have already tried all possible transpositions and exchanges, only to reach a simple, if dismal, conclusion  there is no perfect candidate! Hence, we choose the best available candidate, which is the final provisionalbest found during the evaluation of all candidates as illustrated in B.5.
The Sieve Pairing
A very interesting alternative to this method – not necessarily a practical one, but very important from the theoretical point of view – is the one we shall call “Sieve pairing” (because of its similarity with the famous Eratosthenes' Sieve).
The basic idea is very simple: we build all the possible acceptable pairings (i.e., all those that comply with the absolute criteria). Then we start applying all the pairing criteria, one by one  but this time we start with the most important one and proceed downwards.
Each criterion will eliminate part of the acceptable pairings, so that, as we proceed, the number of candidates becomes lower and lower. If, at some stage of the process, only one candidate remains, we choose that one – it may even be a rather bad one, but there is nothing better.
If, after applying all the pairing criteria, we are left with more than one candidate, then we choose the one that would be the first to be generated in accordance with the sequence defined by Section B.
The absolute criteria correspond to the requirements of Section C.04.1, “Basic Rules for Swiss Systems” in the FIDE Handbook, which we may want to look at closely.
No pairing shall violate the following absolute criteria:
See C.04.1.b (Two players shall not play against each other more than once
Those criteria must be complied with always: they cannot be renounced, whatever the situation22. To enforce them, players may even float down as needed.
If the game is won by forfeit, for the purposes of pairing those two players have never met. As a result, that pairing may be repeated later in the tournament (and sometimes this happens, too!).
22 There are however situations in which no pairing at all exists, which complies with the absolute criteria – in such cases, the arbiter must apply his better judgment to find a way out of the impasse (see A.9).
See C.04.1.d (A player who has already received a pairingallocated bye, or has already scored a (forfeit) win due to an opponent not appearing in time, shall not receive the pairingallocated bye)
Nontopscorers (see A.7) with the same absolute colour preference (see A6.a) shall not meet (see
C.04.1.f and C.04.1.g)
If the current bracket is the PPB (see A.9): choose the set of downfloaters in order to complete the roundpairing
Please notice that, contrary to the previous rules, only PABs and forfeit wins prevent the allocation of a PAB (see A.5) – on the contrary, a player who received a requested bye (usually, half point) may receive the PAB in a subsequent round.
This criterion does not apply to topscorers (A.7) or topscorers’ opponents, who are the only possible exception to C.04.1.f/g.
Two players, who cannot be paired to each other without infringing criteria C.1 or C.3, are said to be incompatible.
This is an absolute criterion too, but it applies only to the processing of the PPB – hence, only after a completion test failure (see A.9). Contrary to ordinary brackets (whose downfloaters are chosen in order to optimise the pairing of the next bracket  see C.7), for the PPB we just require a choice of downfloaters that allows a completion of the roundpairing  independent from the optimization of the next bracket, which is of course the CLB, and hence must be completely paired.
Please note that, since C.4 precedes both C.5 and C.6, the compliance with this criterion may cause a reduction in the number of pairs, or an increase in the final PSD, with respect to the previous pairing23.
The above criteria set conditions that must be obeyed: a candidate that does not comply with them is discarded. The following criteria are of a different kind, in that they establish a frame of reference for a quantitative evaluation of the “goodness” of the pairings, by setting a sequence of “test points” in order of decreasing importance, according to the internal logic of the system. The level of compliance with each one of the following criteria is not a binary quantity (yes/no) but a numerical (integer or fractional) quantity. We will measure it by means of a “failure value”, whose meaning is of course tightly connected to the criterion itself (e.g., the number of pairs less than MaxPairs for C.5, or the number of players not getting their colour preference for C.10, and so forth).
When we compare two candidates, we in fact compare the failure values of the candidates for each criterion, one by one, in the exact sequence given by the Rules. If the two failure values are identical, we proceed to the next criterion. If they are different, we keep the candidate with the better value and discard the other one.
It seems worth noting that a candidate having a better failure value on a higher criterion is selected, even if the failure values for the following criteria are far worse. In other words, the optimisation with respect to a higher criterion may have a dramatic impact on the remaining failure values – and,
23 Of course, since the bracket we are pairing is a PPB, it has already been paired once.
we may add, the optimisation with respect to a criterion is always only relative to the current status, because even a small difference in a higher criterion may change the situation completely.
To obtain the best possible pairing for a bracket, comply as much as possible with the following criteria, given in descending priority:
Maximize the number of pairs (equivalent to: minimize the number of downfloaters).
Relative criteria are not so important as absolute ones, and they can be disregarded, if this is needed to achieve a complete pairing. In general, they are not important enough to make a player float – in fact, the first one of them, and hence the most important, instructs us to do just the very opposite, minimising the number of downfloaters!
Apart from the remaining player in odd brackets, only incompatible (or semiincompatible) players should float. This too is an evidence of the attention of the FIDE (Dutch) system towards the choice of the “right strength opponent”.
The first “quality factor” is of course the number of pairs, a reduction of which increases the number of floaters (and, usually, also of the overall score difference between players).
Maximising the number of pairs actually means, build MaxPairs pairs (see B.1). At the beginning of the pairing process, though, MaxPairs, or the maximum number of pairs that can be built (which is a constant of the bracket), is actually unknown – hence, we need to "divine" it.
