Rigged: The AA test

19RK64 said:

If you notice when you have KK, if you were UTG or UTG+1, you most likely limped or just bumped the bet (1 or 2x BB) same as AA, should of done, in that position, really, when YOU DO lose with KK, it cause 70 sum % of the time someone has AA, or YOU are gettin spiked, not many other hands should be taking out KK,LOL, really your answer is logic.. their just aint no guarentees!!, cards are cards, if it get to postflop 2-7 can win. am I saying I allways lose with MONSTER PAIRS? NO-WAY, they have paid me good..but they to lose also...
If I felt sites or poker, were "rigged or setup"I wouldn't waist my time playing!. either way, it makes you look at board postflop and so on, before committing yourself..

Click to expand...

what

grift said:

When I use the term "we" in the following post I mean the forum in general.

Let's talk statistics.

4dogs is testing a hypothesis. For more on what that means:

Statistical hypothesis testing - Wikipedia, the free encyclopedia

I'm going to paraphrase some of this article just to get to the essentials.

We have a null hypothesis. The null hypothesis is kind of our "assumption." Here, 4dogs is assuming that the sites are rigged, by dealing a different number of a certain hand than expected. He collects data, and attempts to show that this particular assumption is incorrect.

He showed this to a pretty convincing confidence level, using the suitable hypothesis testing method. His data show that it is very unlikely that the major sites rig hands by dealing different amounts of certain hands than the expected amount.

That is what the test is about. It only proves that the site is not rigged using this method. It simply disproves his null hypothesis. These data make absolutely no statement as to whether the site is rigged in another manner. Pocket 2's could hit quads every time they are dealt, be dealt the correct amount of times, and thus the site would be rigged, but we could still disprove our null hypothesis because the expected amount of starting hands hasn't changed. This is why discussion of other forms of rigging is outside of the scope of this thread. 4dogs has acknowledged this. We all have acknowledged this.

Any other discussion of whether a site is rigged is outside of the scope of these data. Which means, why talk about it here? Why not start a new thread, collect new data, and do a new analysis? That way, posts stay relevant, and we're not just throwing theories around, we're actually looking at data and making informed decisions.

I think this is the point you've been missing that 4dogs didn't really elaborate on.

Click to expand...

Well justnorv, this pretty much blows another one of your theories out of the water. You implied that grift and my writing style were suspiciously similar? I'm flattered that you think I write this well.

grift. I've made several attempts to say the same thing, but never as well as you just did. Thanks for the best post so far in this thread.

I love the arguments that 'if you think it's rigged, then why carry on playing?' Or 'if you think it's rigged, then you must be a bad player, who cant handle bad beats?'
FWIW I believe that some flops are tweaked to become action flops, so that any people left in will get a piece of the flop, so to generate betting, and therefore more rake for the poker sites. The turn and river cards are also 'actioned', sometimes, for the same reason.
So, you may ask, if you believe that, then why carry on playing? My reply would be Do you drink? Did you know that a bottle of wine that costs $10 to the consumer, will typically cost in the region of $4 to produce(from planting the vine, to production costs, to wages for everyone involved etc to the bottle hitting the supermarket shelf) so, you're being ripped of to the tune of $6 for each bottle, will that stop you drinking it? No, i didnt think so. We drink wine because we enjoy it, even if we are being conned over the price. Well, I enjoy poker, both online and live. Immensely. And it's cost me nothing over the 3 or so years Ive played online, as I've cashed out more than Ive deposited, so Im happy. Even if I had lost money, thats ok , it's still fun.
As for being a bad player, no, i dont think so. i consider myself ok. Ive beaten 2000 in a MTT freeroll(6 hours of play), amongst other wins, and have frequently cashed at my local casino. I get my bad beats, the same as everyone else, and accept them. I know that any higher pair is roughly 80% to win against a lower pair, so of course the lower pair will win 1 in 5 times.
This isnt about bad beats, and how many times you get dealt AA etc, this is about flops being actioned to hit whatever hands are left in play, so it doesnt matter whether you have AA, or 72os going into the flop, some of the time the flop will be 'actioned' to make you feel you are ahead to generate betting, and therefore, rake.
So IMHO, whether this thread's results prove that AA or whatever has been dealt the statistical amount of times over x amount of hands or it doesnt will make no difference to me, as it's irrelevent to my, and many others'(good and bad players) theory about how poker is rigged.(some of it............)

