How to Review Chess Games Without an Engine (and When to Use One)

Updated October 2026 · 8 min read · Every position checked with Stockfish

Engine-first reviews feel productive, but you mostly end up reading arrows. You see that a move was a blunder without learning why you played it, so you repeat it next week. Reviewing without an engine forces you to find the gaps in your own thinking, which is the part that actually moves your rating. This guide gives you a method that takes 20–30 minutes per game. It shows you how to annotate your own games and explains exactly when the engine (or an AI coach) should come in.

1. Record your thoughts before you forget them

Do your first pass within an hour of finishing, while you still remember why you played each move. Replay the game on the Lichess analysis board or the Chess.com analysis board with the evaluation bar and engine lines switched off.

At every move where you remember thinking, add a short comment:

• What you were afraid of
• What you were trying to achieve
• Roughly how long you spent
• Anything you missed that you only noticed later

These notes are the raw material. An engine can tell you the best move, but only you know what was going through your head. Mistakes come from thinking errors, and you can only fix a thinking error you've written down.

2. Find the three to five critical moments

You don't need to analyse every move. Go through the game once quickly and mark the positions that mattered:

• Material changed hands (a capture, a sacrifice, a piece lost)
• You spent a long time on one move, or far too little on a sharp one
• The plan changed: castling decisions, pawn breaks, trading queens
• You felt surprised by your opponent's move
• The result swung: you felt winning, then suddenly worse

Most games under 1800 are decided by one or two of these moments. Spending 15 minutes on each is far more useful than spending 30 seconds on all forty moves.

3. Analyse each critical moment yourself

Set up the position and give yourself a time limit of about 10 minutes. Then work through it as if it were a puzzle:

• Checks, captures, threats for both sides. Start with your opponent's forcing moves, because that's where most missed tactics hide.
• List two or three candidate moves. Don't just justify the move you played.
• Calculate each one a few moves deep, then write down the line and your verdict in plain words: "wins a piece", "probably equal", "too risky because of the knight check".
• Decide which you prefer before you look at anything else.

You can move pieces on the board for this. The point is that the ideas come from you, not from a computer. If you can't decide, write "unclear" and the reason. Those are exactly the positions the engine will teach you most about later.

4. Annotate and label each mistake

Now turn your notes into a short annotated game. Keep it simple:

• ? a mistake, ?? a blunder, ?! dubious, ! a good move you're proud of
• One sentence per critical moment explaining what went wrong in your thinking

Then give every mistake a category. Use the same handful of labels every time so you can count them across games:

• Hung material: you didn't check what was undefended
• Missed tactic: your opponent had a forcing move you didn't see
• Wrong plan: the move was safe but pointless
• Opening gap: you left your preparation and didn't know the idea
• Clock: you ran short of time or rushed

After ten games, the labels tell you more than any single review. If "hung material" keeps appearing, read how to stop hanging pieces before you do anything else.

5. Check the opening against a database, not an engine

For the opening, the useful question is "what do strong players usually play here?", not "what does Stockfish rate as best?". Use the opening explorer on the Lichess analysis board, which draws on real games, to find where you left normal theory.

Note the move where you deviated and the main idea you didn't know. One idea is enough. Memorising ten moves of a line you'll rarely face is not a good use of review time.

6. Now turn the engine on, and argue with it

Only after your own analysis should you switch on Stockfish. Compare it with your notes:

• Where you agreed: good. Your calculation in that position was sound.
• Where you disagreed: this is the lesson. Play the engine's move on the board and ask what it does that your move didn't. Look 3–4 moves deep, not 20.
• Moments you didn't flag but the engine did: these are your blind spots. You didn't even notice the position was critical.

If an engine line makes no sense to you, don't copy it into your notes. Write the idea in words, or leave it out. Lines you don't understand won't help you in your next game. For a full walkthrough of this stage, see how to analyze your chess games.

How Chessy helps: Chessy imports your Chess.com or Lichess games automatically and runs Stockfish on them. When a line doesn't make sense, you can ask the AI coach in plain English ("why was my bishop move bad here?") and get an answer based on engine-checked moves.

7. Turn one lesson per game into training

End each review with one sentence: "Next game, I will…" For example, "check every capture my opponent has before I move" or "not open the centre while my king is uncastled". One concrete rule is far more useful than a page of variations.

Then drill it. The best tactics practice uses positions from your own games, because those are the patterns you actually miss. Puzzles from your own games beat random tactics for exactly this reason. Across many games, your labels from the annotation step show your real weaknesses. This guide on finding your chess weaknesses explains how to read them.

How Chessy helps: Chessy turns your blunders into puzzles and tracks recurring mistake patterns, so the work from your reviews keeps paying off.

Common mistakes

• Turning the engine on first. Once you've seen the evaluation bar, you can't unsee it. Every "self-analysis" after that is just agreeing with the computer. Keep it off until you've written your own verdicts.
• Analysing every move equally. Long reviews of quiet moves feel thorough but teach little. Pick the three to five moments that decided the game and go deep on those.
• Reviewing only losses. Wins often contain the same mistakes, just unpunished. Review a sample of wins too, especially ones where you were worse at some point.
• Writing variations without verdicts. A line with no conclusion is useless the next time you see it. End every variation with plain words: winning, equal, losing, or unclear and why.
• Blaming the result on vague causes. "I was tired" or "bad opening" isn't a lesson. Name the specific move and the specific thinking error, then file it under a category.
• Skipping review because the game was blitz. Fast games still show your habits. Give them a quicker pass: just the one or two moments where the game turned, and the label for each.

FAQ

Reviewing without an engine is slower at first, but it's how you find out why you make mistakes, not just where. Keep it to a few critical moments, write honest verdicts and label every mistake. Only then let Stockfish mark your work. Do that for ten games and your mistake labels will show you exactly what to train next.

Sources

• Lichess analysis board — Replaying games with the engine switched off and using the opening explorer
• Chess.com Analysis Board — Alternative board for engine-off review and later engine checks
• Wikipedia: Blunder (chess) — Definition of blunders and annotation symbols
• Chess Programming Wiki: Stockfish — Background on the engine used for the final check
• Chessy: How to analyze your chess games — Detailed engine-assisted analysis stage