How do I actually improve from analyzing my games?

November 7, 2026 · 5 min read

A rapid player rated around 1260 asked a question on Reddit that almost everyone who reviews their games ends up asking. They open the analysis after every game, they see where they went wrong, they sometimes understand what the engine wanted. And then the next game is different, and none of it seems to apply.

The feeling is fair. Every game is different, and “on move 23, play Rd1 instead of Qc2” will never come up again. But the reasons behind your mistakes do come up again, often in almost every game. Improving from analysis means moving from the first kind of lesson to the second.

Using analysis is not doing analysis

The most upvoted answers on that thread all said the same thing: the engine report is not the analysis, it is the answer key. If you look at the answer key first, you learn that a move was bad, and nothing about why you played it.

The useful part happens before the engine. You go back to the moment you felt unsure, you take the time the clock did not give you, and you try to find the better move yourself. Then you check. When you were right, you learn you could have found it. When you were wrong, you learn exactly which idea you are missing.

The method, step by step

Right after the game, before any engine, write down the two or three moments where you hesitated or felt the game turn. Your memory of what you were thinking is worth more than any evaluation bar, and it fades within hours.
At each of those moments, think again, without a clock. Write down the move you would play now, and the line you expect.
Only then open the engine. Look at where the evaluation really dropped: it is often a few moves before the moment you noticed.
For each real mistake, answer one question: why did I play it? Pick one cause: I did not see the move, I saw it and misjudged it, I calculated a line wrong, I did not look at what my opponent threatened, I was short on time, I did not know the opening.
Keep a short log: the game, the move, the cause. One line per mistake is enough.
Every ten games or so, count the causes. One or two of them will show up far more often than the others. That is what you work on.
Turn the top cause into a question you ask during your games. “What does my opponent threaten after my move?” if you keep missing threats. “Is anything of mine undefended?” if you keep losing pieces. “Did I check every capture?” if you keep missing tactics.

Why this works when every game is different

Positions do not repeat, but causes do. If your log says that most of your lost points come from moves you played without looking at your opponent’s reply, that is true in every game you will play, whatever the opening and whatever the position. The question you ask at the board is the concrete application the Reddit player was looking for.

The same goes for the opening, where positions do repeat. You probably play the same two or three openings in most of your games, so a mistake in your opening is a mistake you will make again next week. Those are the easiest points to win back: learn the right move in that exact position, and it is fixed.

The honest part: it takes a lot of time

Done properly, reviewing a game takes longer than playing it. Some coaches spend several times the length of the game on its analysis. For a rapid game, that is easily an hour, and the log only becomes useful after ten or twenty games.

That is why most players stop after a week. The method is not wrong, it is just more time than an evening after work allows, every day. And then, most of the time, there is no review at all: the game is lost, you press “new game”, and the same mistake waits for you in the next one.

When you have no time, a video beats nothing

Between a full review and no review, there is a middle ground. You ask a question about your games, “why do I keep losing with Black?”, “what goes wrong in my Italian?”, “what did I miss in my last blitz session?”, and you watch a short video that answers it on your own positions.

That is what Chess X-Ray does. It reads your chess.com or Lichess games, finds the moments where you lost the most, checks them with Stockfish, and explains them in a few minutes of video: the move you played, the one you missed, and the idea behind it, in plain words. Because it reads many games at once, it also sees what a single review cannot: the same mistake coming back across weeks.

It does not replace doing the work yourself. Watching an explanation is not the same as finding the move, and the players who improve fastest still do the full method on the games that matter. But on the days you would otherwise skip the review, ten minutes of video on your own mistakes is worth far more than nothing.

A simple weekly routine

One or two games a week, the ones that hurt: the full method, with the log.
The other days: one question about your recent games, and the video that answers it.
Every game: the one question from your log, before every move.

The short version

Do not start with the engine. Find the critical moments and the better moves yourself, then check.
Write down why you made each mistake. Causes repeat even when positions do not.
Turn your most common cause into one question you ask at the board.
Fix the mistakes in your own openings first: those positions come back.
The method takes time. When you do not have it, watch a short video on your own games rather than skipping the review.

If most of your mistakes happen early, read how to stop blundering in chess or open the guide of your opening in our opening guides.