How to analyze your chess games and actually improve
Reviewing your games is the single most effective way to get better. But only if you do it right — most players do it backwards. Here's the method that works.
Every strong player will tell you the same thing: you improve fastest by studying your own games, especially your losses. Yet most people either skip it entirely or "analyze" by clicking the engine button, glancing at the red moves, and closing the tab. That's not analysis — it's confirmation that you lost. Here's how to turn a finished game into a lesson you'll actually remember.
Replay the game yourself — engine off
Before you switch on any engine, step through the game from memory. Where were you unsure? Where did your opponent's move surprise you? Where did you spend real time deciding? This captures what you actually saw over the board — which is the thing you're trying to improve. Turn the engine on first and you'll just nod along to its answers without ever testing your own.
Mark the critical moments
Games aren't decided evenly — they turn on a handful of moments. Find them: the moves where the evaluation swung, where material changed, or where you felt the position get away from you. You don't need to scrutinize all 40 moves. You need the three or four that mattered.
Bring in the engine for the truth
Now use Stockfish. The engine is far stronger than any human and will show you, without guesswork, exactly where the position turned and what the best move was. This is the "what" — objective, reliable, and the foundation for everything else. Free tools like Lichess or any browser Stockfish analyzer do this well.
Understand why — this is the part that matters
Here's where most analysis dies. The engine tells you 5...Bxd1 was a blunder and that 5...Nxe5 was better. Fine — but why? What idea did you misread? In this case, the "hanging" knight was bait, the pin was an illusion, and you walked into a forced mate. Until you can articulate the idea in words, you haven't learned anything you can reuse. A number can't do this for you; a coach can.
Extract one lesson — and track patterns across games
Write down one or two lessons per game. "Don't capture before asking what my opponent gets in return." Over five or ten games, these notes start repeating. That repetition is the gold: it's your personal weakness, and now you can train it directly instead of studying at random.
Skip the manual work
postgame does steps 3–5 for you — engine truth, plain-language explanation of why, and the patterns it spots across your games.
Analyse a game free →3 free analyses · No card needed · Import from Chess.com & Lichess
How long should this take?
Less than you think. You don't need to spend an hour per game. Even one or two careful reviews a week will surface valuable insights, and a focused 10-minute review of a single loss beats an hour of mindless puzzle-grinding. Consistency matters more than depth here — the patterns only emerge if you keep doing it.
The honest shortcut
The method above works, but it asks a lot: you have to be disciplined, know what to look for, and be honest about your own play. That's why most people don't keep it up. Tools like postgame exist to remove the friction — Stockfish finds the turning points, Claude explains the idea you missed in plain language, and because it remembers every game you run through it, it tells you the pattern across your games so you don't have to keep your own notebook. The discipline becomes automatic, which means you'll actually do it.
However you do it — by hand or with help — the principle is the same. Don't just confirm that you lost. Understand why, find the pattern, and train the one thing. That's how a game becomes improvement.