Better at Chess

Blog · Data, September 2026 ·

Lost on time, or played too fast?

Played too fast, about twice as often. In analysed 10-minute games on Lichess from September 2026, a mistake made in 3 seconds or less, with at least two and a half minutes still on the clock, decided about 1 lost game in 11 for players rated about 1200 on Chess.com’s rapid scale. The clock decided half as many: 1 in 30 ended on time and 1 in 84 was decided in time trouble. In most lost games the loser still had more than half their time when the deciding move was played.

The numbers in brief

Source: the Lichess open database, September 2026, analysed 10-minute games, ratings converted to Chess.com’s rapid scale. How we measured, below, has the details.

How are lost 10-minute games decided?

Lost 10-minute games decided on the clock or by a too-fast move, by chess rating. Analysed games, September 2026, Chess.com rapid scale.

We took every lost game and found the move that decided it: the loser’s move from which the winner’s chances stayed at 70% or more to the end. At 1200, 3.3% of lost games ended on time while the loser still had a playable position, and in another 1.2% the deciding move was played with 30 seconds or less left. In 9.0%, the deciding move was a mistake made in 3 seconds or less while the player still had two and a half minutes or more. That’s about twice as many games as the clock decided at 1000 and 1200, and 1.6 to 1.8 times as many at 1400 and 1600.

The rest were decided another way. In 84% of lost games at 1200, the deciding move was neither played in time trouble nor a mistake made that fast with time to spare. The remaining 2% had no single deciding move. The numbers barely change with rating: from 1000 to 1600, the too-fast share stays between 8.8% and 9.5%.

How much time is left when the game is lost?

The loser’s clock at the move that decided a lost 10-minute game: the median and the middle half of games, by chess rating. Analysed games, September 2026, Chess.com rapid scale.

Most of it. At 1200 the median loser still had 7.8 of their 10 minutes when they played the move that decided the game, and 83% still had at least half their time. At 1600 it was 7.5 minutes and 77%. The deciding move itself took 8 seconds in the median game at 1000 and 1200, and 9 at 1600.

Are fast moves worse moves?

Not on average. From move 11, with the game still close, moves made in 3 seconds or less with at least two and a half minutes left were mistakes 19% of the time at 1200, against 27% for all the other moves. One likely reason is that easy moves get played fast and hard positions take longer, but the data can’t tell. So this doesn’t show that fast moves are safe, and it doesn’t show that slowing down would have saved a game.

What the games do show is narrower: of the mistakes players made while the game was still close, about 1 in 8 was made in 3 seconds or less with time to spare, and those decided about 1 lost game in 11. Players at 1200 made one such mistake every 2.5 games, on average.

Does the clock matter at all?

It goes with more blunders. From a level position, where both players still had chances, rapid players at 1200 (all rapid time controls together) blundered 12% of the time while more than half their clock was left, and 17% to 18% once it wasn’t. Later positions can be harder too, so this doesn’t show how much the clock itself adds. And losses on the clock grow with rating: lost on time or in time trouble, 4.4% of lost games at 1000 and 5.8% at 1600, while the loser’s clock at the deciding move falls from 7.9 to 7.5 minutes.

Blitz is different. At about 1200 on Chess.com’s blitz scale, 7.6% of lost blitz games ended on time and 4.1% were decided in time trouble, against 3.5% decided by a too-fast move. The rule scales with the time control: in 3- and 5-minute games it means 1 to 1.5 seconds with 45 to 75 seconds left, and these blitz numbers pool every blitz time control. More games end on the clock than these numbers suggest: 11% of all rated rapid games, draws included, against 3% to 4% of lost 10-minute games here. These are different populations, so the two numbers aren’t directly comparable. And a game only counts as lost on time here if the player who ran out still had a playable position; one who ran out in a lost position is counted by the move that lost it.

What Better at Chess calls these

Better at Chess looks for these habits in your own games. Itchy Fingers is a mistake made within seconds while you still had time; Auto-Reply is the same right after your opponent’s check or capture; Time Scramble is a mistake made with little time left. It counts how often each one happened in your recent games and shows you the positions. See the habits it looks for. How blunders change with rating is in our first data post, At what Elo do you stop blundering?

Get Better at Chess on the App Store. For iPhone.

All the numbers

Every 100-point Lichess rapid band from about 850 up, shown on Chess.com’s rapid scale, in 10-minute games. The levels named above are these rows: 1000 is 950–1100, 1200 is 1100–1250, 1400 is 1350–1500, 1600 is 1500–1650. Analysed games, September 2026. Download all the numbers as a CSV file, free to reuse with a link to this page (CC BY 4.0).

