Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Past and Present Re: Rough approximation Re: ELO Calculations

Author: Dieter Buerssner

Date: 11:38:29 04/28/05

Go up one level in this thread


On April 27, 2005 at 18:40:23, Eelco de Groot wrote:

>I sure wish I could follow all your C code, but I never did Study C seriously.

The code should be understandable, even for an "Excel is fine, I don't now much
more"-user. The formulas I used are given in comments (and in the post of Odd
Gunnar for the Schmitt formula). How I exactly implemented the Schmitt formula
does not really matter. If you are curious, I used Horner's scheme, and some
obfuscation may be the "return sign ? 0.5-p : 0.5+p", which is just a shorthand
in C for saying

  if the Elo difference was negative, the result is 0.5-p; otherwise it is
0.5+p.

p was exactly calculated with the formula given in Odd Gunnar's post (and
technically avoided some multiplications and all those absolute value terms).

>Are there any good compilers that are not expensive and work under Windows?

There are many free choices available. One can be found at www.mingw.org. This
particular choice has the advantage, that the download size is (almost)
reasonable, even for a modem. Free command line compiler can be found at
www.microsoft.com; also a beta version of their soon to be released new
compiler. If you are really interested, I can find links.



>I think I can guess where FIDE got its numbers, I suspect they are mainly
>interpolations of the original table in Arpad E. Elo's 'The Rating of
>Chessplayers, Past & Present' (Batsford Books, 1978). Prof. Elo gave the
>following table:

Wow! Did you really type this in? I checked a few data points - exactly equal to
the data on fide.com (link is in a comment of the source code, I posted
yesterday).

But now, my mystery is: How did Elo calculate those numbers? (My program
calculates very similar numbers, but they differ significantly in the last
decimal place given). Of course, this all is not too interesting - all numbers
are about right, and the amount they differ, will not really matter.

Cheers,
Dieter



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.