Chess review tools love labels: blunder, mistake, inaccuracy, best. They make a long move list readable at a glance. They also hide a simple fact: the thresholds are arbitrary, and the same move can receive different labels at different depths.
The usual definitions
Most tools measure the drop in win probability after a move. A common scheme is:
| Label | Win-probability drop | Typical meaning |
|---|---|---|
| Inaccuracy | about 5–10% | The move gives away some of the advantage but does not change the result immediately. |
| Mistake | about 10–20% | A clear error that may require accurate defence or loses the initiative. |
| Blunder | about 20%+ | A large swing, often tactical, that can decide the game. |
ChessUrge uses these win-probability bands for full-game reviews. The exact numbers are less important than the pattern across a game.
Why the same move can change label
Win probability is not linear in centipawns. A move that turns +0.20 into -0.10 may be a small drop; a move that turns +1.50 into +0.80 may be a large drop despite a smaller centipawn change. The position and the depth of the search both matter.
Context beats the label
- A "blunder" in a completely lost position may not change the practical result.
- A "mistake" in a must-win game can be more costly than a "blunder" in a drawn position.
- A move the engine dislikes may be the best practical try against a human opponent.
Use the label to find the move, then read the line and decide whether the error was calculation, evaluation or time. The label is the index; the explanation is the lesson.
Full reviews on ChessUrge show the label, the evaluation swing and the engine line for every move.
See the whole game, not just the blunder.
Choose a review depth, see the maximum cost, and get move-by-move accuracy with blunder, mistake and inaccuracy markers.