Author: Uri Blass
Date: 00:50:17 08/26/00
Go up one level in this thread
On August 26, 2000 at 03:31:26, Christophe Theron wrote: >On August 26, 2000 at 00:42:46, Uri Blass wrote: > >>On August 25, 2000 at 21:01:04, Christophe Theron wrote: >> >>>On August 25, 2000 at 17:21:00, Uri Blass wrote: >>> >>>>On August 25, 2000 at 17:11:00, Fernando Villegas wrote: >>>> >>>>>I suppose that Chessbase people cannot feel very good. They putted three horses >>>>>of his corral into the race and the winner was the solitary stead from >>>>>Millenium. I bet somebody there is just now ready for harakiri. >>>>>fernando >>>> >>>>I disagree. >>>> >>>>chessbase did not do a bad result. >>>> >>>>Fritz is number 2 >>>>Junior is number 5 >>>>Nimzo is number 7 >>>> >>>> >>>>All of them finished in the first half of the participants. >>>> >>>>Uri >>> >>> >>>When you enter 3 top programs in a tournament, your chances to get one at the >>>first place is much higher than if you enter only one or two programs. Still, it >>>did not happen. >>> >>>Let's calculate the "team performance" for the following computer chess >>>companies: >>> >>>1) ChessBase >>>2) Millenium >>>3) Shröder BV >>> >>> >>>ChessBase has: >>> Fritz, ranking #2 >>> Junior, ranking #5 >>> Nimzo, ranking #7 >>>Average ChessBase ranking: 4.67 ( that is: (2+5+7)/3 ) >>> >>>Millenium has: >>> Shredder, ranking #1 >>>Average Millenium ranking: 1.00 >>> >>>Shröder BV has: >>> Rebel, ranking #3 >>> Chess Tiger, ranking #4 >>>Average Shröder BV ranking: 3.50 >>> >>> >>>Standings of the teams: >>> >>> 1. Millenium 1.00 >>> 2. Shröder BV 3.50 >>> 3. ChessBase 4.67 >>> >>> >>> >>>The reason for taking the average of the rankings, and not the highest program >>>of each team, is that if you enter several programs in the hope that one wins >>>the tournament, then you have to accept that the result of all your programs are >>>taken into account for your global performance. >> >>I think that it is better to use average rating and not average ranking. >> >>You need to calculate rating based on the tournament(assuming the results repeat >>again and again. >> >>Rebel and tiger are not going to enjoy from it because they won 2 games against >>the weakest programs that everybody wins. > > > >I have never seen a tournament sorted by tournament performance ratings. > >If it was better to use average rating, tournaments would be sorted by ratings, >not by score. I think it is a better way to decide about ranking. This method is more complicated and it is hard to know what is the result that you need in the last round and this is the reason that people do not use it but I believe that the method is more fair. By the method that is used programs that were lucky to play against weaker opponents can getbetter ranking. I do not say that tiger and Rebel deserved worse ranking but it seems clear to me that Nimzo deserved better ranking than Sos because Nimzo did not play against the 2 weakest programs when Sos played against them. > > > > >>Fritz did not play against them. >>Nimzo did not play against them. >>Junior played only against 1 of them. >> >>I do not say that rebel's rating or tiger's rating will be worse than Junior or >>Nimzo. >> >>Rebel and tiger had 4 out of 7 if you do not include the 2 last programs. >>Junior has 4.5 out of 8 and Nimzo has 5 out of 9. >> >>4/7 is bigger than 4.5/8 or 5/9 but the difference is not big and it is logical >>to think that the average rating of Fritz Junior and nimzo is better. >> >>Their scores is 16 out of 26 without games against weak programs. >>tiger and Rebel score is 8 out of 14 (if you do not include the last 2 players) >> >>16 out of 26 is better than 8 out of 14. > > > >By selecting a given number amongst the games played against weaker opponents >you can get exactly the rankings that you want... > >Maybe the top programs were good enough to get the point anyway against the 4 >weakest programs, or maybe against the 6 weakest, who knows... There is a reason that I chose to ignore the 2 weakest programs. They lost against all the rest. You cannot find another team of programs that lost against all the rest. Uri
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.