Actually, the only things we know for sure are the total number N of players in the bracket, and the number M0 of MDPs entering the bracket. We want to observe that the number of pairs can never be greater than N/2; thus, this value should make a good starting point, independent of the kind of bracket (homogeneous or heterogeneous).
The actual value of MaxPairs can be less than that, because some players might be impossible to pair in the bracket. Moreover, if this bracket is a PPB, it must also provide the downfloaters required to complete the roundpairing (see C.4), and that might detract to the number of pairs that can actually be built. Hence, the process to determine MaxPairs value is somewhat empirical and may require some “experimenting”.
If the bracket is heterogeneous (M0≠0), then as many MDPs as possible (M1) must be paired. They will be paired first, before proceeding with the rest of the players (see B.3)  but, as it happened for the value of MaxPairs, we still do not know the true value of M1, and we must divine it too. A first educated guess for its value is M0 – minus, of course, any incompatible MDPs.
If there is no way to make all those pairs, our estimate of the value of M1 was apparently too optimistic – in this case, we will have to gradually decrease it, until we succeed. Any remaining MDPs join the Limbo (see B.2) and shall
Minimize the PSD (This basically means: maximize the number of paired MDP(s); and, as far as possible, pair the ones with the highest scores)
If the current bracket is neither the PPB nor the CLB (see A.9): choose the set of downfloaters in order first to maximize the number of pairs and then to minimize the PSD (see C.5 and C.6) in the following bracket (just in the following bracket).
eventually float (after the completion of the pairing for the bracket).
The number of pairs made in the MDPpairing will be subtracted into the total number of pairs to be made in the bracket, yielding the (plausible) number of pairs to be built in the remainder24.
Here too applies the same line of reasoning: if we cannot make all those pairs, our initial estimation of MaxPairs was apparently too optimistic – hence, we will have to gradually decrease their number. Any remaining players become downfloaters, and will eventually float down into the next bracket.
The same line of reasoning also holds for a homogeneous bracket  which, by definition, contains no Limbo or MDPs, but is otherwise essentially similar to a remainder.
In heterogeneous brackets, even when the same number of pairs is made, different choices of floaters, or different pairings, can lead to different mismatching between players’ scores (for an example, see the many possible ways to pair a heterogeneous bracket containing many players all having different scores). This important criterion, directly related to rule C.04.1:e, directs us to minimise the overall difference in scores. Its location before the colour related criteria (C.8C.11) is suggestive of the attention the FIDE (Dutch) system gives to the choice of a “right strength” opponent rather than a “right colour” one.
The method to compute and compare the PSDs is explained in detail in the comment to article A.8.
When we get here, we have already complied with the absolute criteria (hence the pairing is a legal one) and optimised the most important pairing quality parameters (number of pairs, PSD).
Before going ahead to optimise colours and MDPs treatment, we take a look ahead to the next bracket. We do not want to ever come back to the current bracket again. Thus, we must make sure that the choice of downfloaters we are going to send to the next bracket will be the best possible one to comply with C.5 and C.6.
First, we check that the downfloaters (which will be the MDPs of the next bracket) will allow us to compose the maximum possible number of pairs.
24 We always want to remember that the pairing of the MDPs and of the remainder are two phases of a single operation, which is performed as a unit. Thus, we do not “go back” from the remainder pairing to the MDPpairing, because we are already inside the same operation. For example, let us suppose that the current bracket produces only one downfloater and that the next scoregroup contains an odd25 number of players, one of which has no possible opponent. If we can choose between two possible downfloaters, both compatible in the destination bracket, but only one of them can be paired to the “problematic” player, we must choose that one  because choosing the other one would leave an incompatible player (and hence an unavoidable downfloater!) in the destination target. Only when the number of pairs have been maximised, we proceed to look into the PSD in the destination target. This in practice means that, when we may choose between two or more possible downfloaters, if all other conditions are equivalent, we must choose the downfloater that may be paired with the lowest score difference26. This optimisation is to be extended only to the next bracket. Actually, there are situations in which a small change in a previous pairing would bring in large benefits  but looking several brackets ahead would be too much difficult an operation to be carried on every time. So the rules settle for a practical optimisation, renouncing those that are out of reasonable reach. But the reason is not only this one: in the basic philosophy of the FIDE (Dutch) system, the pairings for the higher ranked players are considered far more important than those for the lower ones. Hence, altering the pairing of the current bracket for the benefit of some player, who is located two brackets below this one, would simply be opposite to that philosophy.
Having already made sure that both the number of floaters and their scores are at a minimum, we now start to optimise colour allocation. Actually, colour is less important than difference in score – and that’s why, consistently with the basic logic of the system, the colour allocation criteria are located after those that address number of pairs and PSD.
Minimize the number of topscorers or topscorers' opponents who get a colour difference higher than +2 or lower than 2.
Article C.3, in accordance with C.04.1:fg, states that when two nontopscorers meet, their absolute preferences must be complied with. Here we have the special case of a topscorer who, for some reason, is bound to be paired with a player (who may or may not be also a topscorer)
25 Please note that if the next scoregroup contained an even number of players, the bracket built with it and the current downfloater would be odd. Hence, it would in any case produce (at least) one downfloater and the choice of the MDP would not be critical for the number of pairs.
