Author: James T. Walker
Date: 12:47:02 05/05/99
Go up one level in this thread
On May 05, 1999 at 11:14:00, Mark wrote: >On May 05, 1999 at 06:42:18, blass uri wrote: > >> >>On May 05, 1999 at 04:34:41, Harald Faber wrote: >> >>>On May 05, 1999 at 03:10:10, blass uri wrote: >>> >>>>>>>The Week in Chess gives what looks to be a mistake in Polgar's TPR for the Fritz >>>>>>>match. Does anyone know the actual calculations used to determine TPR's??? >>>>>>> >>>>>>> >>>>>>>Fritz 5.32 ---- = = = 1 1 1 0 1 5.5 **2818** >>>>>>> >>>>>>>Polgar, Judit g HUN 2677 = = = 0 0 0 1 0 2.5 **1859** >>>>>>>(?) >>>>>> >>>>>>As we can count Fritz made a +141 elo performance ... >>>>>>we can deduce that J polgar made a -141 elo performance ... >>>>>>So Fritz have a 'default' rating of 2000 (Obviously the minimum for the >>>>>>performance calculating program) >>>>> >>>>>So there must be s.th. wrong because Polgar had a rating of 2677 (see above) >>>>>which would give 2677-141=2536TPR and Fritz 2000+141=2141TPR >>>> >>>>Fritz5.32's performance is 2677+141=2818 >>>>Judit polgar's performance is 2000-141=1859 >>> >>>Can you tell me where you get the 2000 for Polgar from? It is neither default >>>nor written above. Above it is 2677. >> >>The 2000 is for Fritz5 and not for polgar. >> >>Polgar's rating is not important when we compute polgar's performance. >>The only important thing is that polgar lost to unrated player 5.5:2.5 >> >>Everyone who lose against stupid 2000 player 5.5:2.5 has performance >>2000-141=1859 >> >>everything is right except the assumption that Fritz has 2000 >> >>Uri > > >All you guys are great, thanks for the information. My final question is where >does the 141 come from? Here in the US you can get a provisional estimated >rating by taking, say, 6 games, adding 400 to your opponents rating if you win, >subtracting 400 if you lose, and no change if you draw. Add those six figures >up and divide by 6 and viola! Your provisional estimated rating. > >Is this general procedure (with possible minor modifications) how the value 141 >is calculated in the above discussion? If it is, then by solving for X (instead >of using 400) I get 376, which is indeed a minor moditication to bring my >calculations in line with the 141 number. > >Thanks for your replies! Can't wait for HIARCS Yermolinsky! > >Mark I hate to tell you guys but you are all wrong. Performance ratings are generallly only given to unrated players. There is no "assumption" of an unrated players rating. The performance rating is found by the formula: Rp=Rc +400(W-L)/N Rp=Performance rating Rc=Average rating of players opponents(2677 in the case of J. Polgar) W=Number of wins(4 Wins) L=Number of Loss (1 Loss) N=Number of games (8 Games) 4-1=3x400=1200/8=150+2677=2827 J. Polgar since she is a rated player would not be given a performance rating. The fact that she performed at (1-4)=-3x400=-1200/8=-150 points below her opponent is about all that can be said.
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.