Official Poker Site Data Analysis and Discussion Thread

[Sodom]

Originally Posted by optimus

It is curious a majority of active posters in these threads (though not necessarily a majority of readers/poker players) appear supremely confident of the randomness of the deal...that when presented with evidence to the contrary, they're quick to dismiss it as ``you must've done wrong then''!.

We have evidence. That's fine. Why not publice them?

In a very fundamental meaning we always have to trust things we are involved in. That is basic for all human interactions. We will never know if there are superusers on a site; we will never know if the RNG is manipulated in a very advanced and subtile way; we will never know if the live dealers we face in a pokerroom collude with some players; and there are probably a lot of other factors we never will know exactly. We just have to trust and we will trust as long as it seems reasonable. It is indeed reasonable to run extremely bad in a poker carrier as well as ridiculously good. The variance in poker can be hard to handle and that is also the reson why very few players will stay as professionals for any longer time.

Poker players are by nature very suspicious because that is part of there job. I am always sceptical to what happen at the tables and I am always assuming something shady going on. But nevertheless I continue to play because of the simple reason that I am winning and have so done on every poker site I have been playing aswell as live. Pokerstars to me does not seems to be more or less rigged than my live games or any other online sites.

[sluggger5x]

[Schwallie]

FYI, I'm starting to wonder myself. I started playing in January of 07, and was at 3/6 and 5/10 by the end of the year. I play like 1.4million hands for the year. They knew I was considering elite this year, and I had a bad run at 5/10 so I cashed out and started back at 2/4.

I am currently down nearly 40k in all-in equity for the year, 100 buyins, yes, 100 ****ing buyins down. I will post the image when pEV comes out, I am using PT3 and only have a simple program that calculates it, so if that program is right and pEV confirms it then I will post this sickening stuff

[Bobo Fett]

Originally Posted by optimus

The burden of proof for the fairness of the shuffle is on the sites. This is my opinion as well as that of other reasonable posters.

Yes, but until we get a site to allow a trusted third-party to scrutinize every aspect of their operation, and release the results in a transparent fashion, what do we do? If people are convinced something is rigged, it only makes sense for them to gather some evidence and discuss it. There's a much greater chance of something being broken open this way than waiting for the sites to prove/disprove things.

Originally Posted by optimus

It is curious a majority of active posters in these threads (though not necessarily a majority of readers/poker players) appear supremely confident of the randomness of the deal...that when presented with evidence to the contrary, they're quick to dismiss it as ``you must've done wrong then''!.

Dismiss it? Dispute it, provide alternative theories, yes...but I haven't seen many people outright dismissing any reasonable evidence. However, if anyone is too quick to dismiss evidence, it likely relates to your next point:

Originally Posted by optimus

The problem of course is there do exist paranoid posters on this topic. They open themselves wide open to ridicule, while cutting away at the credibility of others with worthy observations.

Exactly. This is the reason you'll find me arguing with many of the people that start these threads. The tin foil hat posters aren't helping anyone; they hurt both the credibility of those with worthy observations, as you say, and at the integrity of online poker.

[john.smith]

Originally Posted by Elbow Jobertski

From what I recall is that Mason was speaking of the effect on strategy within a hand. This isn't so much a problem of math as much as that all the human factors in a hand, estimating ranges based on opponent behavior and so on, present a margin of error that will dwarf any difference between a pot equity calculation based on the assumption that the stub is identical in randomness to all unseen cards and one that takes opponent behavior into account.

Who cares if, assuming your ranges are correct, your equity is 50% rather than 49%, as your estimation of opponent behavior is generally not going to be so precise. This is a totally different kettle of fish than using the same calculations to make detailed criticisms of an RNG over a massive amount of trials. The larger source of error that makes the smaller error insignificant is gone.

This also seems to doom any totally precise calculation of expected ev when taking any all in hands in a game with more than two players. To do this one has to make assumptions about what an opponent will or will not fold, and there simply is no way to do this by a universal method. It is still an estimation with flaws as people are different.

This would be all fine and dandy if we assume that every player gets in the same situation the same amount of times. However, the whole point of playing poker is to not let this happen to you.

I imagine some sort of feature that tries to model typical behavior, where the end user can put folding ranges for typical opponents in typical situations will make these sorts of calculations more accurate, but it still is going to suffer from the same problem once the number of trials gets large.

