375 of 376 games finished, 76% of them decisive. 9 teams won 4–0. United States of America had the best team performance, 2671.
games finished of 376
decisive; White scored 56%
moves per game
moves in the longest game
White wins, 45%
draws, 24%
Black wins, 32%
46 checkmates on the board, 19 draws by repetition, 0 stalemates, 4 draws before move 30 and 1 forfeits.
Left out of the statistics: Dotse Sumya Ibrahim – Ella Miceli-Farrugia — the broadcast did not record the moves that were played. The result 1–0 stands (verify, 28 Sep: board 59.4 Dotse – Miceli-Farrugia scored 1-0 (chess-results agrees) but the relay's moves end 32. Qc4 with Black +6.24, a move after Black's 31... Qxd2 had left White a mate in one; the relay mis-registered moves).
Left out of the statistics: Samiro Mohamed Ali – Vernesa James — the broadcast did not record the moves that were played. The result 1–0 stands (verify, 28 Sep: board 82.2 Samiro – James scored 1-0 (chess-results agrees) but the relay's moves end 39... Nxh3+ with Black +6.84 and 28:46 on her clock after the time-control bonus; the recording stopped before the end).
Left out of the statistics: Tiania Kukula – Ginger Jubitana — the broadcast did not record the moves that were played. The result ½–½ stands (by hand, 28 Sep (Sam): board 84.4 Kukula – Jubitana drawn (chess-results agrees) but the relay's moves end 38... Qxf6 with Black +9.09, a queen for a rook up with 27:09 on her clock; the game ended in a way the relay did not record. A draw has no loser, so verify does not name it).
Left out of the statistics: WFM Kler Caku – Salma Lagmou — the broadcast did not record the moves that were played. The result 1–0 stands (by hand, 28 Sep (Sam): board 45.2 Caku – Lagmou scored 1-0 (chess-results agrees) but the relay's moves end 11. hxg3 in a level Caro-Kann Exchange, with no clock time on any move and no Lichess analysis; the recording stopped before the end. It stood as the section's quickest win of the event).
average accuracy, as Lichess scores it
blunders per game
mistakes, as Lichess marks them
blunders, as Lichess marks them
average time per move
of moves played with under 5 minutes left
Lichess analysed 285 of 371 finished games, with 1,603 inaccuracies, 640 mistakes and 914 blunders. Times come from the relay clocks of 357 games. The extra 30 minutes after move 40 showed up on move 41 or 42 for 67% of 287 players, and never for 74 players who went past move 45.
30captures
Most captures in one game of at least 10 moves.
7captures
Fewest captures in one game of at least 30 moves.
101moves
The longest game of the round.
15moves
Shortest decisive game of at least 10 moves; forfeits do not count.
10moves
The earliest checkmate actually delivered, not a resignation.
78moves
The longest endgame of the round, by Lichess's phase divider.
28moves
Two pieces attacked each other for the longest time and nobody took.
8attackers
The piece with the most attackers and defenders pointed at it at one moment.
62moves
The latest real queen trade: both queens captured within two plies of each other.
5moves
The earliest real queen trade in a game of at least 20 moves.
26moves
The shortest draw; before move 30 only repetition or stalemate is allowed.
473rating points
The biggest rating gap where the lower-rated player did not lose.
20moves
The longest stretch played on with nothing but a king.
Not awarded this round: Houdini.
9captures in a row
The longest run of consecutive captures, both sides taking turns.
12checks in a row
Most consecutive checks by one side; a quiet move ends the hunt.
11moves
The earliest rook to leave its back rank for the third rank or beyond, without capturing.
32moves
The latest castling of the round.
36king moves
Most king moves by one side in a game of at least 30 moves.
68squares
The original queen that travelled the most squares; promoted queens do not count. The metres are at human scale, a long stride per square.
33.3% retreats
Highest share of moves that went backwards, over at least 25 own moves.
34knight moves
Most knight moves by one side.
11captures
Most captures landing on dark squares by one side.
12captures
Most captures landing on light squares by one side.
7pawn captures
Most captures made by pawns by one side.
8arrivals
The one piece that landed on d4, d5, e4 or e5 most often.
28squares
The one piece that stood on the most different squares, its start square included.
69.7% of piece moves
Highest share of one piece type among a side's piece moves, over at least 20 of them.
9pawn moves
Most pawn moves by one side while the game was still in its opening.
47.6% edge moves
Highest share of moves from or to the a- and h-files, over at least 25 own moves.
9.7% expected
The win with the lowest expected score by Elo, at most 25 %.
