League B finished, League C started

[George Lyapko]

Geschrieben von: / Posted by: George Lyapko at 26 May 2000 08:23:21:
After incredible last round Chop and Inmi advances to League A,
Chessterfield and Dchess goes to League C...
League C: 3d round running...
More details at http://www.geocities.com/lyapko/winboard.htm
Best regards,
George

[Colin Frayn]

Geschrieben von: / Posted by: Colin Frayn at 26 May 2000 12:58:23:
Als Antwort auf: / As an answer to: League B finished, League C started geschrieben von: / posted by: George Lyapko at 26 May 2000 08:23:21:

After incredible last round Chop and Inmi advances to League A,
Chessterfield and Dchess goes to League C...
League C: 3d round running...
More details at http://www.geocities.com/lyapko/winboard.htm

George,
There are some very strange results there from many programs. Are you using the large opening book for ColChess from my web page? Are you using my executable or did you compile your own?
Example;
From your web page;
ColChess - TSCP +3 =2 -7
Match I played recently;
ColChess v TSCP +37 =3 -0
...and the time limits were 2 games (1 as white and 1 as black) from each of 1 minute to 20 minutes. All the draws were at short time controls. At long controls ColChess never came close to drawing. With the latest opening book it should totally slaughter TSCP as I've coded in most of its common lines of play.
Again;
Your Page:
ColChess - Faile +5 =0 -7
Recent game I played; (10/Blitz)
ColChess - Faile +10 =0 -0
Game I played a fortnight ago; (40/Game)
ColChess - Faile +17 =3 -0
Also your ELOs seem rather different to some others that I've seen, though admittedly these have mainly been from Blitz.
Any ideas why this could be?
I think there is a problem with using data from old versions. For lots of programs there isn't much of a change from month to month, but I know for a fact that ColChess has gained at least 100-150 ELO since January, probably more. It has also gained an opening book, amongst other things.
Perhaps you should only include games played in the last month or something in your ELO calculations?
Cheers,
Colin

ColChess Homepage

[WYx]

Geschrieben von: / Posted by: WYx at 26 May 2000 13:15:26:
Als Antwort auf: / As an answer to: Re: League B finished, League C started geschrieben von: / posted by: Colin Frayn at 26 May 2000 12:58:23:

There are some very strange results there from many programs.
Are you using the large opening book for ColChess from my web page?

Yes, you are right !

In my humble opinion,
chessengine strenght not depend on opening book so much!
ColChess - TSCP +3 =2 -7
against
ColChess v TSCP +37 =3 -0
Or not?
regards
WYx
ps.: colchess is stronger, as tscp, that certainty!

[Inmann Werner]

Geschrieben von: / Posted by: Inmann Werner at 26 May 2000 14:38:29:
Als Antwort auf: / As an answer to: League B finished, League C started geschrieben von: / posted by: George Lyapko at 26 May 2000 08:23:21:

After incredible last round Chop and Inmi advances to League A,
Chessterfield and Dchess goes to League C...
League C: 3d round running...
More details at http://www.geocities.com/lyapko/winboard.htm
Best regards,
George

Nice last round for me and really incredible...
Werner

[Lyapko George]

Geschrieben von: / Posted by: Lyapko George at 27 May 2000 09:54:18:
Als Antwort auf: / As an answer to: Re: League B finished, League C started geschrieben von: / posted by: Colin Frayn at 26 May 2000 12:58:23:

George,
There are some very strange results there from many programs. Are you using the large opening book for ColChess from my web page? Are you using my executable or did you compile your own?
Example;
From your web page;
ColChess - TSCP +3 =2 -7
Match I played recently;
ColChess v TSCP +37 =3 -0
...and the time limits were 2 games (1 as white and 1 as black) from each of 1 minute to 20 minutes. All the draws were at short time controls. At long controls ColChess never came close to drawing. With the latest opening book it should totally slaughter TSCP as I've coded in most of its common lines of play.
Again;
Your Page:
ColChess - Faile +5 =0 -7
Recent game I played; (10/Blitz)
ColChess - Faile +10 =0 -0
Game I played a fortnight ago; (40/Game)
ColChess - Faile +17 =3 -0
Also your ELOs seem rather different to some others that I've seen, though admittedly these have mainly been from Blitz.
Any ideas why this could be?
I think there is a problem with using data from old versions. For lots of programs there isn't much of a change from month to month, but I know for a fact that ColChess has gained at least 100-150 ELO since January, probably more. It has also gained an opening book, amongst other things.
Perhaps you should only include games played in the last month or something in your ELO calculations?

First of all, my contiguous tournament have been started at November, 30 1999.
All results from any version of programs are included in rating calculation.
It is the same like you play on FICS(ICS): you get an account, and you always play with a same account, even if your playing strength desperately changes.
About Colchess:
Here are some results that you mentioned(until round 32 i didn't mark versions in PGN's):
Round 25(10.02 - i think,v5.4)
Tscp(1985)-Colchess(-)=7:1(0 1 1 1 1 1 1 1)(2040-1722)
Round 35(14.03 - v6.0)
Faile(1838)-Colchess(1801)=2:2(0 1 0 1)(1834-1805)
Round 46(26.04 - v6.1)
Faile(1938)-Colchess(1843)=5:3(1 1 0 0 1 0 1 1)(1942-1839)
Round 52(22.05 - v6.2.3)
Tscp(1855)-Colchess(1807)=1:3(1/2 0 0 1/2)(1820-1842)
I use Colchess_623.exe from your package(16.05.2000 - 369070 bytes)
Book.dat size is 1044468, and it is book_large from Dann Corbit's package (though I can't run his executable on my hardware)
About rating's:
As you can see, my ratings are based on FICS server system (with one differnece:
when two newcomers plays against themselves, thet get 2200), so if a program's
strength desperately grows, the rating also grows faster(an example is switching back to Dragon v3.11, or Gromit v3)...
Another thing: Olithink's rating is so low, because if it makes an invalid move, I award a win to opponent. It's real playing strength in blitz is much higher (~2000 by me). Gnuchess4 is excellent at 40moves/5min, but not so good if you play game in 5min, or 40moves/20min, for example, and so on...
By the way, I have decided to play more games with longer time controls...
Sorry for my English and have a nice day,
George