Restrict the data set to only include situations where:
a) there are only two players at the table (e.g. heads-up tables, final stages of a sit & go, etc.)
b) where both players go all-in pre-flop (sit & gos would be ideal for this)

All cards would then be known.

In a large enough sample, the various hands should come close to their expected win rates. I can't see any reason why AA wouldn't win 85% of the time in a large enough sample. In Sit & Gos, it would also be interesting to see how often the bigger stack won vs. their expected win rate.

[redlotus]

Originally Posted by Bobo Fett

Dismiss it? Dispute it, provide alternative theories, yes...but I haven't seen many people outright dismissing any reasonable evidence.

While I can't recall anyone outright dismissing evidence, you can't deny that there are a lot of people who cry "anecdotal" whenever someone tries to present something. Just look at the initial responses to NLL's "Cake Poker SnG" post for evidence of that.

To make matters worse we have this thread where someone tries to put together start a "fact-finding" mission. The first responses are fairly positive, including a member that has the server power and internet connections to help. Unfortunately, a greenie decides to mock the whole idea and things just denegrate from there. How is it possible to get the evidence when we have members that actively stifle any attempt to collect information?

Sorry if this is a little OT, but I'm a little miffed how some posters go out of their way just to argue when it comes to this subject.

[MicroBob]

Originally Posted by Phil153

OK, I've just read MicroBob's objections about the bunching effect as a theoretical flaw in all-in EV calculations. I actually posted about this a long time ago. His objections are theoretically valid.

Like Mason, I don't think it's significant, but the only way to answer this question definitively is to run a simulation with known folded cards and compare to the unknown card calculations. I don't have the time to do this right now but I do have the basic code for it and it won't take terribly long, so I'll do it in the next 3-4 weeks if no one else takes it on. I'll release the results and code open source so that others can look at it.

I haven't seen a lot of huge sample PokerEV graphs. There's fgator's EV graph of 1.5 million hands, which had expected and actual off by something like 5% over a large sample, which is statistically very normal. He's very weak tight and the theoretical bunching effect should be maximized. I've also seen people run hot over hundreds of thousands of hands. So the bunching effect should be less than the order of swings. That's not to say it couldn't be significant - it could - but I don't think the graphs where people run very bad are caused by this. Perhaps contributed to, depending on how large it is, but not the major contributor.

Phil - Thanks a ton for replying and contributing.
I had no idea you had already pontificated on the whole bunching stuff before. I am very glad to hear that. I was going from scratch and really don't know my way in and out of this stuff and am just kind of feeling my way along.

I think your software is pretty darn cool but I know you've heard that before. Definitely interesting to look at in smaller samples. But I do think that it can be dangerous in larger samples because of potential skew for certain players.


1. What Mason was referring to as it being insignificant was regarding calculations for your individual hand as you are playing it. Elbow pointed out that you aren't going to care if you are really 49% instead of 50%.
But over 800k hands this factor will add up.

2. I haven't really looked at fgator's graphs. Only have heard about them. I notice you mentioned fgator's weak-tightness.
That goes for my theory on 1p0kerboy as well. 1p0kerboy is somewhat weak-tightish and plays 9-handed games.
Similar to how you point out that the graphs will skew for a short-stack player I also believe it can tend to skew (although perhaps a bit more slowly) for a weak-tight player especially at full-ring where you say that the theoretical bunching effect could be higher.


Also fwiw I think it's possible that the skew for shortstack players might not just be for multi-player flops. Just a theory but hang with me here.
Some SS'ers perhaps change the dynamic of the game enough with such a massively wider range to potentially exaggerate the bunching effect even on a 6-max table. They will be more likely to get into situations where they are shoving their SB with JT and will be more likely to be called by A7 or something. Getting into more all-in's with wider ranges exaggerates this potential bunching effect. Every player that already folded that DID NOT have AA changes the odds for each of those A7 type hands to suck out.
you don't get nearly as many wider-range A7 preflop all-ins otherwise. Frequently the ONLY time you will see them is when one of the participants has a shortstack.

[Bobo Fett]

For once I think we're actually having a somewhat productive discussion here. I'd love to see a thread dedicated to such a discussion; I'd like to take some of the recent posts from this thread and get one going. I'm going to PM Mike Haven about the possibilities.

