How to review your own chess games

Q: How should I review my own chess games?

Start from your results rather than your memory. Pick the games your data flags: losses where you were ahead on material, losses decided by a single moment rather than a long decline, and games where a forced mate was available. For each one, find the move where the game turned and ask what should have alerted you at that point. The warning sign transfers to future games; the specific move does not.

Q: Should I use an engine to review my games?

Use it second, not first. An engine tells you which move was best, which is the easy half. The harder and more useful skill is noticing that a position had become critical at all. Form your own view of where the game turned, then check it against the engine and see where the two disagree.

Q: How many games should I review?

Two or three a week, properly, beats twenty skimmed in one sitting. Reviewing is slower than playing and the returns come from depth rather than volume. If you only have time for one, make it a game you were winning and lost.

Almost nobody reviews their games, and most who do review the wrong one. The game you remember is rarely the game that cost you. Rookbook picks the ones worth your time.

Why most review doesn't work

The first is selection. You review the game that stung — the one where you hung a queen, or lost to someone rated 200 below you. That's why it's a bad sample: memory keeps the vivid loss and discards the twenty ordinary ones, so the pattern you think you're fixing is just the one you remember.

The second is method. Most review is: open the engine, click through, note the move it disliked. That tells you which move was best, not what you should have noticed — and noticing is what transfers to the next game.

Rookbook's blown leads table:every game where a material lead was lost, with the peak lead, the final margin, and the clock at the biggest lead. — The first list worth having: every game you were ahead in and lost, with how far ahead you got and what was on your clock.

Pick the games your results flag

Three lists are worth more than everything else in your history:

Losses where you were ahead on material. The fastest points available, because you already outplayed the person — nothing else returns as much.
Losses decided by one moment. If your loss accuracy is close to your win accuracy, the game wasn't a gradual defeat — something specific happened, and it's findable.
Games where mate was on the board — even the ones you won. Missing a forced mate rarely costs the game, but it tells you your attention was on material, not the king.

Notice what isn't on that list: the games you lost badly from move one. Those are openings you don't know or days you shouldn't have played, and neither is fixed by staring at the board.

A game open in Rookbook's review page: the board at the move that decided it, with the full move list and the time spent on each move. — A game opened at the moment it turned, not from move one. The clock column shows where the thinking went — and where it stopped.

Find the moment, not the mistake

A review that covers the whole game teaches nothing — most moves in most games are fine. You want the single point where the position changed character, and three signals find it faster than clicking move by move:

Material swing. Where did the balance move, and was it a fair trade or a piece given away?
The clock. A four-minute think marks the moment you knew something was wrong; a run of three-second moves, the moment you stopped checking.
Accuracy split. One bad move in a clean game shows up as a gap between your accuracy and your opponent's.

A useful test: if you can't say what should have alerted you, you haven't finished reviewing it. Knowing the better move is not the same as knowing how you'd have found it.

Ask one question of each game

What should have alerted me?

Nearly always there's a concrete answer: a piece that stopped being defended, a knight reaching a square nobody watched, an opponent's move you'd decided to ignore before they played it.

The move is specific to that game; you'll never see it again. The signal repeats every week.

Don't start with the engine

Form your own view first — where the game turned, what you'd play differently — then check it. You're looking for where you and the engine disagree — that gap is your blind spot. Reading the evaluation first replaces your judgement with its, which feels productive and teaches nothing.

The exception is tactics you didn't see. If the engine finds a forced win you'd have missed with an hour to think, note the pattern and move on.

How many, how often

Two or three games a week, properly, beats twenty skimmed. A single blown lead worked through carefully changes more than an evening of clicking through evaluation bars.

If you only have time for one, make it a game you were winning and lost.

Rookbook's missed mates list: each position where a forced checkmate was available, the move played instead, and what it cost. — The third pile: positions where mate was on the board. Worth reviewing even in games you won — missing one says your attention was on material, not the king.

Do it on your own games

Rookbook reads your chess.com history and builds those three lists — blown leads, losses decided by a single moment, and every missed forced mate. Each opens on a board at the move that mattered, with the material balance, your clock, and what you played instead.

Free, no account, and your games never leave your browser.

Review your own games on rookbook.net →

Common questions

How should I review my own chess games? Start from your results, not your memory. Take the losses where you were ahead on material, the ones decided by a single moment, and any game where mate was available — then find the move where it turned and ask what should have alerted you.

Should I use an engine to review my games? Second, not first. Form your own view of where the game turned, then check it. The gap between your answer and the engine's is the thing worth studying.

How many games should I review? Two or three a week, properly, beats twenty skimmed. If you only have time for one, make it a game you were winning and lost.