26 Since this criterion does not apply for the PPB, the next bracket's resident players will all have the same score. Thus, it is not possible for moveddown players to be paired with players having different scores  but, if they cannot be paired in the bracket, they will have to float again, and this makes the PSD change!
Minimize the number of topscorers or topscorers' opponents who get the same colour three times in a row.
Minimize the number of players who do not get their colour preference.
having the same absolute preference. The outcome of those players’ games may be very important in determining the final ranking and podium positions; and this is an exception explicitly provided for by C.04.1:fg, so we may compose such pairs. Thus, we choose the best possible matched opponent – but there must not be more such pairs than the bare minimum.
The subdivision into two individual rules establishes a definite hierarchy, giving more importance to colour differences than to repeating colours. Suppose that, for one same opponent, we can choose between two possible topscorers, and all those players have the same absolute colour preference. In this case, we must select the components of the pair in such a way that colour differences are minimised (as far as possible).
As hinted above, a player, who has an absolute colour preference without being a topscorer, may happen to be paired with a topscorer having an identical absolute colour preference. These two rules equate the players of the pair  thus, a player might be denied its absolute colour preference just as if it were a topscorer, even if it is not one!
We can have an idea about the minimum number of players who cannot get their colour preference, by inspecting the bracket, prior to the pairing.
Let us suppose that m players prefer a colour and n players prefer the other one, with m ≥ n. We can thus compose no more than n pairs in which the players are expecting different colours; and the colour preferences in these pairs can  and must  be satisfied.
The remaining mn players all expect the same colour; and they will have to be paired among themselves. In each of the pairs thus composed, one of the two players cannot get its preferred colour. The number of such pairs, and henceforth of such players too, is x=(mn)/2, rounded downward to the nearest integer if needed. Sometimes, in addition to those m+n players, the bracket contains also a more players who have no colour preference at all. Those players may get any colour, but, of course, they will usually get the minority colour, so that they will subtract to the number of disregarded preferences. Taking one more step further, we may reason that we can build a maximum of MaxPairs pairs. Among those, n+a pairs can satisfy both the colour preferences, whilst the remaining x=MaxPairs na cannot help but disregard one colour preference. Of course, x cannot be less than zero
Minimize the number of players who do not get their strong colour preference.
(a negative number of pairs has no practical meaning); thus, we obtain the final and general definition for x:
x = max (0, MaxPairsna)
Please take notice that a perfect pairing always has exactly x disregarded colour preferences – no more, no less.
Actually, there might be even more pairs in which a player does not get its preference  because of incompatibilities due to absolute criteria, as well as “stronger” relative ones. Thus, at first we propose to make the minimum possible number of such pairs – but we may need to increase this number, to find our way around various pairing difficulties.
Since the general philosophy of the FIDE (Dutch) system gives more importance to the correct choice of opponents than to colours, the pairs containing a disregarded colour preference will typically be among the first to be made27.
Only now, having maximised the number of “good” pairs, we can set our attention to satisfying as many strong colour preferences as possible.
The minimum number of players not getting their strong colour preference, which is usually represented by z, is of course a part of the total number x of disregarded colour preferences (see note to C.10) – therefore, z is at most equal to x.
For instance, let the number WT of white seekers be greater than the number BT of black seekers (we call White “the majority colour”). The x players will all be White seekers, and as many as possible among them should have mild colour preferences, while the rest will have strong colour preferences28. Hence we can estimate z simply as the difference between x and the number WM of White seekers who have a mild colour preference, with the obvious condition
27 Actually, transpositions swap players beginning with the last positions of S2 and going upwards, causing the bottom pairs of the bracket to be modified early in the transposition process, while the top pairs are modified later. Hence, a “colourdefective” pair located at the bottom of the candidate has a higher probability to be changed soon than a similar pair located at the top – therefore, perfect pairings with top “colourdefective” pairs have a definitely higher probability. Incidentally, we might also mention that players often seem to worry about “colour doublets” (like, for example, WWBB) and think that such colour histories are more frequent with the FIDE (Dutch) system than with other Swiss pairing systems. This is not so. In fact, such histories are usual enough (and unavoidable) in all manners of Swiss pairings – in the FIDE (Dutch) system they may seem more frequent just because they appear more often in the top pairs of the bracket, therefore involving higher ranked players, which makes them more noticeable.
28 We want to notice that, during the last round, some absolute colour preferences might be disregarded for topscorers or their opponents (see C.8, C.9), so that part of x may represent such players. In those instances, our line of reasoning should be suitably adapted. that z cannot be less than zero; hence:
z = max ( 0, x – WM ) if WT ≥ BT (White majority)
z = max ( 0, x – BM ) if WT < BT (Black majority)
With a careful choice of transpositions and/or exchanges, we might be able to minimise the number of disregarded strong preferences29.
For several reasons, however, the number of players who cannot get their strong preference may be greater than that.
The following group of criteria optimises the management of floaters, which is the last step towards the perfect pairing.
Minimize the number of players who receive the same downfloat as the previous round.
Minimize the number of players who receive the same upfloat as the previous round.
