Learning how to analyze a chess game starts with your own decisions: mark the moments you found difficult, explain what you expected, then compare those ideas with an engine. An evaluation graph helps locate turning points, but the useful result is a lesson you can apply in another game.
Load the completed game
Export one PGN from your chess platform and load it into Game Review. Alternatively, request recent public completed games by username. Public imports can fail or be limited by the provider; a PGN is a practical fallback.
Before starting the engine, replay the moves. Mark three positions where you spent time, felt surprised, or changed your plan. Write what you thought the opponent threatened. This gives the engine comparison a specific question to answer.
Run a first pass
Choose Economy for a short review or Deeper for a longer search. The current tool uses 0.75 or 2.5 seconds per nonterminal position. Device speed affects the depth reached, so these are useful search budgets rather than guarantees of perfect play.
The review follows the game’s main line. If your PGN contains study variations, use the analysis board to explore them separately. You can pause a review and resume; the most recent review is saved on the device when browser storage is available.
Inspect changes, not just low scores
Positive evaluations favor White and negative ones favor Black. A score moving from −3 to −1 is an improvement for White, even though both values are negative.
The loss column estimates the moving side’s decline by comparing evaluations before and after a move. It is clamped at zero, and mate transitions are labeled separately. It is not an official accuracy percentage, a performance rating, or a universal definition of a blunder.
Select an interesting graph bar to see the corresponding position. Open that position in the analysis board and try the engine’s alternative. Look at the opponent’s reply as carefully as your own move.
Work through one concrete example
Consider this legal study game:
1. e4 e5 2. Nf3 Nc6 3. Bc4 Nd4 4. Nxe5 Qg5
5. Nxf7 Qxg2 6. Rf1 Qxe4+ 7. Be2 Nf3# 0-1
White’s knight attacks material, but Black’s queen creates threats against the king. Rather than stopping at “White lost,” inspect the position before 5.Nxf7. What did White expect to win? What checks and threats did Black gain? Replay the final sequence to see why collecting material did not address king safety.
This is a worked tactical example, not a claim that one trap is a sound complete opening repertoire. In your own games, a quieter positional mistake may require more explanation than a forced mate.
Turn the finding into one exercise
Use a compact note:
- Position: save the FEN or the move number in the PGN.
- My assumption: what you expected before moving.
- Missed reply: the opponent’s critical resource.
- Recognition cue: a feature you could notice earlier next time.
If you missed a fork, practice relevant chess puzzles. If development stalled in a familiar opening, revisit an opening lesson. Choose the exercise from the actual mistake rather than studying unrelated material.
Recheck important positions
Short engine searches can revise their score or preferred move with more time. Use a longer search for critical tactical positions and read what engine evaluations mean. One unusual result is a reason to investigate, not to label your entire game or playing strength.
Start with one completed game, and finish with one specific lesson.