1. Get the game as PGN
PGN, or Portable Game Notation, is the standard text format for a chess game. It stores the players, result, date, and move sequence. Open the completed game on Chess.com and use its share, download, or PGN option. The exact location can differ between desktop and mobile layouts, but the goal is the same: copy the full PGN text or download the .pgn file.
Keep the headers when possible. They help you connect a position to the opponent, time control, color, and result. If you review several games later, those details reveal patterns: perhaps most losses with Black occur in rapid games, or one opening repeatedly produces bad positions before move ten.
You can also use the game import page for supported public game data. When a direct import is unavailable, pasting PGN into the Plychess Analysis Lab is the simplest universal method.
2. Analyze yourself before turning on Stockfish
This step separates training from entertainment. Replay the game and stop at every moment where you spent significant time, felt surprised, or changed plans. Write down the candidate moves you considered. Do not try to prove that your move was correct; reconstruct what you actually saw during the game.
For each critical position, answer four questions: What did my opponent threaten? What forcing moves did I have? Which piece was worst placed? What pawn break changed the position? These questions create a human diagnosis that can later be compared with the engine's concrete line.
Marking uncertainty is more valuable than marking only mistakes. A move may be objectively fine even when your reasoning was wrong. That hidden weakness will cost points in a slightly different position unless it is corrected.
3. Import the PGN and choose sensible engine settings
Paste the game into Plychess and start analysis. Stockfish runs in the browser, so you can begin without an account. A moderate search is normally enough to identify the shape of the game. Do not wait for an extreme depth on every quiet move: spend computing time on positions where the evaluation changes or where your human notes show uncertainty.
One principal variation is best for a fast review. MultiPV is useful when comparing several plausible plans, but additional lines are weaker and can create noise. Read our guide to Stockfish depth, evaluation, and PV if the output is unfamiliar.
4. Find the first meaningful turning point
Players often jump to the largest late-game blunder. The more instructive moment is usually earlier: the first decision that made the position difficult to play. A missed opening break may create a cramped middlegame; a careless exchange may leave a bad endgame; a slow move may hand the initiative to the opponent.
Move backward from each large evaluation swing. Ask whether the final error was forced by an earlier strategic problem. If you blundered a piece under time pressure, the lesson might be clock management or a poor plan that consumed ten minutes, not the final one-move oversight.
5. Translate engine moves into chess language
Never save a lesson such as “Stockfish wanted 18.Rae1.” That statement is too narrow to transfer to a new game. Instead explain the reason: the a-rook belongs on the open e-file because the f-rook must continue defending f2; the move also prepares e5 and removes a back-rank tactic. A useful note contains a feature, a plan, and a trigger.
Compare the engine's first choice with your candidate. Play the engine response, then try your intended continuation. When your line fails, identify the exact tactical motif: a pin, deflection, weak back rank, or overloaded defender. Add that motif to a puzzle-training list.
6. Review the opening without memorizing the whole database
Locate the first move where the game left your known preparation. Check whether your move violated a plan or was simply a playable alternative. If the position came from the Sicilian, French, Italian, or another covered system, compare it with the Plychess opening guides and interactive model line.
Add only one correction to your repertoire after a game. Save the position just before the mistake, the preferred move, the opponent's most challenging reply, and one sentence explaining the plan. A compact correction reviewed repeatedly is more useful than importing thirty engine moves.
7. Separate middlegame and endgame lessons
In the middlegame, classify the error by decision type: missed tactic, wrong pawn break, poor exchange, unsafe king, inactive piece, or calculation failure. In the endgame, identify the underlying technique: opposition, rook activity, passed-pawn calculation, a Lucenaor Philidor setup, or an incorrect king route.
Recreate the critical endgame with the fewest necessary pieces and play it against the board from both sides. Endgame knowledge becomes reliable only when you can execute it without following arrows.
8. Build a five-minute post-game record
Finish with a short record: one opening correction, one tactical pattern, one strategic idea, and one practical factor such as time management. Tag the position by theme rather than result. After ten games, count repeated tags. The recurring category, not the most painful individual loss, should shape next week's training.
- Write down where you left opening preparation.
- Mark every move where your candidate moves felt unclear.
- Identify the first mistake that changed the character of the game.
- Save one tactical pattern and one strategic lesson.
- Create a position to revisit, not a list of engine moves to memorize.
Common analysis mistakes
Running the engine immediately encourages hindsight bias. Treating every small numerical change as an error creates false precision. Copying a long principal variation creates memory without understanding. Finally, analyzing only losses hides valuable moments from wins where the opponent failed to punish a bad decision.
Review wins, losses, and draws with the same process. Use engine evaluation as evidence, not as a grade. The goal is to predict better moves in future games, not to produce a perfect report about a game that is already finished.