[threeonefour]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]


the computer and the human team plays each set of cards against the same set of cards. So one side cant get a good or bad run of cards.

[/ QUOTE ]

Yes, but for instance, Phil Laak suddenly decided on Hand 250 that he was going to be aggressive for awhile. Presumably Ali is doing his own thing. What if Phil's side gets a good or bad run of cards at that very moment?

Since they have studied this, presumably U. of A. has some reasoning as to why this will not affect things.

[/ QUOTE ]

obviously it will effect things, but the field of statistics is a very robust science in general, you can often violate your own assumptions and still come to valid conclusions. they will almost certainly be able to conclude quite a few things from this sample.

[/ QUOTE ]

I'm just curious as to the quick explanation of the analysis. I do believe them, but I'm curious too. The reason why this is mentioned is that the Limit Hold'em community here has applied enough statistical analysis to have a feel for things. And the conclusion is, that if you are only on one side of things, that it takes hundreds of thousands of hands, and some would say millions, before things are even the slightest bit meaningful. This is mentioned in Morgan Kan's thesis, I got that far.

Also, in a hand with 2 people, some players will take random shots on random hands. One poker intuition would be that the luck of the draw on these hands would affect things a great deal and 2000 hands, which maybe consists of 50 of these "shots", where hopefully the opponent has nothing, isn't enough to eliminate luck.

Skimming the thesis, it looks like he does things like plot various strategies on a large number of hands to get a sense of the statistics. Then they have some empirical results. It would be nice if somebody can summarize all that, for folks like me who aren't smart enough to understand. [img]/images/graemlins/grin.gif[/img]

[/ QUOTE ]

that hundreds of thousands of hands bit is very unrealistic. if you want to prove your winrate within a .1 of a bb/100 then yes.

but if you want to simply prove you are a winning player, it takes much less, which is what they will be doing for polaris they will be testing if his winrate is >0 not whether it is the .5bb/100 to 2bb/100 range, which is a different test that would require relatively more evidence to reject the null. and if you want to determine something like your VPIP %, that converges very quickly.

of course once you have the numbers they only tell you about your performance given those conditions. it obviously does not prove you are a winning play at a different site, or a different limit, or whether you will be a winning player in a couple of years when the games get harder.