Latest cliffsnotes on Absolute soulreading.

[DeeJ]

why has this thread become a stats tutorial? take it elsewhere

[Drag]

flight2q is spot on with the methodology.

IMO it is quite hard to be sure about methodology when we consider individual hands as we need to specify 'suspected cheaters' strategy in advance, and it is hard to do.

I've already suggested a simpler approach which will test the hypothesis of 'cheating' by making fewer assumptions.




Hypothesis: they play fair, but they got lucky.

First step:
We assume that their style: 80+ VPIP is a losing style if they can't see the cards, so we can specify a long term winrate to be something like -20PTBB/100 in NL and -4PTBB/100 in Limit. (IMO it is even lower in reality.)
We also can specify the probability of winning a tournament to be less than that of an average player. Let say 2 times less.

Second step:
We take the hands that they played in NL, Limit and tournaments. Extract SD from these samples, there should be about 1000 hands in NL and limit, so we can have a good estimation for SD.
With the 'real' negative winrate and SD that we have we can easily estimate the probability of a 'fair' win. We just need to take the Gaussian distribution with the figures that we have for winrate and SD.
We would get some small numbers like 10^(-4) for both Limit and No Limit. (May be even less, the more hands we add the lower the probability would be.) We will also have a probability of a fair win of a tournament 1/(2*(Number of participants)). (With the assumptions that with such a loose style; if they play fair, the probability to win is ONLY 2 times lower than for an average player.) Then we will have to multiply these figures to get an estimations that bad players could be so lucky in all the games: about 10^(-9) or 10^(-12).

Result: Probability of simultaneously having such results playing fair is less than 10^(-9).



In this approach the only assumption that we make is that they are losing players, which allowed us to predetermine their 'real' winrate. Obviously, we are not integrating all the evidence in this approach, but we also won't suffer from biases that could appear when we introduce more coomplicated hypothesis.

These approach is equivalent to noticing that someone is constantly winning playing dies, which leads us to the conclusion that there is probably cheating going on. And we don’t need to know how they do it, by using misbalanced dies or by using magnets, or by other complicated techniques.

Notice that with the outlined above approach we can’t say if ‘cheaters’ could see opponents cards or flop, turn and river or they colluded, etc. It doesn’t matter

[knappis]

[ QUOTE ]

knappis, you are much closer to the truth. First, you had some problem entering the numbers into calculator or something. The expression calculates to 1/12,650 [lol, I had problems too]. The odds are much less severe than this, of course. One factor is you need to figure out a fair way to deal with the fact that we don't know if there were other hands where Potripper did vpip while a JJ+ hand was out there. Another factor is that the JJ+ range was chosen on the basis of examining the data. Then there is the data selection bias problem.

Your approach is not perfect, but at least it is trying to compare the probability of events with and without knowledge of hole cards.

[/ QUOTE ]

Thanks. My mistake. It was pretty late and I was tired [img]/images/graemlins/blush.gif[/img].

The correct calculation is as follows:

(4/25)*(3/24)*(2/23)*(1/22) = 1:12650

If we also factor in the likelihood that folding 4 hands of 25 is part of villains distribution of bet/fold patterns we would know the odds of this to happen by chance. The problem is that we don't kow his distribution of bet/fold patterns but I would say the likelihood is probably pretty high, maybe 33%ish. That would give odds in the 50k range indicating that similar "coincidences" probably happens every day in online poker.

[ikestoys]

oh god to the above post...

can we focus on whats really important here? The fact that he played 93% of his hands except when someone else had a big pair is pretty damning, but the smoking gun is the fact he went from 93% of his hand preflop to never calling a river bet while beat...

also, can we stop the half baked attempts at statistical analysis? Half of those above are def wrong and I only took a semester of stats.

[knappis]

[ QUOTE ]
also, can we stop the half baked attempts at statistical analysis? Half of those above are def wrong and I only took a semester of stats.

[/ QUOTE ]

Maybe you should have taken one more semester of statistics. I have been working with quantitative reserach for 7 years now and I have a few papers published in peer-reviewed journals.

[ikestoys]

[ QUOTE ]
[ QUOTE ]
also, can we stop the half baked attempts at statistical analysis? Half of those above are def wrong and I only took a semester of stats.

[/ QUOTE ]

Maybe you should have taken one more semester of statistics. I have been working with quantitative reserach for 7 years now and I have a few papers published in peer-reviewed journals.

[/ QUOTE ]

thats nice, but i bet you blow at poker, because you don't seem to understand the fundamental reasons for our suspicion

[darkcore]

@flight et. al.
why don't you discuss methodology and other probability related issues about the whole thing in a new thread here:

http://forumserver.twoplustwo.com/po...rd=probability

i don't think this is an unimportant topic at all, but it shouldn't be discussed in *that* depth here.

[ChrisV]

Enough goddamn statistics. If you've read through the hand histories and don't think it's cheating, then that's for the same reason that you can't beat 10/20NL, while everyone that can beat it does think this is cheating. Even if you did some whizbang statistical analysis and showed that there's a 2 percent chance doubledrag could do all the things he did basically by chance, you'd then have to explain why several other accounts did the same thing at around the same time; why doubledrag showed up again after the thread was up and did things like call with 4 high on the river, which indicates a random button-mashing retard which contradicts your theories on him being a LAG who got lucky; and why there would be chip dumping going on; and why AP would deny the existence of said chip dumping EVEN THOUGH IT'S AS PLAIN AS THE NOSE ON YOUR [censored] FACE.

[ikestoys]

[ QUOTE ]
Enough goddamn statistics. If you've read through the hand histories and don't think it's cheating, then that's for the same reason that you can't beat 10/20NL, while everyone that can beat it does think this is cheating. Even if you did some whizbang statistical analysis and showed that there's a 2 percent chance doubledrag could do all the things he did basically by chance, you'd then have to explain why several other accounts did the same thing at around the same time; why doubledrag showed up again after the thread was up and did things like call with 4 high on the river, which indicates a random button-mashing retard which contradicts your theories on him being a LAG who got lucky; and why there would be chip dumping going on; and why AP would deny the existence of said chip dumping EVEN THOUGH IT'S AS PLAIN AS THE NOSE ON YOUR [censored] FACE.

[/ QUOTE ]

TY, this is the rant in my head that i couldn't type

[olieovie]

[ QUOTE ]
Enough goddamn statistics. If you've read through the hand histories and don't think it's cheating, then that's for the same reason that you can't beat 10/20NL, while everyone that can beat it does think this is cheating. Even if you did some whizbang statistical analysis and showed that there's a 2 percent chance doubledrag could do all the things he did basically by chance, you'd then have to explain why several other accounts did the same thing at around the same time; why doubledrag showed up again after the thread was up and did things like call with 4 high on the river, which indicates a random button-mashing retard which contradicts your theories on him being a LAG who got lucky; and why there would be chip dumping going on; and why AP would deny the existence of said chip dumping EVEN THOUGH IT'S AS PLAIN AS THE NOSE ON YOUR [censored] FACE.

[/ QUOTE ]

Great post. Now, back to the important thing, which is how to put pressure on AP to address this issue.