KerouacsDog said:

I love the arguments that 'if you think it's rigged, then why carry on playing?' Or 'if you think it's rigged, then you must be a bad player, who cant handle bad beats?'
FWIW I believe that some flops are tweaked to become action flops, so that any people left in will get a piece of the flop, so to generate betting, and therefore more rake for the poker sites. The turn and river cards are also 'actioned', sometimes, for the same reason.
So, you may ask, if you believe that, then why carry on playing? My reply would be Do you drink? Did you know that a bottle of wine that costs $10 to the consumer, will typically cost in the region of $4 to produce(from planting the vine, to production costs, to wages for everyone involved etc to the bottle hitting the supermarket shelf) so, you're being ripped of to the tune of $6 for each bottle, will that stop you drinking it? No, i didnt think so. We drink wine because we enjoy it, even if we are being conned over the price. Well, I enjoy poker, both online and live. Immensely. And it's cost me nothing over the 3 or so years Ive played online, as I've cashed out more than Ive deposited, so Im happy. Even if I had lost money, thats ok , it's still fun.
As for being a bad player, no, i dont think so. i consider myself ok. Ive beaten 2000 in a MTT freeroll(6 hours of play), amongst other wins, and have frequently cashed at my local casino. I get my bad beats, the same as everyone else, and accept them. I know that any higher pair is roughly 80% to win against a lower pair, so of course the lower pair will win 1 in 5 times.
This isnt about bad beats, and how many times you get dealt AA etc, this is about flops being actioned to hit whatever hands are left in play, so it doesnt matter whether you have AA, or 72os going into the flop, some of the time the flop will be 'actioned' to make you feel you are ahead to generate betting, and therefore, rake.
So IMHO, whether this thread's results prove that AA or whatever has been dealt the statistical amount of times over x amount of hands or it doesnt will make no difference to me, as it's irrelevent to my, and many others'(good and bad players) theory about how poker is rigged.(some of it............)

Click to expand...

+1
I don't know if its rigged or not, but I have doubts. I don't care how many time I get AA or KK or flop a flush. I want to know how many times my big hands are up against other big hands. The last two times Ive had Kings, they have been up against Aces BOTH times, pathetic. I'm also running 22 buyins below expectation, pathetic.
The swings online are just INSANE. Anything can happen over ten thousand hands, bad players can win, good players can lose. I'm sorry but I cant see that happening live. If you get a good player and put him in a live casino, I think hes gonna win if he played 6 hours a day for 52 days (I estimated that's about 10K hands).
I know everyone is sick of donks saying poker is rigged because their aces got cracked or because someone hit a gutshot. I also think these people are idiots. I know bad beats are part of the game and I accept them, there not even bad beats to me, because they are supposed to happen. I wouldn't even call a 1 outer on the river on the bubble of a sng a bad beat.

I actually lean towards online poker not being rigged. But I think the chance that it is, is a lot bigger than most people think.
Just think about it, if it was rigged, don't you think these multi million $$ companies would rig it in a way that would be impossible to find out or at least really really hard.
People often say that these huge sites would never risk their reputation and all there money, just to win a little extra. That's not a very good argument imo. Like Daniel Kimberg said in his book, "But we don't live in a completely rational world, and online poker would not be the first parties guilty of an incredibly stupid form of greed."
I think that if the big sites could rig their games (which they obviously can) in order to win a little extra $$, and the chance that they got caught was slim, then they would rig them without thinking twice.

The bottom line is, online poker MIGHT be rigged.
No one can probably ever prove if it is or isn't, therefore I think the people who take the pi$$ out of people who think its rigged, are just as idiotic as the people who insist its rigged. The odds that it isn't rigged definetley out weigh the odds that it is rigged, but you cant be sure.
The only real way you can find out if a site is rigged or not, is to get complete access into their machines and software.
Since none of us will ever get that chance, I think we should just all STFU about online poker being rigged right now, it depresses everyone.

Four Dogs said:

Well justnorv, this pretty much blows another one of your theories out of the water. You implied that grift and my writing style were suspiciously similar? I'm flattered that you think I write this well.

grift. I've made several attempts to say the same thing, but never as well as you just did. Thanks for the best post so far in this thread.

Click to expand...

Four before you reffered to my half baked idea and I asked which one and got no reply. Now I would like to know which theory of mine you are referring to.

Actually, grift could you answer these questions for Four?

grift said:

When I use the term "we" in the following post I mean the forum in general.

