TalkChess.com

[H.G.Muller]

Yes, the statistics is a bit meager, but for 200 games the standard error should be about 2.8%. But I had expected the advatage to increase to 60+%, and this seems to be ruled out with high confidence even with the currnt number of games. But I certainly don't have enough sttistics to decide which of the methods is better. Problem is that the 5 min/game tests take a lot of time. So I am using kind of marginal runs to gat a vague idea of what might work better, and then will concentate on getting better statistics for those. That gives better chances for having something that works available quickly.