2249opponent rating
An untitled player beat or drew a titled one; the strongest opponent wins.
388rating points
The decisive game with the biggest rating gap won by the lower-rated player.
2530average rating
The game with the highest average rating.
10years
The youngest player who played this round, by birth year.
71years
The oldest player who played this round, by birth year.
56years
The largest age gap between two opponents.
Not awarded this round: GM Scalp.
39:32on one move
Most time spent on a single move, from the relay clocks.
63moves under 5 minutes
Most moves played with less than five minutes on the clock.
0:31left at the lowest
Won after dropping below one minute at some point after move 30; the lowest clock wins.
1:38:38left at the end
Most time left at the end of a won game of at least 30 moves.
6:15gained after move 40
Most time banked after move 40 from the 30-second increments, in a game of 50 moves or more.
39:32thinking, then a blunder
Longest think that ended in a blunder, as Lichess marks them.
Not awarded this round: Flag Fall.
From the relay clocks of 357 of 371 finished games; games with missing, hand-entered or corrected clocks are left out, and time left after move 40 counts only once the clock shows the extra 30 minutes.
99.3% accuracy
Highest Lichess accuracy in a game of at least 20 moves.
7combined centipawn loss
Lowest combined average centipawn loss of both players, at least 20 moves.
306combined centipawn loss
Highest combined average centipawn loss of both players, at least 20 moves.
93.7% win chance lost
The Lichess-marked blunder that cost the most winning chances.
#−1at the worst
Won a game that the engine had at −3.00 or worse for the winner; the deepest hole wins.
#1at the best
Lost or drew a game the engine had at +3.00 or better for them; the biggest lead wins.
6swings
Most swings between a one-pawn lead for White and one for Black, in a game of at least 20 moves.
1move mate missed
The opponent let a forced mate slip, as Lichess marked it, and did not win; the shortest missed mate wins.
56moves without a mark
Longest game of at least 30 moves in which one side got no inaccuracy, mistake or blunder mark from Lichess.
From Lichess’s computer analysis of 285 of 371 finished games. 86 more join when the round is updated the next evening. Inaccuracies, mistakes and blunders are the ones Lichess marks on the move.
7moves
The king stepped off its home square by move 15 while it could still castle, was not in check and both queens were on. The earliest walk wins.
15moves
The queen went right after a move Lichess marked ??: taken on the next move, not won back, and still three points down soon after. The earliest one wins.
53% scored by all who tried
Ben Finegold says never push the f-pawn one square. These players did it by move 10 (f3 or …f6) and won anyway.
2points of material given
A real sacrifice: the piece was taken at once and the player stayed at least two points down, yet Lichess found nothing wrong with the move. The biggest gift wins.
18moves down
“Take your opponent into a deep dark forest”: won after the longest stretch at least three points down in material.
65moves
Won a long game from nothing: the eval stayed within ±1.00 through move 40, and the win came after move 60. The longest grind wins.
5moves
Both rooks on the opponent’s second rank, eating everything; the longest stay wins.
6rank reached
The king marched to its fifth rank or beyond before the endgame, with the enemy queen still on (Short–Timman, Tilburg 1991). The deepest walk wins.
Homages to chess memes, famous games and Samarkand lore, meant kindly. Brilliant!!, Botez Gambit and The Carlsen Grind use Lichess’s analysis of 285 of 371 finished games; the rest look at every finished game.
22en passants
Every en passant capture of the round.
3named mates
Checkmates with a name: back-rank, smothered, Anastasia’s, Arabian, hook, dovetail, Boden’s, double-bishop, double check, pawn, promotion, king move or castling.
7forks
A knight gave check and attacked the queen at the same time, from a square where it could not simply be taken. The ones that won the queen come first.
18left alone
En passant was on the board and the player left it alone. En passant is forced, allegedly.
5pawn grabs
The queen grabbed the b-pawn (Qxb7 or …Qxb2) by move 15. Fischer’s favourite snack; the result tells whether it was poisoned.
11sacrifices
The classic bishop sacrifice on h7 (or …h2): the bishop took the pawn with check.
1diamond
Four pawns around an empty square: one below, two beside, one above. The first time for each player is listed.
7rows of pawns
Four or more pawns side by side on one rank beyond their starting rank. The first row of each player is listed, the longest first.
4Irish centres
Three pawns on one file, one behind the other. The first time for each player is listed.
5photo finishes
Both sides promoted a pawn within four moves of each other.
3third queens
A third queen on the board.
Not awarded this round: Underpromotion, Shahrukh, Il Vaticano, Pawn Cube, The Hard One.