How lost 10-minute games were decided on the clock, the loser’s clock at the deciding move and the share of mistakes made in 3 seconds or less, September 2026, games with computer analysis, by rating on Chess.com’s rapid scale converted from Lichess
RatingLost on timeTime troubleToo fastDecided another wayClock leftFast mistakes
850–9503.2%1.1%8.6%84.9%7.9 min12%
950–11003.3%1.1%8.8%84.4%7.9 min13%
1100–12503.3%1.2%9.0%84.3%7.8 min13%
1250–13503.4%1.4%9.5%83.5%7.7 min14%
1350–15003.6%1.6%9.5%83.1%7.6 min14%
1500–16503.8%1.9%9.4%82.6%7.5 min14%
1650–18003.9%2.3%9.3%82.4%7.3 min14%
1800–19004.3%2.8%9.0%81.7%7.1 min13%
1900–20504.8%3.6%8.4%81.0%6.8 min13%
2050–21504.8%4.4%8.4%79.9%6.4 min13%
2150–22504.8%5.5%7.7%79.0%5.9 min12%
2250+4.3%6.8%7.9%76.7%5.4 min12%

Lost on time, time trouble, too fast and decided another way are shares of lost games; the rest had no single deciding move. Decided another way also holds the few games whose clock wasn’t recorded (26 of 183,093 at 1200). Clock left is the loser’s median clock at the deciding move. Fast mistakes is the share of mistakes made in 3 seconds or less with time to spare, among all mistakes made while the game was still close.

Questions

How many chess games are lost on time?

In analysed 10-minute games on Lichess from September 2026, about 1 lost game in 30 ended on time while the loser still had a playable position, for players rated about 1000 to 1200 on Chess.com’s rapid scale, and 1 in 26 at 1600. Counting every rated rapid game that ended on the clock, draws and hopeless positions included, it’s 11%.

Do chess players lose because they move too fast?

More often than they lose on the clock. A mistake made in 3 seconds or less, while the player still had at least two and a half of their 10 minutes, decided about 1 lost game in 11 from 1000 to 1600 on Chess.com’s rapid scale: 1.6 to 2 times as many as the clock decided. That’s what happened in these games, not proof that slower play would have saved them.

How much time is left when a 10-minute game is lost?

Usually most of it. In the median lost game, the loser still had 7.8 of their 10 minutes when they played the move that decided it at 1200 on Chess.com’s rapid scale, and 7.5 at 1600. About 4 losers in 5 still had half their time or more.

Are fast moves more often mistakes?

Not in these games. From move 11, with the game still close, moves made in 3 seconds or less with at least two and a half minutes left were mistakes 19% of the time at 1200, against 27% for all the other moves. That doesn’t show that speed is safe, only that fast moves weren’t the worse ones on average.

How we measured

Games
Every rated standard game played on Lichess in September 2026, from the Lichess open database (public domain): 88,673,670 games, after leaving out games with a bot, games that were abandoned, unterminated or ended by a rules infraction, correspondence games and games without both players’ ratings.
Games with computer analysis
Every number here comes from games that carry Lichess’s computer evaluation, 8,868,664 in all: games someone asked Lichess to analyse. They aren’t a random sample: upsets are far more common among them than among all games. The 11% of rapid games that end on the clock is the exception: it counts every rated rapid game, analysed or not.
Who counts where
How lost games were decided: the 1,763,633 analysed 10-minute games (10 minutes, no increment) that ended in a win, each at the band of the player who lost it. The clock at the deciding move: those of them decided by a move whose time is known. Mistakes made in 3 seconds or less per game: every player in an analysed 10-minute game with at least 20 moves by each side, wins, draws and losses alike (319,190 player-games at 1200). Fast moves against the others: every move from move 11 on in an analysed 10-minute game of any length, in positions within three pawns of level, not on a low clock and not recaptures (724,056 quick moves and 2,236,688 others at 1200). Blunders by clock from a level position: all rapid time controls together. Blitz: all blitz time controls together.
The deciding move
In a game that ended in a win, the winner’s chances stayed at 70% or more from some point to the end; the deciding move is the loser’s move at that point, or just before it. Winning chances use Lichess’s own formula. A loss on time while the loser still had 40% or more counts as lost on time, with no deciding move; a game with no such point is undecided.
Time trouble
The deciding move was played with 30 seconds or less on the clock (10 seconds or 5% of the starting time, whichever is more).
Too fast
The deciding move was a mistake, made in 3 seconds or less while the player still had at least two and a half minutes, from move 11 on, and not an immediate recapture. These are the thresholds Better at Chess uses for a player’s own games, applied here to Lichess’s evaluations and clocks. A few details differ: Lichess records whole seconds, recaptures and checks are read from the move notation, and positions with only one legal move can’t be told apart. Each player’s first and last moves, and games whose clock didn’t start at the time control, are left out.
Mistake
A move that lost a pawn’s worth of evaluation or more from a position within three pawns of level, by Lichess’s evaluation. This is the scale Better at Chess uses; the clock-and-blunder comparison from a level position uses our first post’s blunder instead, a move that lowers the player’s winning chances by 10 points or more.
Clock
Lichess records each player’s clock after every move, in whole seconds. A move’s time is the clock before it, plus the increment, minus the clock after it.
Chess.com rapid scale
We converted each Lichess rapid rating to Chess.com’s rapid scale with a comparison of players who have established ratings on both sites: a Lichess rapid 1576 matches a Chess.com rapid 1200. Single players scatter widely around these matches, so treat each level as approximate. Each level on this page is the 100-point Lichess band its Chess.com rating falls in, with the band’s range rounded to the nearest 50.
Rounding
One decimal for shares of lost games, whole percentages elsewhere, because the conversion is approximate.
Privacy
We kept no usernames, game links or game IDs. Every published number comes from groups of at least 1,000 games, player-games or moves.
Fixed first
The deciding move, time trouble and the too-fast rule were written down on 11 November 2026, before any of these numbers was computed.