H0 is that rating is not dependent on the opponent and you can use one opponent to find CEGT 120/40 rating of players.
H1 is that the rating is dependent on the opponent.
My question is if you can reject H0 based on the data.
Here is the CEGT 120/40 rating list
http://www.husvankempen.de/nunn/40_120_ ... liste.html
Note that even with only 50 games performance of the opponents are usually close to their rating
I do not find many cases of difference of more than 100 elo between rating and performance and I can expect a small number of cases when it happen even only by luck assuming that performance of programs is independent on the opponent.
Uri