How to use the chess engine to analyze move by move
At this point I have all the evaluation scores and alternative evaluations made by the chess engine inserted into the game played. I step through the all the moves, with the computer analysis turned on (in Chessbase called infinite analysis, in Chessok Aquarium called start analysis). This is actually an unusual example of a point at which Chessbase has a worse GUI than Chessok Aquarium, since copying notations from the analysis into the game makes Chessbase start calculation a new, wasting a lot of time if you analyze deep.
What to look for in the move by move analysis
At this point of time, I am paying attention to suggestions on other moves for my color. I am also paying attention on differences in evaluation score where there is no alternative suggestion. Such occurrences often means that the evaluation time set to the chess engine was too low. It is a good chance the difference exist not due to a mistake, but due to the chess engine having a very different evaluation one move later, when now at next step seeing more of the consequence of the move.
Since the analysis is turned on, for each move I see what is currently the top suggested lines (3-5 lines is what I normally have on). I also see if they are the same as the finally made suggestion if such exist. And I see if these moves are changing score and order. If the analysis is fairly stable, and similar to the one written down, then I do not wait for the analysis to recalculate, I just check the existing alternative suggested by the engine during the first go through. I normally do that by first trying to understand why the different move is better, then by stepping through the suggested game play in that suggestion to get a visible picture of what is going on and if I would like to play such a line.
- If during analysis for a certain move the chess engine keeps changing its mind, this is a place where a deeper analysis should be allowed before trusting the advice. Then I let it run until I judge it as stable. This means normally at least 23 24 moves. In the opening or complex situations with a lot of pieces, it might mean up to 28-30 moves.
- If the chess engine states something which I think is wrong based on my own pure positional thinking on what is good to do (no clear blunder and an evaluation score differing less than 0.2), then this can also be a case where I allow the engine to analyze a bit deeper.
- If I do not still really understand, I try entering the moves I think I or the other player should/could do and looks at what the engine then recommends as counter play. It often gives me a better chance to understand why.
Many time after this stage of the analysis, I can now see why the engine thinks a certain way could be better. This is then often a good thing to note down in the annotation, saving it to come back to it later. Some times I conclude I am not going to want to play the advice, it feels wrong, or it is too complicated for me.