Understanding Your Chess Accuracy Percentage

Accuracy percentage summarises how closely your moves matched what the engine considered best, across an entire game. A 100% game means every move you played was the engine's top choice; lower percentages reflect moves that gave some of your advantage away.

It is built one move at a time. For each move, the review measures how much you lost compared with the best move, in hundredths of a pawn, and turns that into a score out of 100 that falls away quickly: playing the best move scores 100, giving away half a pawn scores roughly two thirds of that, and a full-pawn loss scores under half. Your game accuracy is the average across your moves.

Not every move counts. Moves inside known opening theory are skipped, since matching the book proves you studied rather than that you played well, and so are moves where only one legal move existed. That is why two games of the same length can average over quite different numbers of moves.

The shape of that curve explains a lot of surprising results. Because the score falls fast, one blunder can pull a whole game down several points, while a long run of small inaccuracies costs less than you might expect. A single oversight in an otherwise clean game and a game full of vague moves can end up with similar percentages for opposite reasons.

It is easy to assume a higher rating always means higher accuracy, but that isn't quite right. A 2000-rated player pushing for a win in a sharp position might post lower accuracy than a 1200-rated player in a quiet, simple game. Sharp positions have more ways to go wrong and fewer safe moves; quiet ones forgive almost anything. Accuracy measures precision in that specific game, not overall skill.

The same logic means accuracy is a poor way to compare yourself with other players. Your opponent's choices shape the positions you get, and a player who steers towards calm positions will look more accurate than one who seeks complications, even at the same strength.

Where it is genuinely useful is as a trend. Track it across many of your own games rather than reading single results, and ideally split it by phase — opening, middlegame, endgame — to see where the points are going. A rising average over weeks or months is a solid, concrete sign that you are making fewer costly mistakes.

And when a game's number disappoints, don't stop at the number. Find the one or two moves that pulled it down. Those moves, not the percentage, are what you can actually learn from.