What your chess accuracy actually means

Accuracy is a summary of how much you gave away, move by move, in the positions you were actually in. Every move is compared with what the engine would have played, the difference is converted into how much winning chance it cost, and the whole game is averaged into a single percentage.

The thing to hold on to is that it measures losses against your own position. It knows nothing about how hard the position was, how much time you had, whether your opponent was any good, or whether you understood what you were doing. The guide has the arithmetic; this page is about what to do with the answer.

In a quiet, symmetrical game where nothing is ever hanging, almost any sensible move is close to best. Two club players can trade everything off and both finish above 90% without either having made a decision. In a sharp game — an opposite-castling attack, a messy middlegame with three pieces loose — every move has a real cost attached, and 75% can be an excellent performance.

So the number is only comparable between games of similar character. Comparing your accuracy in a 40-move Berlin with your accuracy in a bullet Sicilian tells you about the two openings, not about you.

There is no standard formula. This site uses the Lichess curve — win percentage lost per move, then a weighted average that gives more weight to the moves played while the position was volatile. Chess.com uses something of its own and does not publish it. Both are defensible; neither is correct in a way that makes the other wrong.

What follows is worth remembering: an accuracy figure only means something next to other figures produced the same way. Chasing a number across two sites is chasing two different measurements.

One number for a whole game is a headline. The things that actually change how you play are underneath it:

• The two or three moves that decided it. Most games are settled by a handful of moves. The rest is noise around them, and it is the moves that cost the most that are worth an hour of your attention.
• What keeps happening. One blunder is bad luck; the same blunder in six games is a habit with a name. Reviewing several games at once is the only way to see it.
• When in the game it happens. Accuracy by phase — opening, middlegame, endgame — and by how long you spent on the move says far more than the average does.

Paste a PGN or type a username and every move is graded in your own browser — free, no account, nothing uploaded. Analyse a game, read how the grades are worked out, or find out what makes a move brilliant rather than good.