TalkChess.com

[Tom Giampietro]

So I am forever curious about trying to improve my analysis method for my post game analysis of my real time or OTB games.

I have gotten better at it of late but am very curious about what some more experienced players might have to say.

Preferred method for doing post mortems on one's own games?

Preferred engines? Why?

My current method is as follows:


After going through my most recently played game a few times on my own and annotating what I can figure I then turn to some computer help.

I take one engine and hide the engine's move choices (able to do this in the Shredder Classic GUI). With the move choices hidden I go through the game backwards and pay attention to the engine's evaluation (which I can still see), when a spike in evaluation happens I stop the engine and try to figure out what the better move and or plan might have been on my own. After giving that I go, then and only then do I look at the engine's suggestion.

Once I have gone through the game in that fashion I go back to the critical sections of the game and see if a few other engines have any alternative thoughts or plans. I also will give the opening a going over with a few engines as well.

This has proven to be beneficial so far but I would love to hear other people's thoughts for post game analysis geared towards improving one's own chess game.

[H.G.Muller]

Well, I never play games myself, but when I would like to analyze one I would load it into WinBoard, use its Analyze Game function to score each position with an engine, look in the Eval Graph for sudden drops of my score. I would then go to the position just before the earliest one first (or perhaps the largest, if there really is one that dominates them all), byclicking on the eval graph. There I would let the engine analyse somewhat longer, to make sure that the score drop was not just because I was already doomed but the engine could not see it yet in the original annotation.

Then I would probably immediately start looking at the PV of the move recommended by the engine (lazy as I am), right-clicking it in the Engine-Output, and walking through it on the variation board to see why the engine thought it was so much better than the actual game. If, towards the end of that PV the engine would start to suggest moves that IMO do not make sense, I would release the PV walk just before that point, to let the engine continue analysis from there (pressing Shift down during the release of the mouse button, to make sure the added PV moves would be added as a variation, and not erase the loaded game). This to confirm that the engine did not just think this variation was good for me because there were some very poor opponent moves towards the end of it. After that, if the good score before my move would be confirmed, I would Revert to the original game, and probably switch the engine to multi-PV (by clicking in the Engine-Output window), to see if I failed to find a unique refutation, or whether I picked an exceptionally poor move.