Hello! Out of curiosity: How would proving a rigged RNG mathematically work?Would there be any simple ways to prove that a RNG isn't random? Would we have to know all holecards or for example, would community cards be enough?

Without getting technical, there are statistical tests for randomness. Use Google and search for Tests for Randomness. What is looked at depends on what particular type of rigging is being checked.

In this forum, I have several times answered a question similar to the following:

“Hey, in the past several days I had a higher pair X times and lost to a lower pair Y times. How can that be without some type of rigging?

In every case, either the sample was too small to do any reasonable test, or the probability of that happening was greater than the critical test for significance value, say 5%, or the poster interpreted the data incorrectly.

Check the “The Great Poker is Rigged” thread in the Internet forum. I understand that no one has ever presented convincing proof of a bad RNG while checks that have been done by credible sources found no indication of rigging.

Cheating such as a superuser is a different issue but that is not what is asked here.

Form a hypothesis, measure a sample of hands of predetermined size, or get someone else’s hand database. Test your hypothesis on the sample.

For example, maybe you feel that you don’t get AA dealt as often as you should. You play 10,000 new hands and should expect AA about 45 times, but only get it 30 times. The probability of this occurring by chance is only 1.04%. This isn’t enough to constitute proof, but would be enough to merit further investigation.

Thanks for the feedback! Assuming the manipulation works in a way that starting hands and community cards are dealt randomly but the timing isnt random. For example:

Player A gets his AA dealt as often as he should and he flops a set as often as he should but when that happens he loses to a flush more often than Player B does when hes dealt AA and flops a set against player A. Assuming Player A is better than player B, he would lose more money in this spot than player B does and therefore his edge becomes smaller. Would that make sense?

Would it be possible to create such a scenario while community cards and starting hands are dealt at normal frequencies?

If that would be possible: How would it be possible to figure out if we dont know our opponents hole-cards all the time?

It is possible to favor one player over another while keeping a perfectly "fair" frequencies of hands for each player, in the sense that each player gets the number of AA, flushes, sets etc. that they are supposed to. However I don't think the distribution would be randomly uniform, but I'm not sure how hard it would be to detect a slightly non-uniform distribution of hand types. In order to detect what you are looking for you would want to be able to see all hole cards and see if certain matchups occur at a frequency they aren't supposed to.

Without having opponents' hole card knowledge, you could compare their fold/call/raise frequencies and see if there is any correlation with you own hand strength that can not be accounted for by considering blockers.

You've done your tests. AA shows up with the right frequency both for individuals and collectively. AA also flops a set as often as it should. But, you have a feeling that in particular situations it loses to a flush more often than it should. Do you think you could set up a new test to see if that is true?

<3

Or you could do some bad science, and check the same data for this hypothesis, and get a result of statistical significance.

This is part of the challenge with testing RNGs. If you test for 20 biases, you have a good chance of finding one or more significant bias.

thanks guys! I guess without handhistories it becomes impossible to do any research. I would have to count these scenarios myself and I would probably have to play forever to get a significant sample :/


Edit: Assuming the manipulation is poorly done and community cards arent dealt at random frequencies. Would it make sense to pick out for example K-high flops and count all these and see how often an ace hits on turn / river runouts and then see if it deviates? If so, how many occurences would make it statistically significant?

That would be entirely useless

Care to explain why?

Such a study was done for Pokerstars many years ago and it was random. When those cards come is another issue. The studies also look for repetitions and patterns as far as I know. In case you are aware that the RNG has generally been investigated by 3rd parties; another thing then is what the site runs after that.