Minimize the number of players who receive the same downfloat as two rounds before.
Minimize the number of players who receive the same upfloat as two rounds before.
Minimize the score differences of players who receive the same downfloat as the previous round.
Minimize the score differences of players who receive the same upfloat as the previous round.
Rule C.04.1:e states that, in general, players should meet opponents with the same score. This is (of course) best achieved by pairing each player inside its own bracket. However, there are some situations, in which a player cannot be paired in its bracket  and then, by necessity, must float. These criteria limit the frequency with which such an event can happen to a same player
 but they are “very weak criteria”, in the sense that they are almost the last to be enforced  and almost the first to be ignored in case of need.
Here, each criterion establishing a certain protection for downfloaters is immediately followed by a similar one establishing the very same protection for upfloaters.
Because of this, there is a certain residual asymmetry in the treatment; viz. downfloaters are (just a little bit) more protected than upfloaters. Please note that, in some other Swiss systems, floaters’ opponents are not considered floaters themselves, and therefore enjoy no protection at all.
The four previous rules minimised the number of players who, having floated in the last two rounds, may get a float again in this round. However, those rules do not give any special protection either to a player who, being already a MDP in a bracket (in this round), cannot be paired and must float down again, or to its opponent. Such players, and their opponents, will
29 Of course, since the total number of disregarded preferences must remain the same (we cannot have it smaller, and do not want it to grow larger!), this may only happen at the expense of a same number of mild preferences. A brief example may shed some light on the matter. Consider the bracket {1Bb, 2b, 3Bb, 4b}, where we have x=2, but z=0. The latter means that we can build the pairs in such a way that any one of them contains no more than one strong colour preference – and, in fact, a simple transposition allows us to obtain just this result.
Minimize the score differences of players who
receive the same downfloat as two rounds before.
Minimize the score differences of players who receive the same upfloat as two rounds before.
The criteria C16C.19 are for protected players whose protection has already failed once or more, and try to prevent such players from further floating. When we must make some players float down, we try, as long as possible, to choose those players who are not MDPs. Sometimes, however, this is not possible, and we must make some MDP float down. In this case, we should, as far as possible, choose those MDPs that are not (or are least) protected because of previous floats. Of course, the same holds (almost) symmetrically for the MDPs’ opponents.
For example, in a CLB (see A.9) that contains players with many different scores, the effect of these rules is that, if we have two possible prospective floaters and only one of them is protected, we try to pair the latter with a SD as little as possible30.
This section states the rules to determine the sequence in which transpositions, exchanges, and MDPexchanges must be tried, in order to generate the candidates in the correct order. The general basic principle is, as always, that of “minimal disturbance” of the pairing. This means that we have always to move that player (or those players) whose displacement will cause the least possible difference of the pairing from the “natural” one31  while at the same time allowing the best possible quality of the pairing itself.
Before any transposition or exchange take place, all players in the bracket shall be tagged with consecutive inbracket sequencenumbers (BSN for short) representing their respective ranking order (according to A.2) in the bracket (i.e. 1, 2, 3, 4,...).
A transposition is a change in the order of the BSNs (all representing resident players) in S2.
All the possible transpositions are sorted depending on the lexicographic value of their first N1 BSN(s), where N1 is the number of BSN(s) in S1 (the remaining BSN(s) of S2 are ignored in
The use of pairingids, in this phase, may sometimes be confusing. Therefore, we give temporary sequence numbers to the players, as a very handy remedy to simplify the application of the rules below.
All transpositions are sorted or compared based on the dictionary (“lexicographical”) order, so that one given transposition precedes or follows another one if the string formed by the players BSNs of the first one precedes or follows that of the second one. The method to compare the strings is the very same already illustrated for the comparison of PSDs 32.
30 Another example is the case of two MDPs with different scores, and a protected resident who must be paired with one of those two MDPs: the resident should be paired to the MDP who has the lower score of the two.
31 But, to avoid misunderstandings, we should keep in mind that any change in the order in S2 (transposition) is by definition preferable to even a single exchange between S1 and S2.
32 See the comment to C.6 [page 24] for details. Please note that the use of alphabet letters would be completely equivalent to that of numbers, at least for brackets with less than 26 players. The use of numbers, however, allows an identical treatment for all brackets, whatever the number of players they contain.
this context, because they represent players bound to constitute the remainder in case of a heterogeneous bracket; or bound to downfloat in case of a homogeneous bracket  e.g. in a 11 player homogeneous bracket, it is 678910, 6 78911, 6781011,..., 6111098, 7689
10,..., 1110987 (720 transpositions); if the bracket is heterogeneous with two MDPs, it is: 3 4, 35, 36,..., 311, 43, 45,..., 1110 (72
transpositions)).
The subgroup S1 may or may not have the same number of players as S2. For the comparison to have a meaning, we must define the number of elements of each of the two strings of BSNs that we are comparing.
We are looking for mates for each element in S1 (which of course represent a player each). Thus, we consider the number N1 of elements in S1– while the remaining players are (for the moment) irrelevant.
A simple example will help us clarify the matter: consider a heterogeneous bracket {S1=[1]; S2=[2, 3, 4]}. All the possible transpositions of S2 (properly sorted, and including the original S2) are:
[2,3,4]; [2,4,3]; [3,2,4]; [3,4,2]; [4,2,3];
[4,3,2]33.
