Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: estimating opening strength: how the heck does chessbase do it

Author: Joseph Ciarrochi

Date: 15:38:43 01/17/06

Go up one level in this thread


>The above formula assmes that the relationship is linear

Yes, I'm afraid so. Welcome to the simplified world of psychology:)

 We could test for other kinds of relationships. Maybe the key variable would be
a difference between black and white ELOS (this would simplify our analysis).
Then we could look for say, a logarithmic relationship between this difference
and performance. e.g., from 0 to 100 point difference, you get big difference in
predicted performance. In contrast, from  500 to 600 point difference, you get
little difference in predicted performance.

On January 17, 2006 at 18:33:32, Dann Corbit wrote:

>On January 17, 2006 at 18:22:02, Joseph Ciarrochi wrote:
>
>>actually, i think a more accurate discreption of the formula is:
>>
>>prob win = B0 +b1 white ELo - b2 black elo + b3* opening+ E
>>
>>B0 is a constant, b1 , b2 the relative importance of ELO in impacting wining
>>
>>YOu could look at interactions between ELO and opening, if you have a truley
>>massive data base (e.g., b1*b3  b2*b3)
>>
>>Now i must work, though this is more fun
>
>The above formula assmes that the relationship is linear



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.