Rigging question

[guesswest]

Question for the math people, because I'm not one.

I don't believe for a second that online poker is rigged. But when this subject comes up, as it invariably does, I always find myself saying something like 'the deal and hand histories are a matter of public record, if deals are rigged there should be statistical evidence to show it'. Yet I've never actually seen a study on this - I feel sure that with the amount of money in online poker someone somewhere must be crunching numbers.

Basically my question is twofold. Firstly, how easy is it to prove that a deal is fair? I'm talking about all kinds of supposed rigging - action hands, house bots sucking out etc. Our information is incomplete since we only see our own cards and showdowns - but I assume there has to be some way of making that information statistically significant, right? And the second part, does anyone know of any studies on this that I can link people to when this subject pops up, as it does again and again? Thanks.

[FortunaMaximus]

A preflop deal? You wouldn't need a large sample size. Several tens of thousands, presumably. Enough to account for standard deviation.

Board deals in HE? Well. Same principle as above applies, only you'll need a sample size in the millions probably.

Is this truly a subject of interest on an intellectual reasoning basis? Because THAT has merits.

[guesswest]

As I said, I'm talking about all kinds of supposed rigging. A preflop deal is obviously quite easy to get stats on, I assume it gets much more complicated factoring board deals and ideas like 'action hands', house bots with advance knowledge etc - I don't know how much more complicated, that's why I'm asking.

I have plenty arguments as to why online poker wouldn't be rigged, in terms of motivations for doing so and how difficult it would be to keep secret. I have none to prove that it in fact isn't. I'm wondering if those proofs are out there. I don't have an agenda here, I'm asking because I don't know the answer - I'm not a statistician.

[FortunaMaximus]

The key is to remember, using 52 cards, there are a finite number of eventualities.

[ QUOTE ]
I assume it gets much more complicated factoring board deals and ideas like 'action hands', house bots with advance knowledge etc - I don't know how much more complicated, that's why I'm asking.

[/ QUOTE ]

It wouldn't be more complicated to arrive at a result, you would just have to define a different set of criteria to filter out the results, so it is a more complex method to arrive at those conclusions.

[ QUOTE ]
I have plenty arguments as to why online poker wouldn't be rigged, in terms of motivations for doing so. I have none to prove that it isn't.

[/ QUOTE ]

Of course. And the wider the market base, the more need there is for the provider to hold its processes above and beyond possible criticism, for its revenue comes from what it takes for providing the service, not funnelling relatively small sums of revenue towards specific individuals, even if those individuals are allied with the organization. If only because that would end up costing them more in the long run both in monetary costs and threats to reputation and possible devastating declines to its customer base.

[ QUOTE ]
I don't have an agenda here, I'm asking simply because I don't know - I'm not a statistician.

[/ QUOTE ]

Yes, I assumed that. No worries where I'm concerned. /images/graemlins/tongue.gif

[guesswest]

Bump...anybody know?

[Darryl_P]

To the second question, all I can say is that I've never seen any reliable studies on the cheating issue. Then again, I've never looked too hard for any.

To the first question, it's fairly easy if you have access to the raw data in a good manipulatable format. It's a pain if you have to input everything manually.

All you'd have to do is focus on one thing at a time, like action flops say. For each possible starting hand, define precisely what it means to "hit the flop". For example, with a 97 off you might define it as getting a made hand of one pair or better, involving at least one of your cards, or an open-ended straight draw. This definition doesn't have to be strategically sound. In other words you don't have to sure that it's correct to play those and fold all others. As long as it loosely satisfies that criterion it's ok. The key is that it be defined precisely so that you can compute a precise theoretical probability of hitting the flop to compare to your observations. If your observations fall outside the range of +/- 2 standard deviations, then you can say with some confidence there is cheating going on. The more hands you test, the more reliable your study will be. But you have to make sure you define the number of hands to be tested in advance to avoid bias.

That's more or less how such a test would be done.

[CORed]

I think it's highly unlikely that online poker is rigged, but there is more than one way to rig it, and at least one way that I can think of that would work with a random deal. In games like holdem and omaha, once you have shuffled the deck, the hole cards of each player and the board are determined. If any player has this information before bets are placed, he has a huge advantage. So, assuming a shuffling algorithm that works like a live card game, rather than just randomly selecting each card as it's dealt, players, be they house bots, humans playing for the house, or friends of rogue programmers or sysadmins who have been given or sold a "back door", would be taking money from legitimate players. In order to catch this, you would have to analyse players' betting patterns as well as cards dealt. If the money from the players with information they shouldn't have is being funneled back to the house, it would show up in a financial audit, but if they are freelancers with a back door, it might not.

Of course, such a scheme would be much harder to work with stud, etc, where the cards a player gets will differ depending on whether and when other players fold. So, if your're really paranoid, play stud.