How Chess Players Lose at Every Rating

We took 8,000 blitz and rapid games from Chess.com, played in August 2026 by 7,134 different players, and graded every move with Stockfish. Then we split the players by rating and asked a simple question: how do games actually get lost at each level?

Key findings

• By move 25, half of all players rated under 1000 have already blundered. At 2400+, only 7.5% have.
• Players rated 600–799 blunder 3.6 times per 100 moves. Players rated 2400+ blunder 0.9 times, four times less often.
• Two thirds of games at 2400+ contain no blunder at all. Under 800, two thirds contain at least one.
• About 40% of games below 800 end in checkmate. Above 2200, it is about 15%, and resignation takes over.
• Under 600, a quarter of completely winning positions are not won. From 1000 up, about 9 in 10 are.

The first blunder comes much earlier at lower ratings

The gap opens immediately. By move 15, 31% of players under 1000 have blundered, compared with 2% of players at 2400+. By move 25, it is 51% against 7.5%. Strong players do blunder, but much later. For players rated 2400+, the median first blunder comes on move 36. For players under 600, it comes on move 15.

Blunders per move fall four times from the bottom to the top

The rate peaks at 600–799, with 3.6 blunders per 100 moves, and falls to 0.9 at 2400+. The middle of the chart is flatter than you might expect. Between 1000 and 2000, the rate stays close to 2 per 100 moves. What changes across that range is when the blunder happens. The median first blunder moves from move 18 at 1000–1199 to move 22 at 1800–1999, so stronger players get through the opening and early middlegame clean and make their mistakes in harder positions.

Clean games become normal

Under 800, roughly two out of three games contain at least one blunder. Around 1000 it is a coin flip. At 2200 and above, two out of three games are blunder-free.

Checkmate is how beginners lose. Resignation is how everyone else does.

Below 800, about 40% of games finish with checkmate on the board. Lower-rated players play on in lost positions, and they walk into mate more often. Under 800, about 9 in every 100 player-games contain a blunder that allows a forced mate. At 2400+, it is under 3. Above 2200, only 13–16% of games end in mate, and most players resign once the position is lost. Losses on time are also common at the very bottom: a quarter of games under 600 end on the clock.

Winning a won game stops being the problem early

We counted every game where a player reached an advantage of +5 or more, roughly an extra rook. Under 600, 24% of those games were not won. By 1000–1199, 93% were. Above that, the number stays around 88–96% without a clear trend. Converting winning positions is a real skill, but by about 1000 most players can do it. The bigger difference between a 1000 and a 2000 is not throwing the game away in the first place.

What this means for your games

If you are under 1000, the numbers point to the first 20 moves. Half of the players in this group have blundered by move 25, so checking every capture and every check before you move will save more games than another opening line. Between 1000 and 2000, the blunder rate barely moves, but stronger players blunder later. Getting through the middlegame without one is what separates the bands.

All the numbers

Download the full dataset (CSV). It is free to use with a link to this page.

Method

• Games. 8,000 blitz (4,038) and rapid (3,962) games from Chess.com's public game archive, played in August 2026. The games were collected by starting from a set of players and following their opponents' archives. There are 7,134 distinct players. Each side of a game counts as one player-game, filed under that player's own Chess.com rating for the time control.
• Engine. Every move was graded with Stockfish 18 at depth 12, using the same grading as the Chess It Up game analyzer. A move counts as a blunder when it throws away a large share of the player's winning chances. Hanging a piece in a position that was already lost often does not count, because there was little left to lose.
• First blunder. The median move number counts only games with a blunder.
• Winning positions. Counted when a player's position was +5 or better after one of their own moves. Abandoned games are excluded.
• Limits. The sample is not perfectly random. Lower bands are mostly rapid games and higher bands are mostly blitz, so the bands are not perfectly comparable. Where we have enough of both, neither time control shows a consistently higher blunder rate. The 1400–1599 band is small (345 player-games), so treat its numbers with more caution. Depth 12 can miss deep tactics. The game records had no clock times, so this study says nothing about time pressure.

Cite this study

Chess It Up (2026). How Chess Players Lose at Every Rating. https://chessitup.com/research/how-chess-players-lose