Aduke.1929
net.chess
utcsrgv!utzoo!decvax!duke!bcw
Sun Mar 28 14:16:13 1982
Re: Elo rating system
It's quite true that there are strange things which can happen with
the provisional rating formula. The problem is that before someone
has built up a sufficient history, there isn't much that can be done
with improving the accuracy of provisional ratings.
The provisional rating hasn't been the only problem; recently the
USCF changed from an approximation of what was believed to be the
"true" winning probabilities to the exact formula (it is believed,
based on empirical and some theoretical considerations, that the
probability of winning can be expressed as an S-curve centered about
the difference in ratings). The approximation used flat lines at
the asymptotes and a linear approximation in between (making it a sort
of Z-curve). This lead to peculiarities in the range of 250-350 rating
points difference between the players and the USCF changed it to the
true S curve about two years ago.
There's a fairly extensive literature on rating systems but as far as
I know the Elo system is considered to be the best that anybody has
been able to come up with to date.
Bruce C. Wright @ Duke University
-----------------------------------------------------------------
gopher://quux.org/ conversion by John Goerzen
of http://communication.ucsd.edu/A-News/
This Usenet Oldnews Archive
article may be copied and distributed freely, provided:
1. There is no money collected for the text(s) of the articles.
2. The following notice remains appended to each copy:
The Usenet Oldnews Archive: Compilation Copyright (C) 1981, 1996
Bruce Jones, Henry Spencer, David Wiseman.