HOW TO READ A STOCKFISH BLUNDER REPORT

By the ChessDiary team

Running an analysis produces more than just a list of blunders — it's a move-by-move report with a quality label, an estimated cost in centipawns, and sometimes the tactical pattern behind the mistake. Knowing what each field actually measures turns the report from a red-and-orange list into something you can actually study from.

Here's what each part of a ChessDiary analysis report means, and how to read them together instead of in isolation.

1. THE FIVE QUALITY LABELS

• Best — the engine's top choice, or effectively equal to it. Not necessarily a flashy move; often the quiet, correct one.
• Good — a reasonable, sound move that isn't the engine's first choice but doesn't meaningfully worsen the position either.
• Inaccuracy — a small, usually low-cost slip. Rarely game-deciding on its own, but a pattern of repeated inaccuracies in the same phase of the game is worth noticing.
• Mistake — a move that meaningfully worsens your position without necessarily losing material outright. Worth reviewing closely, since a mistake often sets up a blunder a few moves later.
• Blunder — a move that hands over a significant, usually decisive advantage. Every blunder comes with the engine's recommended alternative so you can see exactly what you missed.

2. CENTIPAWN LOSS, IN PLAIN TERMS

A centipawn is one-hundredth of a pawn's worth of advantage — the engine's standard unit for position evaluation. Centipawn loss on a given move is the difference between the position's evaluation before and after that move, from your perspective: how much of your advantage (or how much further behind) that specific move cost you. A small inaccuracy might cost 20–40 centipawns; a real blunder frequently costs 150–300 or more. It's a precise number, but the quality label next to it is usually the faster thing to scan for.

3. MOTIF — WHAT KIND OF MISTAKE IT WAS

When a mistake or blunder fits a recognizable tactical pattern, the report tags it — a fork, a pin, a hanging piece, a back-rank issue, and so on. This is the field worth tracking across multiple games: a single blunder is one bad move, but if the motif field says "hanging piece" on three of your last five losses, that's not bad luck, it's a specific habit worth fixing. (Our common blunder patterns guide walks through what each motif actually looks like on the board.)

4. OPENING, MIDDLEGAME, OR ENDGAME

Each move is also implicitly tied to a phase of the game based on the move number — roughly the first 15 moves count as the opening, up to move 35 as the middlegame, and everything after as the endgame. Reviewing where your mistakes cluster by phase is often more useful than reviewing them move-by-move in isolation.

5. TIME PRESSURE FLAGS

If your game includes clock data, a mistake or blunder made with under 30 seconds left gets flagged as a time-pressure move rather than folded in with the rest. This matters for how you read your own report — a blunder made scrambling on a 10-second clock and one made with twenty minutes to think are different problems with different fixes.

6. WHICH ENGINE PRODUCED THE REPORT

A quick note after analysis tells you which engine actually ran: Stockfish, or Gemini AI if the Stockfish server wasn't reachable at the time. Both produce the same report structure — quality labels, comments, and best-move suggestions — but Stockfish's numbers come from genuine engine search, while Gemini's are a language model's best estimate. See our guide on what happens when the analysis engine is unreachable for the full explanation.