How to Analyze a Chess Game Move by Move: A 4-Step Method

You just lost a game, and you want to know why. The quick reflex is to open an engine and scroll through the evaluation bar. It feels like analysis, but it rarely teaches you much. You see the engine's suggested move. You don't see why you missed it.

This guide gives you a method for analyzing a chess game move by move that builds real understanding. It has four steps. Write down your thoughts first. Check the critical moments. Let the engine confirm. Then log each mistake by type, so single games turn into patterns you can train.

Step 1: Write down your thoughts before any engine

Do this soon after the game, while your memory is fresh. Open the game on a board or in a viewer with the engine turned off. Go through it from move one.

For each phase, write short notes in plain language. You are recording what you thought, not what was true.

• Opening: Did you know the line? Where did you leave your preparation?
• Plans: What were you trying to do in the middlegame? What did you think your opponent wanted?
• Candidate moves: At key moments, which moves did you consider, and why did you pick the one you played?
• Clock: Where did you spend a lot of time? Where did you play too fast?
• Feelings: Were you confident, worried, or impatient at any point?

Be honest. A note like "I didn't look at his knight check at all" is far more useful than a polished excuse. These notes are the raw material for everything that follows.

A simple note format

Keep it short. Something like this works:

• Move 14, Black: I thought my king was safe, so I grabbed the pawn on b2.
• Move 21, White: I considered Rxe6 but rejected it because I saw a back-rank problem.
• Move 30: Low on time, I moved fast and didn't check his checks.

Step 2: Find the critical moments

Not every move deserves the same attention. A game usually turns on a handful of moments. Your job is to find them before the engine points them out.

Look for these signals in your notes and in the game:

• Moves where you spent a long time thinking.
• Moves where the position changed character: a pawn structure fixed, queens came off, or the center opened.
• Moments where you felt the game slipping, or suddenly felt winning.
• Captures and recaptures where you had several options.
• The last few moves before the result became clear.

Mark each critical moment and try to analyze it yourself first. Set up the position, list your candidate moves, and calculate the main lines on the board. Write down your conclusion. This is a good way to train calculation with your own games.

It is fine to be wrong here. The point is to have a clear guess that the engine can confirm or correct.

Step 3: Let the engine confirm, not lead

Now turn the engine on. Go back to your critical moments first, not to move one. Compare its verdict with your own.

A few terms you will see:

• Evaluation: the engine's estimate of who is better, usually shown in pawns or centipawns (hundredths of a pawn).
• Inaccuracy, mistake, blunder: labels for moves that worsen the evaluation, from minor to severe.
• Suggested line: the sequence of moves the engine expects after its top choice.
• Depth: how far the engine searched. More depth usually means a more reliable answer.

For each critical moment, ask three questions:

1. Did the engine agree that this moment mattered?
2. Was my move good, and if not, what was better?
3. Do I understand why the better move works, in words, not just in moves?

If you can't explain the engine's move, play out its suggested line on the board. Then try a natural human alternative and see how the engine refutes it. Stop when you can state the idea in a sentence.

Next, scan the rest of the game for big evaluation swings you didn't flag. Those findings matter a lot. They show what you don't see yet.

Common traps at this step

• Chasing tiny differences between two good moves. If both keep a similar evaluation, move on.
• Accepting a move you don't understand. Copying the engine's choice without the idea teaches nothing.
• Judging a decision only by the result. A sound decision can still lose, and a bad one can get away with it.

Step 4: Log each mistake by type

This step is what turns analysis into improvement. One game shows you a mistake. Ten games show you a habit.

Keep a simple log in a spreadsheet or a notebook. Give each real mistake one row and one type.

For each row, also note the game, the move number, what you thought (from Step 1), and what was true (from Step 3). Add a one-line lesson, such as "Check all opponent checks before grabbing pawns."

After a few weeks, sort the log by type. The categories that keep filling up are your training priorities. If tactical oversights dominate, do puzzles. If endgame technique shows up again and again, study basic endings.

From one game to a pattern: where a report fits

The method above works with one game at a time, and the thinking you do yourself is the part no tool can replace. The hard part is volume. Logging enough games to see reliable patterns takes time.

That is the job Rookprint does. It produces an AI-generated report from your last 200 online games analyzed by Stockfish. It names your 3 recurring weaknesses with numbers, builds 50 exercises from your own mistakes, and adds a 30-day plan.

Here is a worked example with the prices shown on the site on the day of writing. The Report is $29 (VAT not applicable, article 293 B du CGI). Progress is $49 and includes everything in Report with 100 exercises, plus a second report 30 days later that compares before and after. The Club pack is $149 for five reports for five players, each with 50 exercises and a 30-day plan. The full grid is on the pricing page, and you can order from the order page.

What it doesn't do: it works from your online games only, so it can't know what you were thinking at the board. It is not a human coach. Your own notes from Step 1 still matter. Other questions are answered in the FAQ.