As we want to pair #1 with the first element of S2, it is at once apparent that [2,3,4] and [2,4,3] have the very same effect34; and the same holds for [3,2,4] and [3,4,2]; and for [4,2,3] and [4,3,2]. Hence, the actual sequence of transpositions is as follows (elements between braces “{…}” are ‘irrelevant’ and are ignored in this phase):
[2]{3, 4}; [3]{2, 4}; [4]{2, 3}
An exchange in a homogeneous bracket (also called a residentexchange) is a swap of two equally sized groups of BSN(s) (all representing resident players) between the original S1 and the original S2.
In order to sort all the possible resident exchanges, apply the following comparison rules between two residentexchanges in the specified order (i.e. if a rule does not discriminate between two exchanges, move to the next one).
The priority goes to the exchange having:
The exchanged sets must of course have the same size  because, were it not so, we would be changing the sizes of S1 and S2.
However, to evaluate the “weight” of the change, we must take into consideration not only the size of the exchanged sets but also the choice of players. To do that, we need a set of criteria addressing the various aspects of this choice. The aim is, as always, the “minimal disturbance” – viz. to try and have a pairing as similar as possible to the natural one.
The first criterion is, of course, the number of involved players: the less, the better!
From a theoretical point of view, all players in S1 should be stronger than any player in S2 is.
33 Please note that, in the very simple case where every BSN is a single digit, the string may be interpreted as a number, which becomes larger and larger as we proceed with each new transposition: 234, 243, 324, 342, 423, 432.
34 Of course, this equivalence is in no way general – it depends only on the fact that we are looking for just one element!
to S1 and the sum of the BSN(s) moved from the original S1 to S2 (e.g. in a bracket containing eleven players, exchanging 6 with 4 is better than exchanging 8 with 5; similarly exchanging 8+6 with 4+3 is better than exchanging 9+8 with 5+4; and so on).
Therefore, when we have to swap two players across subgroups, we try to choose the weakest possible player in S1 and swap it with the strongest possible one from S2.
To do so, we can use the BSNs to choose a player as lowranked as possible from S1, and a player as highranked as possible from S2, and then swap them, assuming that a higher rank should indicate a stronger player.
Thus, the difference between exchanged numbers is (or, at least, should be) a direct measure of the difference in (estimated) strength and should therefore be as little as possible.
When two possible choices of players to be exchanged show an identical difference in the sum of their respective BSNs, we choose the set which disturbs S1 as little as possible, i.e. the one in which the (highest BSN) player from S1 has a lower rank.
In the example, 52 is better than 43 because exchanging #5 is better than exchanging #4. Similarly, (5,4,1) is a better choice than (5,3,2), because exchanging #4 is better than exchanging #335.
Finally, having optimised the difference in ranking and the disturbance in S1, we can optimise the disturbance in S2 too.
Contrary to S1, now we try to exchange the lower possible BSNs. Hence, 69 is better than 78, because exchanging #6 is better than exchanging #7 – and so forth.
An exchange in a heterogeneous bracket (also called a MDPexchange) is a swap of two equally sized groups of BSN(s) (all representing MDP(s)) between the original S1 and the original Limbo.
Here we are changing the composition of the set of pairable MDPs. Of course, this alteration may only occur when M1 < M036, because only in this situation does a Limbo exist. This means that we must choose which MDPs to exclude from the pairing. Sometimes the decision is easy – e.g. there may be some incompatible MDP, and we may have no choice at all37.
35 Sometimes, just as it happens in the above example, we might end up exchanging a higherranked player, as a side effect of enforcing the exchange of the lowest possible player. To understand this, we want to remember that, in the exchange, we do not operate on “several single players” but on a whole set of them, and we just have to decide if a set is better or worse than another one. In this case, (5, 4, 1) is better than (5, 3, 2) – therefore, we exchange #1, who is the topplayer, because this is the way to exchange #4 rather than #3.
36 See B.1, p. 15.
37 We want to remember that, because of C.7, the downfloaters from the previous bracket (i.e. the MDPs of the current bracket) have already been optimised. Thus, if we have an incompatible here, it means that there was no alternative at all. Hence, there is no going back to the previous bracket (“backtracking”).
In order to sort all the possible MDPexchanges, apply the following comparison rules between two MDPexchanges in the specified order (i.e. if a rule does not discriminate between two exchanges, move to the next one) to the players that are in the new S1 after the exchange.
The priority goes to the exchange that yields a S1 having:
Any time a sorting has been established, any application of the corresponding D.1, D.2 or D.3 rule, will pick the next element in the sorting order.
When we have a choice, we start by trying to pair as many MDPs as possible, and as high ranked as possible [B.2]. If we must change this original composition, we need to apply an MDP exchange. The following criteria allow us to determine the priority among all the possible exchanges. Please note that this result is achieved by inspecting the composition of the new S1, not that of the Limbo. To a hasty reader, it might seem that, pairing a player with lower score would yield a lower score difference, and thus a lower PSD. Of course, this is definitely wrong! When we put a higher scored player in the Limbo, that player will float – hence, the corresponding SD, which is calculated with the artificial value defined in A.8, will be very high. To minimise the PSD, the Limbo must contain a minimum of players, and those must have as low a score as possible. Hence, complying with C.6, which instructs us to minimise the PSD, automatically satisfies this criterion too.
