Good post! This sums up the main problem very nicely.

Juk

i don't expect the rng to be statistically true (if there is such a thing), and here's why:

the sites make money by collecting rake. the more hands a customer plays, the more rake the site will collect from that customer. therefore, the site has a non-zero interest in seeing their customers play as many hands as possible, or at least more hands than they would if the site were statistically "true." and since the sites can "prolong the lifespan" of their customers (most of whom are fish who get their money in bad regularly) by throwing them a single, random big pot from time to time, they probably do.

I cannot count how many times i've seen someone hit their miracle card with what might very well be the last of their roll (assuming that's why you buy in for 100, let it drop to 34, and not reload) and go on to play for 2-3 more hours. in other words, their suckout enables them to play 150 more hands -- and sometimes they even leave with some cash to play another day.

i know the above argument is lacking statistics, but since we will never satisfy the requirements to make statements with anything approaching mathematical confidence, i submit that it not be dismissed out of hand. it would be easy to put together >50 million hands from 2p2ers, but the sites have dealt what, 40 times as many hands?

as for myself, i truly believed that i was getting screwed by a site on 3 different occasions:

the first time, it turned out that i sucked at poker (and probably ran a little bad there). i also realized that since i was getting my money in good more often than my opponents, the reality was that i was going to get sucked out on more often than i sucked out on others.

the 2nd time, it turned out that i was above average at poker, but ran bad on that site. i learned this by running good on another site. i also realized that if i'm "always getting my money in good," i'm not gambling enough.

and the last time it turned out that i sucked really bad (25% conversion rate) at hitting my flush draws, but was average at hitting straight draws and hitting the flop with my "big ace" hands, and a little above-average at hitting sets. and i realized that it felt like i was getting screwed because i tended to bet the hell out of those flush draws, at the wrong times.

in other words, it was variance -- the same variance that can and will ultimately be used to explain away any statistical deviation we might find (along with "bunching," selection bias, and a host of other things), and the same variance that long-time posters tell newbies is the explanation for what they're going through. once i started getting decent volume, i saw all sorts of ridiculous **** happen on a monthly basis, for better and for worse. and i stopped worrying about it as much, because at the end of the day, the better i play, the less i "suffer" from variance.

as for mttr/pokerboy's argument, it seems to me that it would be a good, unconfirmable way for sites to prolong the lifespan of their average customer, so i wouldn't be shocked if it were occurring. but then again, i've seen plenty of regulars who were entirely unimpressive yet seemed to rocket up the levels, most likely because they ran hot, so....

lolaments, we're back where we started

cliffnotes: read the bolded section

Even as I confidently typed my cunning scheme to prove PokerStars is (not) rigged, I knew there had to be a catch...

What if you looked at the stats for each hand vs hand in isolation, e.g., AK vs 55. Since each all-in is an independent event and the times 55 is folded won't effect the results, should they be in line with expected values over a large enough sample? How many hands would be needed for each individual hand vs hand?

bob. i dont think your examples show anything. because youre probably just as likely to fall on one side as the other, in these situations.

that is to say...sometimes youre the one getting too much equity, and sometimes youre not getting enough equity. over a large sample, these should pretty much cancel out. any effect would be...scant, to say the least.

myturn2raise, while you might make a valid point that stars doesnt datamine and maybe that in itself is wrong, i know of someone who datamines games and has HH from up to almost 2yrs back for very cheap price. Maybe they do that so that players dont have a distinct advanatge over others.

stars doesn't have a firm doing audits..what kind of audits?? i hope you dont mean rng..as there is another site that hasn't any mention of there rng

As others have said there are so many examples and counterexamples that the only way to know if the Microbob Effect is significant is to run a simulation. Typical counterexample: Short stacker going in on the button with Ax (a fairly high probability holding), BB calling with a pair (or a wider range because he's getting odds, or has a big stack, or has a small stack and needs to gamble). Each one of these negates one of the other examples and reduces the size of the effect.

Yes, I don't really know. But you can't rely on 'probably' and 'should' cancel out. You actually have to see.

I think that for certain players they will NOT cancel out and I believe the graphs are beginning to show that. This was also alluded to by Phil (the guy who made the pokerEV thing in the first place).
It will skew downward on the equity thing particularly for shortstack players and for tighter full-ring players.
I'm going to explain this aspect as best I can and hope you're following what I'm trying to get at.