The awards that list every case instead of crowning one, read from the position on every move; the overview counts them across the rounds. The Hard One, Il Vaticano, the pawn shapes and Royal Court may well stay empty for a round or two.
9matches
Won the match 4–0, forfeits included.
12matches
board 1 lost
board 1 lost
board 1 lost
board 1 lost
board 1 lost
Lost on board 1 and still won the match.
2671performance
opponents averaged 2271
Highest performance over the four boards: average opponent rating plus 400 × (wins − losses) ÷ 4, all four played against rated opponents.
252points lower on average
Won a match against the team that looked strongest on paper: the largest gap in average rating, all four boards played and rated.
285moves
34 + 87 + 63 + 101
The match whose four boards ran longest, in moves added together.
83moves
18 + 30 + 20 + 15
The match whose four boards were over quickest, in moves added together; forfeits never count.
104captures
The match with the most captures over its four boards.
28captures
The match with the fewest captures over its four boards.
55opening moves
The match that stayed in the book longest: the opening phases of its four boards added, in moves.
157½endgame moves
The match with the most endgame moves over its four boards.
9eval swings
The match with the most swings between a one-pawn lead for White and one for Black over its four boards; every board needs Lichess's full analysis.
Not awarded this round: Treaty of Samarkand.
Rates over each team’s finished games of the round, four needed. The event-wide team awards, the category leaders and the board race are on the standings page.
45.97per 100 moves
97 of 211 moves crossed the middle
Most moves into the opponent's half per 100 of the team's own moves.
7.75checks per game
31 checks in 4 games
Most checks given per game over the team's finished games.
5.75per game
23 late knight moves in 4 games
23 late knight moves in 4 games
Most knight moves from move 31 on, per game, over the team's finished games.
5promotions
Most promotions by the team's players.
23.94retreats per 100 moves
51 of 213 moves went backwards
Most retreating moves per 100 of the team's own moves; castling never counts.
6.49retreats per 100 moves
10 of 154 moves went backwards
Fewest retreating moves per 100 of the team's own moves.
38.96pawn moves per 100 moves
60 of 154 moves
Most pawn moves per 100 of the team's own moves.
| Knights | 26.43 % | Armenia ARM37 of 140 moves |
| Bishops | 25 % | Slovenia SLO34 of 136 moves |
| Rooks | 36.82 % | Belarus FID81 of 220 moves |
| Queens | 30.05 % | Bosnia & Herzegovina BIH64 of 213 moves |
For each piece, the team whose own moves have the highest share of it.
18.75king moves per game
75 king moves in 4 games
Most king moves per game over the team's finished games.
5.5average castling move
castled in 4 of 4 games
Earliest castling on average, over the games in which the team castled at all.
3checkmates
3 of 3 wins by checkmate on the board
Most checkmates delivered on the board.
2½identical moves
with Black: 1. e4 e6 2. d3 d5 3. Nd2
with Black: 1. d4 Nf6 2. c4 e6 3. Nf3
with White: 1. e4 e6 2. d3 d5 3. Nd2
with White: 1. e4 d5 2. exd5 Qxd5 3. Nc3
with Black: 1. e4 d5 2. exd5 Qxd5 3. Nc3
with White: 1. d4 Nf6 2. c4 e6 3. Nf3
Two of the team's games with the same colour that followed the same moves for longest; both are listed.
3:06per move
134 moves timed in 4 games
Most seconds per move over the team's games with usable clocks.
1:05per move
122 moves timed in 4 games
Fewest seconds per move over the team's games with usable clocks.
44.8% under five minutes
95 of 212 moves in 4 games
Highest share of moves played with under five minutes on the clock.
1:13:33left on average
over 4 games
Most time left on the clock after the last move, on average.
Over the relay clocks of 357 of 371 finished games; a team needs four such games.
94.9% accuracy
4 analysed games
Highest average Lichess accuracy over the team's analysed games.
3games saved
Most games won or drawn from a position the engine had at −3.00 or worse.
Over Lichess’s analysis of 285 of 371 finished games; a team needs four analysed games.
moves of opening
moves of middlegame
moves of endgame
Averages per game, with the phases split as Lichess splits them. 81% of the games reached an endgame; the first capture came on move 7 on average.
| Endgame | Reached | At the end |
|---|---|---|
| Rook | 67 | 43 |
| King and pawn | 39 | 23 |
| Bishop | 33 | 19 |
| Queen and rook | 25 | 9 |
| Queen | 20 | 19 |
| Knight | 14 | 8 |
A rook endgame has nothing but kings, pawns and rooks left; the others likewise, as the Lichess puzzle tagger defines them. A game can pass through several types, so the first column adds up to more than the games.
