Author: Peter Fendrich
Date: 02:08:41 08/03/98
Go up one level in this thread
On August 02, 1998 at 16:09:33, Don Dailey wrote: >On August 02, 1998 at 15:36:35, Robert Hyatt wrote: > >>On August 02, 1998 at 14:46:34, Don Dailey wrote: >> >>>On August 02, 1998 at 13:00:56, Robert Hyatt wrote: >>> >>>>On August 02, 1998 at 09:37:24, Don Dailey wrote: >>>> >>>>>On August 02, 1998 at 08:01:28, Robert Hyatt wrote: >>>>> >>>>>>On July 31, 1998 at 20:51:56, Bruce Moreland wrote: >>>>>> >>>>>>> >>>>>>>On July 31, 1998 at 14:00:04, Peter Fendrich wrote: >>>>>>> >>>>>>>>The ratings on ICC and FICS are really swinging. IMHO they are using a >>>>>>>>completely wrong approach to handle this. The ELO formula is not at all suited >>>>>>>>for the kind of events taking place on these real time severs. The original ELO >>>>>>>>formula is using a kind of constant a 'dampening' factor for varitaions during >>>>>>>>time. At least on ICC they have tried to use another, more dynamic but similar >>>>>>>>method to handle this problem. There are new, much better, ways to deal with >>>>>>>>this. Nowerdays used within some of the space, aero and automotive applications. >>>>>>> >>>>>>>Both ICC and FICS tried to use the Glicko system. >>>>>>> >>>>>>>On ICC, they made it an "extra" rating system, but I don't think anyone paid >>>>>>>much attention to this, so it is removed from the "finger" notes. It is still >>>>>>>there, but to see it you have to use "yfinger". >>>>>>> >>>>>>>On FICS, they made this the only rating system. I am not there much, but I >>>>>>>remember seeing a lot of posts about this, people were upset about this system >>>>>>>because if you played a lot, your rating tended to stay constant no matter how >>>>>>>you did, and people wanted to have more movement in their ratings. So I believe >>>>>>>that on FICS they patched Glicko somehow, so that ratings would still move a >>>>>>>bit. >>>>>>> >>>>>>>bruce >>>>>> >>>>>> >>>>>>the problem with "Elo" is that the "K" factor was statistically derived from >>>>>>the typical number of rated games a person would reasonably play in a year, and >>>>>>the max expected rating change of a person over that time frame. It is totally >>>>>>wrong for a server where (say) a program plays 20,000 games per year. Because >>>>>>we see the huge swings that result from this. Chances are that if you are a >>>>>>2,000 player today, you will be a 2,000 player in 6 months, regardless of how >>>>>>many games you play, So it would be difficult to pick a formula that is fair >>>>>>to those playing a dozen games a year and to those playing thousands. >>>>> >>>>>I wonder why they don't give people the option to use a smaller >>>>>K factor? >>>>> >>>>>- Don >>>> >>>> >>>>You can't do that.. It would grossly distort ratings... someone plays a group >>>>of opponents and decides which he can beat regularly, and then adjusts K to >>>>maximize his rating increase... expect to lose? small K. there's already >>>>plenty of rating abuse there. :) >>> >>>I don't recommend that people have the option to change it whenever >>>they feel like it, perhaps they simply are allowed an initial choice >>>and given one chance to change their minds later. Or maybe they >>>are allowed to change after 6 months with a given one. >>> >>>But a better idea might be to start with a high K and have it adjust >>>dynamically depending on how much you play. It shouldn't get >>>ridiculously high or low but should be limted in either direction to >>>reasonble numbers. I also advocate that if you play someone who is >>>not well established, your K factor should drop too for that one >>>game, since the results should not be weighted too heavily. >>> >>>I know these things can get tricky so I don't know if this is a >>>reasonable suggestion or not (not that anyone is listening.) >>> >>>- Don >> >>At one point in time, FICS was doing something like this... the "K" factor >>is reduced by playing games, and increased by passing time. If you play >>frequently, your K stays low because it is very doubtful that someone's rating >>would fluctuate very much over the span of a few hours or a few days. If you >>don't play frequently, your K stays "up" since it is possible tht your rating >>could change over a period of weeks or months... >> >>It seemed to be quite good (to me) but a human would win 5 or 6 games in a >>row against someone and complain when their rating did not jump like it would >>in a USCF event. The problem is everyone wants + changes, but no one wants big >>- changes. :) >> >>Bob > >Ok, > >One other suggestion. Post 2 ratings, one with high K and another with >low K. Better yet, use only low K and publish a performance rating >based on the last 20 games and call it "recent performance" or something >like this. Then players can see good results get immediately rewarded! > >- Don
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.