Analyzing a game
Posted: Sun Jan 08, 2023 1:24 pm
I found a game of interest and imported it into COW. Certainly a good way to proceed is to play through the game, turning on the engine as deemed appropriate.
However, I decided to use COW features to help automate the process. I wanted to find major turning points, if any. These include major improvements and/or errors. This is what I did.
I imported the game to COW (one COW file, one game). I went to the EPD menu and issued the command "Export Positions ..." and exported all the positions of the game to a file. I turned on the engine and went to the EPD menu and issued the command "Analyze with Engine." COW proceeded to analyze each position, pasting the result into comment box of each position.
This is a useful process, but I think it still falls short and can be improved upon. I think players would like to see where there is a significant change in evaluation between that of the move played and the move recommended. If there is a very large gap, it would indicate that the move played was a either a losing move, or that the player missed the winning move.
As described above, COW analyzes each position and provides the score for the best move. Moving forward one ply shows the score for the actual move played. A simple subtraction of scores should highlight any blunder (say a difference greater than two), error (say a difference greater than one), or an inexact move (say a difference greater than .5). COW could then annotate the moves appropriately. To put icing on the cake, one could then jump to the next "game changer" which would be a mistake or blunder.
However, I decided to use COW features to help automate the process. I wanted to find major turning points, if any. These include major improvements and/or errors. This is what I did.
I imported the game to COW (one COW file, one game). I went to the EPD menu and issued the command "Export Positions ..." and exported all the positions of the game to a file. I turned on the engine and went to the EPD menu and issued the command "Analyze with Engine." COW proceeded to analyze each position, pasting the result into comment box of each position.
This is a useful process, but I think it still falls short and can be improved upon. I think players would like to see where there is a significant change in evaluation between that of the move played and the move recommended. If there is a very large gap, it would indicate that the move played was a either a losing move, or that the player missed the winning move.
As described above, COW analyzes each position and provides the score for the best move. Moving forward one ply shows the score for the actual move played. A simple subtraction of scores should highlight any blunder (say a difference greater than two), error (say a difference greater than one), or an inexact move (say a difference greater than .5). COW could then annotate the moves appropriately. To put icing on the cake, one could then jump to the next "game changer" which would be a mistake or blunder.