# What kind of statistics should I run on a database? [closed]

I have written a python script to verify the strategical rule of "controlling the center".

I have parsed the free 2.2 million games database (it takes about 5 minutes) and I have found that the winner of a game controls the center at any given time with (on average) 1 more piece than the losing side.

I'd like to look at the correlation of other strategical principles with winning. What would you suggest would be an interesting list to look into? (I'm thinking of castling vs no castling, knights on the rim...)

EDIT: I would like suggestions for interesting statistics that you would like me to run. I could then send an article (say, to chess.com, or TWIC) with the results, referencing the suggestions posted here.

• What is your definition for "controlling the center"? Also, do you calculate this for the whole game or only for the opening phase (until which move?) where it is most important? – user1583209 Mar 8 '17 at 10:57
• Right. My definition is "number of pieces attacking the D4,D5,E4,E5 squares". For each move I take that number for the winner and I subtract that number for the loser. Then I average over the whole game. I could also average over the first N moves, of course. – Ziofil Mar 8 '17 at 15:40
• If you want to draw any conclusions from it, I am not sure "control of the center" is of any relevance in an endgame and you should probably consider some cut-off for the number of moves as otherwise you might just measure spurious effects like winners having more material and therefore on average more control of the center (or of any other part of the board). Actually for fun, you could see what statistics you get for the control of the whole board. – user1583209 Mar 8 '17 at 20:11
• You are right, in fact what I'm doing now is I'm checking the correlation for each cutoff (the code is almost identical), to see up to which move number control of the center matters (of course, on average). – Ziofil Mar 8 '17 at 22:09