From: Larry Kaufman WIZARD NET> Date: 7 sep 1999 Subject: Re: Unified ratings -----Original Message----- From: Pieter Stouten INAME COM> To: SHOGI-L TECHUNIX TECHNION AC IL TECHUNIX TECHNION AC IL> Date: Tuesday, September 07, 1999 6:11 PM Subject: Re: Unified ratings >On 99/09/07 at 16:30 -0400, Larry Kaufman wrote: > >>I have updated the USSF ratings for the four American players who >>participated in the recent events in Europe (Pan-Atlantic and European >>Open shogi tournaments). I used European Elo ratings for those opponents >>who had them, and approximate performance ratings for those who didn't. >>The results: I gained 19 points to 2348, George Fernandez lost 12 points >>to 2067, Fassil Bekele gained 29 points to 1991, and Ray Kaufman gained >>15 points to 1741. >> >Did you apply the (reversed) correction formula for European ratings >below 1900 to Ray's rating prior to calculating his gain? No, but I did count the unrated players at their approximate performance ratings, which were generally lower than the nominally assigned ratings at the tournament. I did not use the proposed corrections because at this point they are still just proposed. It is now my opinion that the differences between the two systems are much less, even at the low end, than I thought a year ago, and so I would now advocate a milder correction to the low end, perhaps only below 1500 or so rather than 1900, together with a mild upward correction to the low end of the US rating scale. Although the data suggests that even at the top the US ratings are a bit low relative to Europe, this may be due to the fact that only four players were tested, and there may be special factors at work. For example, Ray's US rating has not yet reflected the improvement resulting from playing about 200 games this summer in Japan just before the European events, and my own US rating may be artificially depressed by the fact that a substantial percentage of my rated games are with my most difficult opponent, Mr. Jiro Yoshinari. If I never played him, I would probably be over 2400. Regards, Larry Kaufman