Author: KarinsDad
Date: 17:28:33 01/29/99
Go up one level in this thread
On January 29, 1999 at 19:09:32, Peter Kappler wrote: >On January 29, 1999 at 18:43:49, KarinsDad wrote: > >>On January 29, 1999 at 18:10:53, Dann Corbit wrote: >> >>>On January 29, 1999 at 17:47:08, James T. Walker wrote: >>> >>>>It seems to me that it's not what Arpad Elo said that matters. It's his formula >>>>that counts. Since it is what determines the ratings it has to be true! Not >>>>that there is no probability of error but that since the formula gives you the >>>>rating based on the WE (Winning Expectancy) then the WE is correct by >>>>definition. Of course because of statistical probability people or computers >>>>will not always perform as per the WE. The point is that people and computers >>>>get their ratings from the same formula. The only thing imperical data will do >>>>is show that there is of course only probability not perfection. Maybe if you >>>>could gather the statistics from millions of games the the actual data and the >>>>probable data would be equal but don't bet the house on it. >>>>Jim Walker >>>His method *is* statistically valid. In fact, it has a sound mathematical >>>basis. Of course, as with any mathematical model applied to concrete things >>>(and especially people) it will not be a perfect predictor. >> >>Not even with an extremely large sample set. >> >>KarinsDad > > >In general, I think the ELO formula does an excellent job. > >I can think of only one or two areas where the current system could be improved. > >The first is to modify the system so it can identify players who are improving >rapidly. The typical case is the young player who is making rapid improvement. > >Back in the 80's the USCF had a scheme that attempted to address this problem. >If a player gained more than 64 rating points in an event, every point past the >64th was doubled. So, if your tournament performance merited an 80 point rating >gain, you actually received 96 points. Likewise, if you were unlucky enough to >lose to somebody who had a fantastic result, you were given a few rating points >back to compensate you for facing an opponent who was clearly underrated. > >I'm not sure why this system isn't in use anymore. > >The other issue is the inactive player who returns to the tournament chess >scense after a long absence. In many cases, this player may be playing at a >strength 200 or more points below his listed rating. > >Except for these 2 cases, I think the current rating system is very accurate. > > >--Peter I agree that for the most part, the ELO rating modification system works well (although it could use some minor improvements). It's the ELO statistcal equation that starts to falter once you get +-300 rating points. There are other weird cases as well. For example, if an unrated player wins all of his games or loses all of his games. What should his rating be? What if he won all of the games against 1200 rated players. He could be 1600 or 2600 calibre. If he lost against a bunch of 2000 players, he could be 1800 or 800. The USCF rating committee is putting out a new set of rules for voting by the delegates (I think this year), but it is 6 postscript pages of mathematics and way too complex. Your average player cannot understand it. You can download it from: http://math.bu.edu/people/mg/ratings.html Good luck understanding it all. KarinsDad :)
This page took 0.01 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.