Quote:
Originally Posted by Wiki
Certainly very odd behaviour but I can't see why you assume bots. If they were all running the same bot I'd have expected a little more consistency over their results.
Definitely not. Variance is a much bigger deal than a lot of people think.
There was some graph that I remember seeing that showed a bunch of computer simulations for some typical player with the exact same long-term winrate. Something like 1BB/100 and SD of 18. Anyway, they did a bunch of different shorter term samples for this player. But by "shorter term" samples" I mean it was a bunch of 100k hand samples or something like that (a relatively short term as far as this stuff is concerned). And the graphs and win-rates were all over the place. +2BB/100 for one stretch, then -0.6BB/100 for the next stretch, etc. For the exact same computer simulation with the exact same SD.
I don't know if anyone knows the graph I'm talking about but if you know the thread I would be interested in seeing it again (PSMichael might remember this one)
Anyway the point is that trying to determine that the players are different people based on slight differences in their winrates just isn't going to work.
Last edited by MicroBob; 07-13-2010 at 07:15 AM.