We also want to take notice that the number of exchanged players is not allimportant. For example, consider an S1 with three players and a Limbo with two: in some circumstances, exchanging the two lower ranked players may give better results than exchanging just the top one.
This is the criterion we must strive to comply with. When the involved players have the same scores, we have to choose the lower ranked players. This is easily accomplished by comparing the BSNs of the players comprised in S1 after the exchange  in the very same way as we did in the previous cases.
If we are lucky enough, the first attempt to a transposition, exchange, or MDPexchange will yield the desired result. Often, though, we must persevere in the attempts until we get a successful one. In this case, we must follow the order (sequence) established by the three rules above illustrated.
Ideally, we should start by establishing a full list of all the possible transformations  be them transpositions or exchanges of any kind  sorting that list by D.1, D.2 or D.3 (as the case may be), and then trying one after another until we find the first useful one38.
38 In common practice, exchanges and transpositions will be tried together (for each exchange, we will likely try one or more transpositions). To avoid mistakes, it is most advisable to annotate the last transformation (of each kind) used so that, on the following attempt, we can be sure about which element of the sequence is the next one.
It is the colour determined by drawing of lots before the pairing of the first round.
For each pair apply (with descending priority):
Grant both colour preferences.
Grant the stronger colour preference. If both are absolute (topscorers, see A.7) grant the wider colour difference (see A.6)
Taking into account C.04.2.D.5, alternate the colours to the most recent time in which one player had white and the other black.
The Initialcolour is not referred to any particular player. It is actually a parameter of the tournament – and the only one left to fate! – that allows the allocation of the correct colour to each player who has not a preference yet.
When two absolute preferences are involved, rule
1: WWBWBW
2: BBWBWW
Here, player #1 has a colour difference CD=+2, while player #2 has CD=0. Thus, we try to equalize the colour differences by assigning to player #1 his preferred colour.
Please note that this rule applies only to pairs in which both players have an absolute preference, while in all other cases the rule does not apply – e.g., in the pair:
1: BWWBWBW (strong preference, CD=+1) 2: =BBWBWW (absolute preference, CD=0)
the absolute preference shall be satisfied, no matter how large the colour difference is.
To correctly manage colour assignments when one or both players have missed one or more games, we often need comparing colours histories by means of rule C.04.2:D.3.
For example, in the comparison between the colours histories of two players, the sequence
== WB is equivalent to BWWB and WBWB (but the latter two are not equivalent to each other!).
Grant the colour preference of the higher ranked player.
Note: Always consider sections C.04.2.B/C (Initial Order/Late Entries) for the proper management of the pairing numbers.
We may want to pay particular attention to this point: in all other conditions being equal, the higher ranked player gets not white but its own preferred colour!
When we get here, both players of the pair have no colour preference. Therefore, we use the Initialcolour decided by lot before the start of the tournament, to allocate colours to the players.
Of course, this rule will be used always in the first round (obtaining the usual results39), but it will be useful also in subsequent rounds, when we have a pairing between two players who did not play in the previous rounds (e.g. late entries or forfeits).
We ought to remember that players, who are actually entering the tournament only at a given round after the first – and who therefore were not paired in the previous rounds – in fact, do not exist, even if (seemingly) listed in the players’ list. An obvious side effect of this is that we cannot expect all “oddnumbered” and “even numbered” players to have the same colour as would be usual (viz., as they would have in a “perfect” tournament).
Actually, such late entries may have different effects on the pairing numbers, depending on how they are managed.
If we insert all the players in the list straight from the beginning, the pairing numbers will not change on the subsequent rounds, but the pairing of the first round will have to “skip” the absent players. For example, if player #12 is not going to play on the first round, players #13, #15, and so forth, who should seemingly get the initial colour, will actually have the opposite colour; while players #14, #16, and so on will get the initialcolour.
If, on the contrary, we insert a new player only when it actually enters the tournament, we must find the correct place to put it. All the subsequent players will therefore have their pairing numbers changed, in order to accommodate the new entry. For example, if the newly inserted player gets #12, the previous #12 (who had colour opposite to the initialcolour) will now be #13; and so on for all subsequent players.
39 Please note that, if we are using an accelerated pairing system, the usual colour alternation is disrupted unless the first score group contains a number of players multiple of four.
This chapter illustrates a stepbystep example of pairing procedure for a six rounds Swiss tournament by means of the FIDE (Dutch40) Swiss pairing system, in the hope to help those who wish to improve their knowledge of the system or get more familiar with it.
During the FIDE Congress in Abu Dhabi 2015, the Swiss Rules for the FIDE (Dutch) system were partially modified and reworded in order both to avoid misunderstanding in some points and to correct some peculiar behaviours in particular situations  and thus get better pairings in some instances41. In the following Congress in Baku 2016, the Swiss Rules were completely reworded, with the aim to make them clearer and easier – but this time without introducing behavioural changes.
Only a general knowledge of the FIDE (Dutch) system is required to follow the exercise, but keeping a handy copy of the Rules is advisable.