First I'm kind of covering the tighter full-ring players:
These are the types that are more likely to run into bad and random Ace high hands when they are getting it in. So their 67% chance of winning the hand according to pokerEV is really only 65% or something.
If they're playing 10/6 or something then they are going to have premium hands pretty much all the time (except in the blinds I guess). Their opponents will have worse hands a greater percentage of the time.

If everyone played their hands the same way then it would even out.
But everyone does not play their hands the same way.
Weaker players get it in more frequently with a bad ace for a hole-card for example. Better players do not.
This is why the fishy players who get it in with the worst of it can end up doing slightly better in long-term all-in equity. So many of their 32%'s were really 33 or 34%'s because of their love for the ace.

Every time they hang in there with A6 or even unimproved AQ vs. the tight player's KK on board of 632 they have a better chance of improving then the all-in equity would calculate.
The better player will not get in these situations as often where he's supposedly 32% but is really 34%.
There may be other scenarios and playing styles where even when he's supposedly 32% he is really down to 30% but I'm just focusing on the AA thing right now because that is most measurable with the all-in equity but I think the same can be said for somebody who is more likely to get it in with KQ or KT or something when you have QQ. They need help to beat you...and the fact that everyone else folded again indicates that nobody had KK or AK and this there is an ever-so-slightly greater chance that a K will ccome to help them suck out on you.

It's just a theory. But I believe the tighter full-ring players' graphs probably do tend to skew downward and I don't think it's because Stars is rigging it against them or that all tight players just HAPPEN to run bad but rather that the nature of their play is skewing the results.

Similar can be said for shortstack player even at 6-max. When it's folded to him in the small blind or on the button he has okay equity to shove with QJ or K7 or 44 or something. When you get to particularly tight players in the blinds who aren't paying attention then it becomes much closer to having enough fold equity to shove with any 2 randoms. Some good shortstack players will actually do this when they identify super-tight players.
They aren't calling with NEARLY as wide enough a range as they should (Mathematics of Poker and Kill Everyone books as well as Miller/Sklansky NLHETP and the charts all cover this in different ways).
Some of these bad-tight players (bad as far as countering effective shortstack strategy) will still be far more likely to call the all-in shove with AT or AJ or something. And a better player will see what's up and know to call the 20 shove even with A4 or something.
The shortstack can better read the deeper stack's range and doesn't necessarily have to call as much with A4 (might be some shortstack's style anyway, not sure).

The shortstack shoving with QJ or whatever is counting on folding equity + having 40% chance of sucking out if they get called by some sort of ace-high hand. But they really don't have 40% equity if everyone has folded before him because there is going to be a slightly greater preponderance of aces in the remaining cards (if the players themselves had aces and especially AA they would have been less likely to fold).

So we're essentially looking at 2 situations here and both of these players who have fared badly on the equity graph and think "it just doesn't look right" happen to be a shortstack specialist (or, at least, that's what myturn used to be) or tighter-than-average full-ring (1p0kerboy) players.
And they are both more likely to run into random Ax hands on all-ins when an A just happens to be slightly more likely to come (because everyone else in the hand folded).

Phil - I agree a simulation would be best.

The shortstacker scenario is interesting though.
A good shortstacker won't be afraid to get it in with QJ or something if he thinks he has folding equity.
But many deeper players are not going to counter by calling an all-in with QJ no matter how much the Sklansky/Miller charts tell you it may be correct to do so.

I'm an example of one who has a too-tight calling range against shortstacker pushes. I'll be more likely to 'give it a shot' with A5 though. And then I see that they shoved with J6 and I think, "Wow...he must have some stats on me that show I'm REALLY tight about calling in the blinds!!"

But for a good shortstacker against a player like me it is NEVER going to happen the other way. If he shoves with A5 then I'm not going to call with J6 much less J9 (because I'm a fraidy-cat and an ignoramus who hasn't figured out the correct calling ranges).

So I do think that the aggressive play of some shortstackers (especially hyper-aggressive ones) combined with the weak-fraidy-cat play of some of their opponents put them in more situations where they're QJ is running up against A7 than vice-versa.

Not to mention that if it hasn't gotten to the blinds yet then all the deeper players need to worry about each other too (pretty significant effect..it's a big reason why shortstacking can be profitable in the first place...all the deep players still need to worry about each other).

And this is why I think certain shortstack players can pretty much work themselves into this lesser equity on the pokerEV graph.
For some players the numbers of times they are all-in with the A as opposed to against an A does NOT even out. That's the theory anyway.

Would love to see if this is really happening for shortstack players and it actually shouldn't be too hard to look into.

