Quote:
Originally Posted by Thinking Out Loud
To reference your exact quote, how would individual HHs show this, what is the formula or method by which HHs can show this and it be proof? Just repeating the claim "HHs will catch it" without proof is no different than saying 'oh clever software would be able to hide this signal in the random noise of the deal'
As MJ said, every suspected rig can be checked with hand histories. Computers can count how many aces are "supposed" to hit the board, or compare how often a flush draw hits.
There is a logical expectation for board runouts in the long run, just as there is a logical expectation for flipping coins in the long run.
i.e. If you suspected a coin was loaded, you could flip it 1 million times, and see if it landed heads up more than 1 standard deviation above 50% of the time. Common sense would say that if the coin came up heads
significantly more often than half the time, it's not a fair coin.
Poker is the same. If flush draws are hitting more often than expected in the long run, the deck might not be fair.
To read more on the expectation in poker hands, see
http://www.spadebidder.com/
The guy that ran that site tested millions of hand histories to see if the results (of various things like "number of paired boards" or "monotone flops") matched expectation over a large sample size.
Quote:
Originally Posted by Thinking Out Loud
OK, how many hands are needed to be statistically relevant and what range of standard deviation from the mean is acceptable?
According to an average rigtard, one or two hands is enough!
To people with more common sense and logic, it depends on which "suspicion" you're testing, but a million hands and a couple of standard deviations seems like a decent sample size for me.
Happily, the
http://www.ispokerrigged.com/Is%20Po...%20Rigged.html website has checked rather more than 1 million hand histories from various sites. In tests to see if there were more "bad beats" than expected, Pokerstars passed.
It's somewhat interesting that in the sample HHs tested, pre-flop underdogs won slightly less than expected, but underdogs on the turn won slightly
more than expected. This is most likely due to variance (it turns out that 1 million hands isn't really a massive sample in a game with such small edges), and I suspect that running the tests on another 1 million hands would have similar results within the expected standard deviations.