Let's talk statistics.

4dogs is testing a hypothesis. For more on what that means:

Statistical hypothesis testing - Wikipedia, the free encyclopedia

I'm going to paraphrase some of this article just to get to the essentials.

We have a null hypothesis. The null hypothesis is kind of our "assumption." Here, 4dogs is assuming that the sites are rigged, by dealing a different number of a certain hand than expected. He collects data, and attempts to show that this particular assumption is incorrect.

He showed this to a pretty convincing confidence level, using the suitable hypothesis testing method. His data show that it is very unlikely that the major sites rig hands by dealing different amounts of certain hands than the expected amount.

That is what the test is about. It only proves that the site is not rigged using this method. It simply disproves his null hypothesis. These data make absolutely no statement as to whether the site is rigged in another manner. Pocket 2's could hit quads every time they are dealt, be dealt the correct amount of times, and thus the site would be rigged, but we could still disprove our null hypothesis because the expected amount of starting hands hasn't changed. This is why discussion of other forms of rigging is outside of the scope of this thread. 4dogs has acknowledged this. We all have acknowledged this.

Any other discussion of whether a site is rigged is outside of the scope of these data. Which means, why talk about it here? Why not start a new thread, collect new data, and do a new analysis? That way, posts stay relevant, and we're not just throwing theories around, we're actually looking at data and making informed decisions.

I think this is the point you've been missing that 4dogs didn't really elaborate on.

Click to expand...

Grift thanks for explaining what Four meant. I didn't realize that all he was doing was agreeing with me. He could have just said that.

jstnorv07 said:

Grift thanks for explaining what Four meant. I didn't realize that all he was doing was agreeing with me. He could have just said that.

Click to expand...

All he was disagreeing with is your overwhelmingly annoying personality.

Cheers,

grift

grift said:

All he was disagreeing with is your overwhelmingly annoying personality.

Cheers,

grift

Click to expand...

Was that an insult?

I was kind of hoping that everyone else here with other theories about how a site may rig the game would see the basis that this sort of statistical analysis is for further investigation. Now that we've done something basic like this, why couldn't we move on to designing experiments for the more complex forms of rigging that have been discussed.

For instance, let's say you've flopped a set, what are the chances the other guy also flopped a set? You've flopped a boat, what are the chances the other guy flopped quads? This type of hand would generate lots of action, and would always go to showdown resulting in us being able to see both players cards and thus generate usable data. Furthermore, with a couple of assumptions we could come up with a pretty accurate expected value for this type of hand.

We could look through some hand history data and see how often on average people go into hands with certain pocket pairs. This is obviously not rigged, because the decision is totally on the user. Then we can look at how often these pocket pairs are dealt. Again, we're assuming this part is not rigged because we've just disproved that in this thread (see how useful this thread is becoming?). Now what do we have? We know how often people go to the flop with these pocket pairs, and we know how often they are dealt, so we can generate an expected value for these types of situations, and compare the results to the actual data. And thus we would be able to see if the poker sites rig in this manner.

The only problem is we need to find a way to sort hand histories based on the actual showdown hand. This should be pretty easy using a simple script. I would probably write it in matlab because I'm more familiar with it, but if someone else here is proficient in a more efficient language for parsing files such as these they could post and let us know.

Anyways, long rant, but this is the reason why I'm interested in 4 dogs analysis. Like I said before, it's a good start, and we can come up with lots of interesting things based on the "fair start" assumption proven by these data.

jstnorv07 said:

Was that an insult?

Click to expand...

Was what an insult?

Cheers,

grift

jstnorv07 said:

Four before you reffered to my half baked idea and I asked which one and got no reply. Now I would like to know which theory of mine you are referring to.

Actually, grift could you answer these questions for Four?

Click to expand...

This was kind out there.

jstnorv07 said:

Here is a pretty out there idea that I just thought of that I reaize would probably be very difficult to achieve. Again please don't attack this as I know it is pure conjecture. What if a US player was playing someone in Berlin. There is an ever so slight delay between him seeing his hand and you seeing yours. Software is written that determines his hand after you make your move. i.e. you push all in with A's, instantly and unbeknownest to either... Berlin player sees K's and flop is set to drop a K. I hope I don't regret posting this one.