Regarding the simulation:

Would very much like to see a simulation with correct shortstack (or even hyper-aggressive shortstack) vs. weak-tight calling range on a heads-up table. And then compare those results with having those very same two players in the blinds on a table where 6 or 7 other players are sitting and they play 'normally' meaning that anytime somebody else has AA or AK or KK we immediately know that such a hand would not have been included in the heads-up cash-game sample.

I absolutely cannot imagine that the equity results would be unchanged in those two situations.
Especially if the weak-tight calling range just happened to be a little bit more top-heavy with Ax hands (maybe he doesn't call with 22 or 33 even if it's correct to do so because he is just too weak-tight to KNOW that it's correct).

Others think that it would pretty much cancel out but I think there are certain styles of play where it does not end up canceling out.
I hope my theories as to why this might be are making some amount of sense.

Thanks again for your time in this thread.

[LEFT]I would personally love for Pokerboy or anyone else who is unsure to send their databases to someone else and let them do their own analysis and post. Throughout this thread I have read and read and keep thinking there is something important missing.

Not to say that I know much about this software because I do not. But is it possible that he ran good in non-all in showdown pots and ran bad in all-in showdown pots. Is the rake calculated as part of the equity here? Because if it is over Pokerboys sample that would make all the difference.

Just an idea, anyone willing to look at Pokerboys database and would you be willing to share it with us?

I think pokerboy said he los the whole database when his hard-drive crashed (might be thinking of someone else though). So that option may be out.

About the rake...yes, the software takes the rake into account and, for the most part, can be trusted to be 100% accurate about all the available information. But unfortunately it doesn't have all the available info so the situation of not being able to make assumptions about what each players' folding tendencies are may cancel out for all players i the long-run or it may not.

A thought:

I don't see posts putting forward possible theories of how the unfair RNG would actually rig the games favouring the poker sites. How would the program run in order to favour the bad players, say, in all in situations? What would the shuffler be told to do? We need plausible theories in order to give meaning to the possible, even if scant or circumstancial, evidence.

How about this though? I know there are ways to calculate how much increased their chance of getting an A or K if we also happen to know conclusively that every single player who folded did not have AA or KK.

As I mentioned before, I believe I read somewhere (maybe in Super System) that the chances of having AA is 1/220 randomly but the chances of the BB having AA after 7 players fold to you in the SB shoots up to 1/140 or 1/160 or something.

So perhaps there's a way for Phil to construct a pokerEV graph option to include the scenario where every single player who folded did not have AA or KK (as well as the possibility of DEFINITELY not folding AK or QQ if nobody has entered the pot yet).
If there's a way to look at a graph that somehow can incorporate those assumptions into the EV equation then we could actually see if there's any difference in the results.

It would still not yield accurate results of course. But if there is any semi-significant change at all really then that should serve as proof that the bunching theory over larger samples (MicroBob effect...meh...whatever) is valid.

Another simulation or run of stats that would be interesting to see would be to look at hands where two players get all-in preflop and everybody else folds (without having entered the hand at all) and then determining how much more common aces are to come on the board.
Doesn't even matter if either of the all-in players have an ace or not.
If it's AK vs. AK then there are two A's left to come on the board...and my theory is that the A's will appear ever-so-slightly more than they're due.

Would have to eliminate AA vs. AK type hands because that would mean there is only 1 left and thus the normally 'incorrect assumption' that pokerEV has that everyone else DEFINITELY didn't have AA would actually now be a correct assumption.

imbecil - I disagree. So many people have their own theories how and why the sites would do such a thing and most of it is nonsense and silly and kind of people making up crap.
I've heard about 1000 different theories of how the site does it (many of them contradicting each other).
If ANY of the theories were correct then the proof would be in the cards.
Babbling on about one theory or another is kind of a big time-waster I think and a big reason why people normally hate threads like these. Because the theories are, most of the time, just a bunch of anecdotes or people making stuff up.

That's an interesting idea, I was planning to do it differently. I was going to run a simulation of a million played hands with various stack sizes and player styles, then calculate the EV on these hands in two ways. One without knowing any of the hole cards except for the all-in players (just like the regular program), and the other including the known dead cards in the EV calc. The results will be very different over the short term but will converge to the same results over the long term if there isn't a bunching effect (or skew if there is).

Your method removes any of the statistical uncertainty in the result, but has the downside of not being representative of the folding/pushing choices at real table. Not sure which is best.

Phil;

I don't know if anyone has mentioned this, but here it goes.

Take a 6max regular at the midstakes, a grinder who is on pace for SNE. Or a FR regular. I would think 2/4nl would be the best spot, since many of the 2/4 players play and post here, and many are grinders. I guess 1/2 could work as well.