Each band runs from the games that reached the endgame to those that still had it on the last move; the rest went on into another type. One scale for both columns.
captures, 18.5 per game
checks
castled kingside
castled queenside
games without castling
promotions, 0 under
en passant captures
Share of all finished games. Opening names and ECO codes come from Lichess, based on the position reached.
The round’s openings as a tree: the five volumes of the opening encyclopaedia (A flank openings, B semi-open games, C open games and the French, D closed games and the Grünfeld, E Indian defences), their families, and every variation played. A line grows with the games through it; the red thread leads to the round’s rarest openings.
Point at a line or a name to name the opening; click to open it on Lichess.Tap a name to name the opening; tap it again to open it on Lichess.
A bar per family: White’s wins, the draws, Black’s wins. The tick is the family’s score for White, the grey line the round’s; a tick right of it, and White did better than usual. Framed in red, the best for White and the best for Black of the families with at least 5 games.
| Family | Games | White wins | Draws | Black wins | White’s score |
|---|---|---|---|---|---|
| Sicilian Defense | 69 | 27 | 19 | 23 | 53% |
| Queen's Pawn Game | 35 | 20 | 5 | 10 | 64% |
| Caro-Kann Defense | 32 | 12 | 10 | 10 | 53% |
| French Defense | 27 | 11 | 7 | 9 | 54% |
| Queen's Gambit Declined | 23 | 8 | 9 | 6 | 54% |
| Italian Game | 19 | 8 | 2 | 9 | 47% |
| Ruy Lopez | 14 | 5 | 6 | 3 | 57% |
| Indian Defense | 11 | 5 | 1 | 5 | 50% |
| English Opening | 10 | 5 | 1 | 4 | 55% |
| King's Indian Defense | 10 | 1 | 5 | 4 | 35% |
| Scandinavian Defense | 9 | 7 | 0 | 2 | 78% |
| Slav Defense | 9 | 5 | 1 | 3 | 61% |
| Catalan Opening | 7 | 2 | 3 | 2 | 50% |
| Zukertort Opening | 7 | 2 | 4 | 1 | 57% |
| Bishop's Opening | 6 | 4 | 0 | 2 | 67% |
| Nimzo-Indian Defense | 6 | 3 | 2 | 1 | 67% |
| Philidor Defense | 6 | 2 | 2 | 2 | 50% |
| Pirc Defense | 5 | 4 | 1 | 0 | 90% |
| Four Knights Game | 4 | 2 | 0 | 2 | 50% |
| Queen's Gambit Accepted | 4 | 3 | 1 | 0 | 88% |
| Grünfeld Defense | 3 | 0 | 3 | 0 | 50% |
| King's Knight Opening | 3 | 1 | 0 | 2 | 33% |
| Modern Defense | 3 | 1 | 1 | 1 | 50% |
| Queen's Indian Defense | 3 | 2 | 0 | 1 | 67% |
| Rapport-Jobava System | 3 | 1 | 0 | 2 | 33% |
| Opening | Games | White’s score |
|---|---|---|
| B00 Pirc Defense | 5 | 90% |
| D53 Queen's Gambit Declined | 5 | 70% |
| A45 Indian Defense: Accelerated London System | 5 | 40% |
| D00 Queen's Pawn Game: Accelerated London System | 9 | 39% |
| Opening | Games | White’s score |
|---|---|---|
| B13 Caro-Kann Defense: Exchange Variation | 9 | 33% |
| D00 Queen's Pawn Game: Accelerated London System | 9 | 39% |
| A45 Indian Defense: Accelerated London System | 5 | 40% |
| D53 Queen's Gambit Declined | 5 | 70% |
Openings with at least 5 games.
pawn moves, 25%
knight moves, 17%
bishop moves, 15%
rook moves, 18%
queen moves, 13%
king moves, 13%
| Piece | Captured | Still on the board at the end |
|---|---|---|
| pawns | 3,193 | 45% |
| knights | 1,230 | 17% |
| bishops | 1,129 | 24% |
| rooks | 860 | 42% |
| queens | 468 | 44% |
23 by White, 23 by Black
The mating piece is read from the final position. A piece opens onto its mates.
A dot per checkmate on the move it fell, filled for a mate by White, hollow for one by Black, the fastest in red. Each piece in the list opens onto its games.
Darker means more moves, on a square-root scale.
Quietest square with any traffic: a1, 138.
Where the pieces went in the finished games of the round.