[MicroBob]

Originally Posted by allurit

lol if that truly does skew the results, that forever needs to be referred to as The MicroBob Effect imo

allurit - Considering that the designer of pokerEV himself has already written on the potential for weak-tight full-ring players to have this bunching effect I don't think it should be named after me.
I just brought it up because it made sense to me just as one possible reason (and there very well could be others but I don't know) and nobody here knew that this was not the first time this idea had been raised.

[MicroBob]

elbow - Thanks for contributing.
I knew somebody out there who could explain it more concisely (more meat...less rambling) would understand what I was talking about. Hopefully your posts will have helped some of the posters here who are having a tougher time following my long-winded train-of-thought type logic.

[MicroBob]

Originally Posted by Bobo Fett

For once I think we're actually having a somewhat productive discussion here. I'd love to see a thread dedicated to such a discussion; I'd like to take some of the recent posts from this thread and get one going. I'm going to PM Mike Haven about the possibilities.

Don't forget to include tautometer's posts!


I don't know about a new thread for this. But there is probably enough meat in here just on pokerEV graphs and bunching effect stuff that may be of interest to some people (and worthwhile knowledge perhaps) that is not going to be seen by many user who don't bother to look in your typical thread with subject line "lets prove pokerstars is rigged".

[Tachyon]

Originally Posted by john.smith

Restrict the data set to only include situations where:
a) there are only two players at the table (e.g. heads-up tables, final stages of a sit & go, etc.)
b) where both players go all-in pre-flop (sit & gos would be ideal for this)

All cards would then be known.

In a large enough sample, the various hands should come close to their expected win rates. I can't see any reason why AA wouldn't win 85% of the time in a large enough sample. In Sit & Gos, it would also be interesting to see how often the bigger stack won vs. their expected win rate.

There is a fundamental flaw with this... whilst yes AA will hold up 85% of the time against a *random* hand, in the situation you describe you are not going to commit all your chips to the pot with a random hand. What I mean is that some of the times you have AA the other guy folds his trash instead, thus the sample is somewhat biased.

Thus, you have to evaluate the EV of each all in situation independently, and sum the expectation of each confrontation against the actual result in the long run...

To do this take the EV calculation of the heads up confrontation and put that in one column. In the other column put the outcome (0=lose, 1=win, 0.5=split). Over a large sample the two columns will converge...

[Elbow Jobertski]

Originally Posted by john.smith

Restrict the data set to only include situations where:
a) there are only two players at the table (e.g. heads-up tables, final stages of a sit & go, etc.)
b) where both players go all-in pre-flop (sit & gos would be ideal for this)

All cards would then be known.

In a large enough sample, the various hands should come close to their expected win rates. I can't see any reason why AA wouldn't win 85% of the time in a large enough sample. In Sit & Gos, it would also be interesting to see how often the bigger stack won vs. their expected win rate.

(a) would work, and (b) is overkill.

In a two player game all cards are known whether the all in comes pre-flop, the flop, or on the turn. As long as all players dealt go all in the MicroBob effect is eliminated because no cards are unnkown. A nine handed flip would follow a classic calculation.

Unlike a mucked hand which can be narrowed down a bit, a burn card is totally random and would be included as part of the stub in calculating equity.

[Bobo Fett]

Originally Posted by MicroBob

Don't forget to include tautometer's posts!


I don't know about a new thread for this. But there is probably enough meat in here just on pokerEV graphs and bunching effect stuff that may be of interest to some people (and worthwhile knowledge perhaps) that is not going to be seen by many user who don't bother to look in your typical thread with subject line "lets prove pokerstars is rigged".

That's why my wish would be to start a nice clean thread with a new title and hopefully an OP that will induce some good discussion...then move over the most of the last couple of pages of posts from this thread. Mike PM'd me and we're coming up with a plan; any suggestions for post #'s we should move? I'd like to just do a whole range to make it easier...if there's a bit of crap, I can pull those posts out.

[MicroBob]

tachyon - or you could just plug them into the pokerEV thing since it already does that for you.
But your overall point is correct. If AA wins 85% against any two randoms it should not win 85% in a heads-up match every time it gets all-in because the opponent is going to be far more selective with what he shoves than 'any two randoms'.