Author: Peter Fendrich
Date: 01:29:54 11/24/98
Go up one level in this thread
On November 23, 1998 at 10:36:21, Luis E. Cabral M. wrote: >Which is the difference between SSDF, ELO and USCF system ratings? >Which is more accuracy.? > >If my rating is around 2175 SSDF what does it mean in ELO of USCF? > >What about BCF rating.? > >Thank you. > >Luis Basically both SSDF and USCF are using the Elo system but... If you are looking for some easy applied formula you will be disappointed. SSDF, ELO and USCF are all using the same algorithm to compute the ratings. That is the rating system worked out by Dr. Arpad E. ELO. The main difference is their pool of players. SSDF is computer games only, USCF is USA players only (I believe) and ELO is top international players only. The overlap in players between these different pools is very small. So you can't really tell the difference. There are lot of opinions however and have one below in this post... :) The USCF and the ELO should be able to 'normalise' by just adding/subtracting a fixed amount of rating points but even that is not so clear because of small differences in applying the same rating system. The SSDF rating is another story because of the pool of chess programs. The profile of one chess program is very similar to the other with a very few exceptions. What you see in the SSDF rating list is very accurate ratings in computer vs. computer play, but we can't be sure of how that would be changed if humans were counted in as well. I don't think that the rating list would change too much but some of the programs would probably gain quite a lot from counting human games so it would change however. How about the level then. The 5 top programs is somewhere between 2500 and 2600 in rating, Would they get that rating against humans? I think so at least in the beginning. Here is what I think would have happened if computer programs competed with humans in real tournaments on real tournament conditions. Each program would start off with quite high performance and get the 2500-2600 rating. After a while, killer lines and killer themes would be generally known among humans and the rating for the programs would drop. But then we would have new releases and the same show would start all over again. Furthermore, humans would better learn how to play against computers in general and that would also make the computer ratings in general to go down a bit. Say 50-100 points. How do I know this? I don't ... :) //Peter
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.