Before ending this short introduction, two side notes about language are in order: first, this work has not been intended for, nor written by, native speakers
 hence, the language is far from perfect, but we hope that it will be easy enough to understand, and that any possible native speakers will forgive its many flaws. Second, and possibly more important, is that we definitely do not want to address a player as either man or woman. Luckily, English language offers a very good device to this end in the use of neutral pronouns  therefore, our readers are advised that our player will always be “it”.
Warm and heartfelt thanks go to IA Roberto Ricca for his valuable and patient work of technical review and the many useful suggestions.
Happy reading!
Notice: to help the reader, the text contains many references to relevant regulations. These references are printed in italics in square brackets “[ ]”  e.g., [C.04.2:B.1] refers to the FIDE Handbook, Book C: “General Rules and Recommendation for Tournaments”, Regulations 04:
40 The FIDE (Dutch) Swiss pairing system, so named with reference to its promoter and developer, Dutch IA Geurt Gijssen, was adopted by FIDE in 1992. Its rules are codified in the FIDE Handbook, available on www.fide.com.
41 For details about the changes, see the minutes of Abu Dhabi 2015 Congress in the Swiss Systems of Pairings and Programs Commission webpage, http://pairings.fide.com.
“FIDE Swiss Rules”, Section 2: “General Handling Rules”, item (B), paragraph (1). Since a great deal of our references will be made to section C.04.3: “FIDE (Dutch) System”, these will simply point to the concerned article or subsection  e.g., [A.7.b] indicates item (b) of Article
(7) of section (A) of those Rules. All regulations can be downloaded from the website of FIDE (www.fide.com).
The preliminary stage of a tournament consists essentially in the preparation of the list of participants. To this end, we sort all players in descending order of score42, FIDE rating and FIDE title43 [C.04.2:B]. Homologous players (i.e. those players who have identical scores, ratings and titles) will normally be sorted alphabetically, unless the regulations of the tournament or event explicitly provide a different sort rule.
Here we face our first problem: the FIDE (Dutch) system belongs to the group of rating controlled Swiss systems44, which means that the resulting pairings depend very closely on the rating of the players  therefore, to get a proper pairing for the round, the players’ ratings need to be the correct ones – i.e. they must correctly represent each player’s strength. Because of this, the Rules require us to carefully verify all of the ratings and, when a player does not have one, to make an estimation as accurate as possible [C.04.2:B.2]. When a player has a national rating, but no FIDE rating, we can convert the first to an equivalent value  in some cases directly, in others by using appropriate formulas. For instance, when a player has no rating at all, we shall usually need to estimate its strength according to current practices or national regulations.
42 Of course, at the beginning of the tournament all players have a null score, unless an accelerated pairing is used.
43 The descending order for FIDE titles is GM, IM, WGM, FM, WIM, CM, WFM and WCM  followed by all untitled players [C.04.2:B.3.b].
44 The “Rating Controlled Swiss Systems” belong to a more general class of “Controlled (or Seeded) Swiss Systems”, in which the initial ranking list is not random or assigned by lots, but sorted according to given rules.
After we prepared the list as indicated above, we can assign to each player its
Pairing Number 
Player 
Title 
Rating 
1 
Alice 
GM 
2500 
2 
Bruno 
IM 
2500 
3 
Carla 
WGM 
2400 
4 
David 
FM 
2400 
5 
Eloise 
WIM 
2350 
6 
Finn 
FM 
2300 
7 
Giorgia 
FM 
2250 
8 
Kevin 
FM 
2250 
9 
Louise 
WIM 
2150 
10 
Marco 
CM 
2150 
11 
Nancy 
WFM 
2100 
12 
Oskar 
 
2100 
13 
Patricia 
 
2050 
14 
Robert 
 
2000 
Our tournament is comprised of 14 players. The players’ list, already properly sorted according to [C.04.2:B], is on the right.
Because of a perhaps a bit controversial (but none the less almost universal) language convention, players who are first on this list (“higher ranked” players) are said to have the highest pairing numbers  in short, number 1 is higher than 14... This is somewhat odd – but, in time, it will become a habit.
The number of rounds is established by the tournament regulations, and cannot be changed after the tournament has started. We may want to notice that this number is, or should be, in close relation with the number of players, because a Swiss tournament can reasonably identify the winner only if the number N of players is less than or at most equal to 2 raised to the number T of rounds: N ≤ 2T. As a rule of thumb, each additional round enables us to correctly determine one more position in the final standings. For example, with 7 rounds we can determine the strongest player (and, therefore, the player who deserves to win) among at most 128 players while we will be able to correctly select the second best among only 64 players, and the third best only if the players are at most 3246. Thus, it is generally advisable to carry out one or two rounds more than the theoretical minimum: e.g., for a tournament with 50 players, 8 rounds are adequate, 7 are acceptable  while, strictly speaking, a 6 rounds tournament
45 Sometimes a player may be registered with a wrong rating, which needs to be corrected (this is necessary for rating purposes too). In such cases, the pairing numbers may be reassigned, but only for the first three rounds; from the fourth round on, pairing numbers shall not be changed, even if players’ data have to be adjusted [C.04.2:B.4].
