A concret way to analyse your games.

The benefits of the analysis are multiple:

• comparisons of our ideas with the hard reality of tactics and positional play.
• development of our thinking process.
• progressive improvement of our knowledge about openings (and eventually about positions and endgames).
• detection of recurring errors.

According to Yusupov, what to expect from an analysis:

1. "find the critical moments" in the game, those moments when mistakes were made, the nature of the position changed, etc.
2. "Look for the reasons for your own mistakes" - not just what went wrong, but why.
3. "Look for new possibilities." What moves did you miss in your analysis? What ideas might you have considered?
4. "Reflect on the opening.

Step 1: An analysis is anticipated.

During the game, record your thinking time. (Well... you have to admit that sometimes there is not enough space on these tournaments sheets and that in the heat of the moment it is not always easy)
Annotate your game sheet with cabalistic signs in order to spot particular moves (a cross for a questionable tactic, an exclamation mark to spot an opponent's move that surprised you, etc.)
Just after the game, write down your impressions in more detail (useful to have a notebook with you)
If possible during a tournament: post-mortem analysis with your opponent (in fact this should be mandatory, if only for conviviality reasons). Note down the ideas, what he saw on his side.

Step 2: right after the game

As soon as possible after the game anyway. During a tournament, it will be in the evening if you have the courage! During an online game, right after.
Put the game in a database and transcribe your notes of the day. Of course, without launching the analysis engine.

This is the time to spot the critical moments. First, in a simple way: identify the opening-mid game and mid game-final transitions. The first critical moment is at the exit of the opening (whether it is you or the opponent who causes it). The loss of a piece is another. In a more subjective way, it is also necessary to look for this moment when the two positions seem in balance and that each camp arrived at the maximum of the exploitation of the position (possible breaking point). When the engine is started it will also indicate important options that have been missed. It is not useful in this step to make a novel of it, but to put one of these cabalistic signs if you did not have the opportunity to do it during the game. (^^, *, # or any other special sign)

Then you should also look for the moments when you felt you were not doing well, the moments when you had no idea, the unclear positions that made you choose a simple move instead of deepening your calculation. These are the key moments that you will analyze in detail later.

Finally, and only for the opening: check what can be improved or avoided (Analysis engine, database). Later on, focus on the reasons why you didn't play the best moves during this opening.
During a tournament, it is not useful to do more in order to keep confidence. In any case, in order not to devalue yourself: never say to yourself "What an idiot not to have seen that!" or "Shame, to have been beaten by a player with 200 elo less than me!", but "It was hot in the room" or "I was thirsty, I should bring another bottle of water".

The analysis engine was used only to check the first few shots of the opening.

Step 3: Several days later.

Of course the help of a stronger player than you is the best solution, but otherwise, the analysis engine will be the reference. Especially since it is almost impossible (with 24-hour days) to review the game, move after move, spending as much time as the game itself. Stockfish's analysis allows us to adjust the evaluation we feel at critical moments (I think it is useful to give an evaluation on each of these positions), then to compare it later with the analysis of an engine (each engine will have its own evaluation, so we will have to rely on the same engine each time).

The situation has settled down, you can approach your game with a new eye. It is now a matter of calmly and slowly reviewing the game, considering the options you did not develop (variants and sub-variants), or understanding how you could have avoided the loss of a piece or the control of a square (column, diagonal).

When a position seems unclear, it is useful to use a database and look for similar positions played by strong players in order to observe their decisions. But the most important thing is to understand why you are in error. If calculation can be the cause, it is also possible that our state of mind leads us to make bad decisions. We think we are late? we attack at full speed by taking risks. We think we are ahead? We relax and the opponent takes over. Game after game, it is then possible to identify some main flaws.

And finally... : Stockfish, Komodo, Lc0 ! Drop the cavalry. Set the error detection according to your objectives (in my opinion, a threshold of 0.66 to 0.75 points is probably enough at a small level, 1 point if you are a beginner). Especially for beginners, if an error is detected every 4 to 5 moves, you should seriously think about reviewing your thought process and working on your tactics. Or set the error threshold to a slightly higher value! Conversely, if the analysis engine does not detect anything... refine the options.

Step 4: The assessment.

It's time to ask yourself some questions. Did you evaluate the position correctly? Did you choose a good plan ? How many tactical errors did you make? (lost piece or failed tactic, game prolonged when a mate in 3 moves was possible). It is very important to verbalize this step, to write it down.
For each game, write down three things you learned. For example:
"For this opening, the bishop exchange is not to be feared."
"In tactical, or desperate situations, always look to see if it is possible to check the opponent's king."
" In a complex R+P endgame, it is sometimes useful to block a pawn island before moving your king."

Look for flaws in your thought process, diagnose failed tactics, the times in the game when you were weak (opening, middle game, endgame). Then, when your tournament games have been analyzed, adjust your training program accordingly.

It is useful to submit your final analysis to a player (at least of your level) in order to limit bias (recursion). For example: self-flagellation for not having found a mate in 5 moves (with complicated variants and sub-variants), when a mate in 7 was possible with a simpler forced continuation. The forums allow to publish these analyses, if someone pays a careful and benevolent eye to them. It is a double-edged sword.

As you can see, a serious analysis takes time. For an OTB tournament, it's a good week's work in perspective. Outside of a tournament, your long online games (at least 30 minutes) are worth the effort. To a lesser extent, faster games should not be neglected.

Of course this activity is time consuming, but that's why there are so few Grandmasters!
Stejpan explains it all on Hanging Pawns : " How to detect and correct your middlegame mistakes " and " How to analyse chess games".

Step 5 : and after ?

The gain, if there is any, is probably on the long term. As far as the interest is finally to correct oneself, it is necessary to adapt one's training program to these analyses. By counting the types of errors (tactical - offensive or defensive, visualization concerns, strategy, endgame - rooks, pawns -, opening - with white, black -) one can thus focus on the weak points. Even simpler: find the last mistake that made you lose a game. Do it on at least 10 games and see what comes out.

From my french blog : "Les échecs sans peine"
In english here. Thanks to google (rough) translation.

You do not agree ? I forgot something ? Tell me !!