Computer Chess Club Archives




Subject: Re: ELO performance?

Author: Peter Fendrich

Date: 15:35:13 05/23/99

Go up one level in this thread

On May 22, 1999 at 21:27:34, Robert Hyatt wrote:

>On May 22, 1999 at 18:11:52, Peter Fendrich wrote:
>>On May 22, 1999 at 15:25:52, Robert Hyatt wrote:
>>>On May 22, 1999 at 12:53:55, Stefan Meyer-Kahlen wrote:
>>>>Just two questions concerning ELO performance and probabilities:
>>>>* How do I calculate my ELO performance if I get x percent out of y games
>>>>against an opponent with z ELO points?
>>>>* In which range is my own ELO rating with 95% probability after that match?
>>>>I guess those are FAQs with 95% probability :-)
>>>>   Stefan
>>>Performance rating is trivial.  compute the following:
>>>number_wins * (opponent's rating +400)
>>This is not the *ELO* performance rating but probably one of all the national
>>variation that there is...  USCF?
>>>number_draws * opponent's rating
>>>number_losses * (opponent's rating - 400)
>>>add those up, divide by number of games.
>>>there is no "95%" probability for a performance rating...
>That is the only "performance rating" formula I have ever seen used.  There are
>alternative ways to represent it... ie
>TPR = opponents_rating + (400*(W-L))/N
>and so forth.. but they are all the same.  If you know of another one, please
>post it...  (btw "elo performance rating" doesn't (to me) imply that this is
>in Elo's book, just that it is an estimated "Elo rating" based on nothing but
>your opponent's rating, regardless of what yours is to start with).

That's true and the elo performance rating formula below is just that.
What you are proposing is a linear approximation of the Performance Rating.
It is not USCF as I thought. In fact it is also mentioned by Dr. Elo.
The purpose with this, from what I can understand, is to easyly calculate the
performance rating by hand.
The original performance rating is:
Rc + D(P)
where Rc is the opponents average rating. D(P) is the expected rating difference
with the current win percentage P. D(P) is based on the Bell curve.


This page took 0.04 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.