Take that one player, have him send you his hands for as long as he has. Someone hopefully that has played for a couple years grinding pretty hard, but a couple million hands would be great.

Now put an announcement in the internet gambling forum. Have everyone else put announcements everywhere else. Ask for hands played against this player, any hands you have that you played vs him. Have people upload them on a place on the internet, likely on your site, it would be simple to have a browse box that people can use to upload the hands, .zip format.

I would def. agree to do mine, but I really doubt I have enough, I played like 1.4mill hands last year rand prob 600k this year, but I'm all over, starting at 25nl last year and going up to 5/10nl FR, and now playing 2/4nl 6m since last month, but if I am a good enough candidate I don't have any problem sending you my hands(it wouldn't put me at any type of disadvantage or anything).

Basically take the hands in which you have a decent amount of holecards shown, and run the studies when you have 3 or 4 sets of holecards. Dunno if you could get enough hands, but I mean it can't be too hard. Many people on here would also probably even request hands from Pokerstars if they didn't still have them on their machine

phil, i haven't used poker ev in a long time, but i was wondering if it shows how many hands of a particular sample actually involve an all-in confrontation. if a person had that information, it would be much easier to make standard deviation calculations as different players will have these confrontations at much different frequencies.

the HEM sample that i posted was over 218k hands, and i'm pretty positive the reason that its possible for my results to be so skewed (positively in my case) is that i was hardly ever all-in, due to the style that i was playing, making the sample much smaller than just the volume of hands would indicate.

phil - I really like your idea too and don't think it's such a bad plan to just go with both of them.
Calculating the equity of different players at different styles on a table and looking at both 'regular pokerEV equity' vs. 'removed-card knowledge equity' could definitely get us somewhere.


For your thoughts on the bias in my shortstack suggestion:
Defnitely.
Different shortstackers play differently and will even adjust based on opponents.

You can get to a point where a big-blind player is 9/4 overall and he doesn't know when he's supposed to call and maybe only does so 10% of the time (maybe AQ+ and 99+ or something like that).
And the shortstacker seeing this situation will open-shove for 20BB's in the SB with pretty much any-two cards (or close to it).

That's how wide the difference can be.
And, of course, the 10% guy is going to have a much higher percentage of A-high hands make up his range.
And if he gets up to 15% then he probably added even more A-high hands to the mix.

Considering the vast differences that are out there perhaps you can experiment with different ranges including more ace-high heavy calling ranges.

Have 60% shoving shortstacker vs.20% calling big blind.
And then 40% shoving shortstacker vs. 30% calling big blind. etc etc.

The idea here is to see if ANY shortstack players of a certain type can actually play themselves into some all-in equity skew as well as which types of players would tend to be the most likely to see this the most.

I can't read any of your posts Micro, but what about MY idea? I think it's genius.

schwallie - That's a pretty good idea too.
So many of the NL400 full-ring and even 6-max players have played against each other so much that you would end up with a lot of hands where you know more than 50% of the hole-cards.
Real hands with may more information.

Just getting 6 or 7 of the PokerStars NL400 types to do this would end up with a lot of hands with 4 or 5 of them at the same table.


edit - Yes schwallie. Definitely genius. I really like it.

The biggest concern for the skewing of results is the unrandom distribution of cards of folded players. Therefore, if any serious analysis can be done at this moment using pokerEV, it should be done using Heads Up Cash/ SnG Hand Histories.

I think it would serve to show an example of how anything except exactly perfect analysis could result in skewed certainty results.

QQ vs AKs is going to be 53.951% for QQ to win if they were heads up.

I cannot determine myself what the chances are QQ vs AKs when folded AA and KK are not possibilities. But we should see for how many hands it takes for this new scenario to be 3 deviations away from 53.951%.

This should make it pretty clear that, although the bunching affect cancels out to some extent, if it doesn't cancel out perfectly, then after a certain number of hands, it would put you 3 deviations away from the norm.

playing on FT today i was literally rivered 32 times after i had put my money in good. I'm talking 3, 4 outters, all ****ing day.

I finally got so tilted that i cleaned out my remaining balance playing ring games since it was too little to cash out, then i proceeded to clear out all my ftp points in 600ftp sitngos, one of those i was rivered on a 3 outter heads up to win.

So now i have no points, no balance, its deleted from my hardrive, and i will never return to FTP. Ever. I've been a regular player for two years there, but the shear amount of money i lost on the river today was too much for me take.

FU FT!