Author: Kirill Kryukov
Date: 03:20:16 11/13/05
Go up one level in this thread
On November 12, 2005 at 16:16:55, Uri Blass wrote: Hi Uri, >As far as I see rating and number of games were slightly changed relative to >previous list. > >Maybe you only fixed small errors and did not add games but it is a fact that >the list was slightly changed. > >Changes in the top 4: > >Deep Shredder9 went up by 2 points with 2 less games relative to previous list. >Fruit2.2.1 went down by 2 points with 1 less games relative to previous list. >Fruit2.2 went down by 4 points with 4 more games relative to previous list. >Fritz9 went up by 2 points with 5 more games relative to previous list. I'll try to summarize all differences (known to me) between Heinz' list and my list: 1. Heinz sorts the list by ELOstat rating, I sort the list by Bayeselo rating. 2. ELOstat ratings in my list are produced by different version of ELOstat from that used by Heinz. I use ELOstat 1.3, he uses ELOstat 1.1, AFAIK. (If I'm not correct perhaps he'll correct me). 3. I shift the ELOstat ratings to be in same range with Bayeselo ratings. This is how I do it: I find average of all Bayeselo ratings, also average of all ELOstat ratings, and then I shift all ELOstat ratings by the difference between the average values. 4. This shifting occurs _before_ adapting the ratings to Shredder 9 with 2750, when the adaption takes place, only Bayeselo is adapted to 2750 for Shredder 9, then ELOstat and Bayeselo ratings are shifted by the same value. 5. There are some minor differences in our game collections. Heinz' database includes a few losses on time. They are also in my database, but I don't use them for ratings. 6. Also it's possible that my database contains a few duplicates, this was claimed by Heinz, but I still did not develop an automatic way to remove them, because of lack of time. If you have any other questions, and time to dig, you can always download Heinz' database and my database and see if you'll find some more differences. I will be curious about your findings too. I don't know another so thorough person like you, Uri. Best, Kirill
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.