NL Bots on Full Tilt

[neverforgetlol]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
If someone who actually understands statistics could actually explain the important of their stats converging and explain why players using a similiar system or working together would be unable to reach this level of convergence over the sample we have we can just end this thread, and make a new one about hating bots right? What am I missing?

[/ QUOTE ]

That would be nice to see. Unfortunately, all the people using stats so far (myself included) don't seem to have enough experience to be considered trustworthy experts.

[/ QUOTE ]

I'm mostly an expert, I have a masters' in stats.

What I can say, at least, is that if you consider testing equivalance of proportions, is that n is large which means you are more likely to reject Ho: p1=p2, etc. If you did not reject, say at the .01 level, that would be pretty damning evidence that these were bots. You would have a very powerful test meaning that the probability of Type II error (not rejecting Ho, when Ho is false) would be very low. This quantity however is difficult to calculate precisely.

[/ QUOTE ]

Hey nflol, would you mind putting that into laymans terms for people like me?

[/ QUOTE ]

I guess. You are interested in comparing two proportions, for example vpip for two of the suspected bots. Since we have thousands of hands to base these proportions on, any difference in them is very unlikely to be due to measurement error. So if we do not reject the null hypothesis that states that the two proportions are equal, that will be pretty strong evidence in favor of them being bots (equivalence means bots), since the true difference between p1 and p2 will be insignificant.

So if someone could repost those stats I can do some tests.

[/ QUOTE ]
If those were the Layman's Terms, I'm an idiot.

[/ QUOTE ]

You need some stats background, it's hard to explain hypothesis testing from the ground up in a few paragraphs.