46 This is true only if in every game the highest rated player ends up as winner. In practice, the occurrence of different results, such ad draws, forfeits and so on, may change the situation, making the individuation of a definite winner (the so called “convergence”) either slower or faster, according to specific circumstances.
(which are the “bare minimum” with respect to the number of players) would not be advisable47.
The preliminary stage ends with the possible preparation of “pairing cards”, a very useful aid for the management of a manual pairing. They are sort of a personal card, the heading of which contains player’s personal data (name, date and place of birth, ID, title, rating and possibly additional useful data) and of course the pairing number of the player. The body of the card is comprised of a set of rows, one for each round to be played, in which all pairing data are recorded (opponent, colour,
float status48, game result or scored points, progressive points). The card may
be made in any of several ways, as long as it is easy to read and to use. Here, we see a typical example.
The basic advantage of pairing cards is that we can arrange them on the desk, sorting them by rank, rearranging and pairing them in an easy and fast fashion. Nowadays, anyway, actual use of pairing cards has become pretty rare because an arbiter is very seldom required to manually make a pairing from scratch  but it’s not unusual that an unhappy player asks for detailed explanations, so that the arbiter has to justify an already made pairing (usually produced by computer software). With a little practice, we can work out such an explanation right from the tournament board  which, in this case, needs to contain all of the necessary data, just like a pairing card. In this paper, we too will follow this latter method.
47 Of course, this is just a theoretical point of view. In practice, many tournaments are comprised of 5 rounds, because sometimes this is the best we can put together in a weekend. Thus, the determination of the players who end up in the winning positions of the final standings must be entrusted to tiebreaks, which should therefore be chosen with the utmost care.
48 See “Scoregroups and brackets”, page 40.
Now we will draw by lot the Initialcolour49 [see section E]. The colours to assign for the first round to all players will be determined by this [A.6.d, E.5]. After that, we will be ready at last to begin the pairing of the first round. Let us say that a little child, not involved in the tournament, drew White as Initial colour.
S1 = [ 1, 2, 3, 4, 5, 6, 7]
S2 = [ 8, 9, 10, 11, 12, 13, 14] 51
Now, we pair the first player from S1 with the first one from S2, the second one from S1 with the second one from S2 and so on, thus getting the (unordered) pairs {18, 29, 310, 411, 512, 613, 714}. Since this is the first round, unless there is some very special reason to do differently52, there is nothing to stop these pairings – so, to complete the pairing process, now we just need to assign to each player its appropriate colour.
Since no player has a colour preference yet, all colour allocation shall be regulated per [E.5]. Hence, in each pair, the higher ranked player (who comes from S1) gets the initialcolour if its pairing number is odd, while it gets the opposite colour if the pairing number is even. Thus, players 1, 3, 5 and 7 shall receive the initialcolour, for which we drawn white, while players 2, 4, 6 shall receive the opposite, which is black.
49 Some arbiters, misinterpreting the drawing of lots, assign colour at own discretion. It should be emphasized that the Rules explicitly require the drawing of lots (which, by the way, may be at the centre of a nice opening ceremony).
50 When the number of players is odd, S2 will contain one player more than S1.
51 Since names are inessential, from now on we will indicate players only by their own pairing numbers.
52 For example, in certain events, we might have specific rules, or reasons, to avoid players or teams from the same federation or club meet in the first round(s), or at all. Such cases usually occur only in major international tournaments, championships, Olympiads and so on, while in “normal” tournaments, in practice, nothing of the kind happens.
The opponents to each player from S1 shall receive, out of necessity, the opposite colour with respect to their opponents; therefore, the complete pairing will be:
1 : 1  8
2 : 9  2
3 : 3  10
3 : 3  10
4 : 11  4
5 : 5  12
6 : 13  6
7 : 7  14
Before publishing the pairing, we have to put it in order [C.04.2:D.9] with the following criteria: 1) the score of the higher ranked player in the pair, 2) the sum of scores of both players, 3) the rank according to the initial order [C.04.2:B] of the higher ranked player. In the vast majority of cases, the FIDE (Dutch) system already generates pairings in the right order (but we always want to check).
At last, we are almost ready to publish the pairing. Before that, we want to check it once again and with extreme care, because a published pairing should not be modified [C.04.2:D.10]53, except when two players should play with each other again.
In the event of an error (wrong result, game played with wrong colours, wrong ratings…), the correction will affect only the pairings yet to be done54 and only if the error is reported by the end of the next round, after which it will be taken into account only for the purposes of rating calculation [C.04.2:D.8]. This means that, in such cases, the final standings will include the wrong result, just as if it were correct!
The last thing to do (and the Pairings Controller may do it while everyone is playing) is the compilation of the tournament board, on which we will post pairings and results for each player. When we renounce the use of pairing cards, as we do here, the board should also contain any other relevant information needed to compose the pairings for following rounds.
For each game, we should indicate at least opponent, assigned colour, and result
– the choice of symbols is free, as long as it is clear, unambiguous, and uniform.
53 But, in this regard, see also FIDE Handbook C.05: “FIDE Competition Rules”, item 7.4.
54 Please note that this rule explicitly forbids the making of new pairings – which is a somewhat frequent request of players  in the event of an error.
Here we will show each pairing by means of a group of symbols comprised of the opponent&rs