[Lyrrad]

[ QUOTE ]
[ QUOTE ]
My suggestion would be to emphasize the number of correct river decisions in a row in a limit game. For example, say there is a 75% chance that a player is ahead if you're betting on the river. Then, if he makes, for example, 50 out of 55 correct decisions in a row, then he's several standard deviations away from the norm.

[/ QUOTE ]

I just tried your idea. Since I couldn't come up with a way to calculate the standard deviaion I used a bootstrap technique drawing 10.000 samples. The mean is ofc 41.25 and the standard deviation was estimated at 3.22.

The odds of 50 out of 55 correct descisions was estimated at 288 to 1

[ QUOTE ]
The evidence should be overwhelming in favour of this cheating

[/ QUOTE ].

These kind of assumptions are very common in this thread. I guess to few people bother to check the validity of these claims. The evidence is far from overwhelming.

However, I do think the model is too simple. All descisions don't have the same probability of beeing correct (how hard is it when you have the nut flush? or 23o that missed?). One thing to keep in mind also is that if this guy wasn't cheating he was running hot and part of running hot is having easy decisions.

[/ QUOTE ]

Okay, so we have a baseline. Can we find out how many hands of data that we can extrapolate from? For example, if he played 550 hands in a row like this (making 500 correct river decisions, with 75% chance), would the odds against that be about 3925770232266214525108224 to 1?

It's still a bit confusing over how many hands we have data showing that he almost always made the correct river decision and how often it was made.