Author: Dann Corbit
Date: 14:15:11 05/21/01
Go up one level in this thread
On May 21, 2001 at 15:09:22, Uri Blass wrote:
>On May 21, 2001 at 14:04:40, Dann Corbit wrote:
>
>>On May 20, 2001 at 13:55:15, Theo van der Storm wrote:
>>
>>>1st International CSVN Tournament
>>>
>>>1. Fritz 8.0 45.5
>>>2. Gambit Tiger 2.0 8.0 45.0
>>>3. Diep 5.5 47.5
>>>
>>>Congratulations to Frans Morsch and Mathias Feist,
>>
>>The PGN program (originally from your site) puts Gambit Tiger 2.0 first instead
>>of Fritz. Has there been some bug corrected in the program?
>
>This pgn is not fair.
>The opponents of tiger that did not play against Fritz scored 11.5 out of 26
>games
>The opponents of Fritz that did not play against tiger scored 11 out of 24
>games.
>
>11/24>11.5/26 so it is clear that the opponents of Fritz scored better
>Comparison of only 11.5 and 11 is clearly unfair.
>
>If you do not include Celes ( I think that a program that gets 0% should not be
>included in the games of the opponents of tiger and Fritz for tiebreak decision)
>then Fritz's situation is even better.
>
>In this case we have 11/24>8.5/23
Here is the ranking by ELO calculation {using ELOStat}:
Program Elo + - Games Score Av.Op. Draws
--- ----------------- ---- --- --- ----- ------ ------ ------
1 Fritz : 2782 228 543 9 88.9 % 2421 0.0 %
2 Gambit Tiger 2.0 : 2755 228 281 9 88.9 % 2394 22.2 %
3 Diep : 2478 278 284 9 61.1 % 2400 11.1 %
4 Patzer : 2442 291 229 9 55.6 % 2403 22.2 %
5 The King : 2429 272 272 8 50.0 % 2429 25.0 %
6 Yace : 2413 278 284 9 61.1 % 2335 11.1 %
7 Insomniac : 2402 291 229 9 55.6 % 2364 22.2 %
8 SOS : 2346 291 229 9 55.6 % 2307 22.2 %
9 Tao : 2332 272 272 8 50.0 % 2332 25.0 %
10 Goliath Chess X : 2319 291 229 9 55.6 % 2280 22.2 %
11 Ant : 2294 198 325 8 43.8 % 2338 37.5 %
12 IsiChess X : 2281 198 325 8 43.8 % 2325 37.5 %
13 Quark : 2234 323 291 9 44.4 % 2272 0.0 %
14 SpiderGirl : 2203 229 229 9 50.0 % 2203 33.3 %
15 GoldBar : 2092 387 310 8 37.5 % 2181 0.0 %
16 Xinix : 2083 323 296 8 31.2 % 2220 12.5 %
17 Mat(h)! : 2035 323 296 8 31.2 % 2172 12.5 %
18 Crux : 2029 323 296 8 31.2 % 2166 12.5 %
19 31337/Celes : 1578 0 0 8 0.0 % 2178 0.0 %
The ELO value would support Fritz for first (if only just and hardly
conclusive). I suspect your assertion is correct, which indicates a bug in the
program pgn.c (perhaps).
Of course, a Swiss format tournament is the worst to calculate ELO from, except
for the first place entry. The middle calculations (especially) will be pretty
much a random number generator.
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.