Click to expand...

grift said:

I was kind of hoping that everyone else here with other theories about how a site may rig the game would see the basis that this sort of statistical analysis is for further investigation. Now that we've done something basic like this, why couldn't we move on to designing experiments for the more complex forms of rigging that have been discussed.

For instance, let's say you've flopped a set, what are the chances the other guy also flopped a set? You've flopped a boat, what are the chances the other guy flopped quads? This type of hand would generate lots of action, and would always go to showdown resulting in us being able to see both players cards and thus generate usable data. Furthermore, with a couple of assumptions we could come up with a pretty accurate expected value for this type of hand.

We could look through some hand history data and see how often on average people go into hands with certain pocket pairs. This is obviously not rigged, because the decision is totally on the user. Then we can look at how often these pocket pairs are dealt. Again, we're assuming this part is not rigged because we've just disproved that in this thread (see how useful this thread is becoming?). Now what do we have? We know how often people go to the flop with these pocket pairs, and we know how often they are dealt, so we can generate an expected value for these types of situations, and compare the results to the actual data. And thus we would be able to see if the poker sites rig in this manner.

The only problem is we need to find a way to sort hand histories based on the actual showdown hand. This should be pretty easy using a simple script. I would probably write it in matlab because I'm more familiar with it, but if someone else here is proficient in a more efficient language for parsing files such as these they could post and let us know.

Anyways, long rant, but this is the reason why I'm interested in 4 dogs analysis. Like I said before, it's a good start, and we can come up with lots of interesting things based on the "fair start" assumption proven by these data.

Click to expand...

Grift, there are a couple of problems i can see with this. One is that once you get passed your own hole cards, there are way too many variables to consider. The way you play your hand, and the way your opponents play theirs makes any results too subjective to be meaningful.

Poker Tracker actually does track how your own cards would have played out had you continued to the end, but a sample size needs to be at least several hundred times greater than the total number of possible outcomes to have merit. When your dealing with only 2 cards, there are 1326 possible outcomes, when your dealing with 7 there are about 133 million.

You saw how difficult it was to get a sample size from PokerStars that is about 500 times larger than the 1326. Imagine trying to achieve the same degree of accuracy with 133 million. I think the beauty of the AA test was its simplicity. But, if I were to try another I think I'd do it with something less obvious; 72o would be poetic.

Thanks grift and yes, justnorv too for some good debate.

There's probably a way to get around the large sample size we just need to think about it a bit.

One tool that comes to mind is re-sampling:

Resampling (statistics - Wikipedia, the free encyclopedia)

Anyways, even if we can't get something nice and neat, it'd still be fun to at least see just how small of a sample size we can really get away with.

Or, we could find a way to test for randomness. I mean, if these sites are rigging and are supposedly "action generating" there would definitely be some sort of unexpected trend in the size of the pot.

Yeah I'm just throwing ideas out there, I think it would be kinda cool if we could come up with a way to measure this stuff.

Well after looking through my handy stat book the smaller sample size/low probability of a "success" does make it pretty hard to do.

Four Dogs said:

This was kind out there.

Click to expand...

Oh, so you were saying that the idea that I said was out there was out there. OK

Four Dogs said:

Grift, there are a couple of problems i can see with this. One is that once you get passed your own hole cards, there are way too many variables to consider. The way you play your hand, and the way your opponents play theirs makes any results too subjective to be meaningful.

Poker Tracker actually does track how your own cards would have played out had you continued to the end, but a sample size needs to be at least several hundred times greater than the total number of possible outcomes to have merit. When your dealing with only 2 cards, there are 1326 possible outcomes, when your dealing with 7 there are about 133 million.

You saw how difficult it was to get a sample size from PokerStars that is about 500 times larger than the 1326. Imagine trying to achieve the same degree of accuracy with 133 million. I think the beauty of the AA test was its simplicity. But, if I were to try another I think I'd do it with something less obvious; 72o would be poetic.

Thanks grift and yes, justnorv too for some good debate.

Click to expand...

Are the A's winning as often as they would be expected to?

jstnorv07 said:

Are the A's winning as often as they would be expected to?

Click to expand...

That would be a good test, but who's to say how often AA is expected to win? We know how pocket aces should fare vs a random hand; something like 80% (I'll run it through Poker Stove when I get home) but once a player has the chance to make a decision, bet, fold, raise, check, the hand is no longer random. In holdem, the only thing that is (or should be) random is your own hole cards and those of the players around you. But we can't see theirs, and if we do then their randomness has been tainted.

However! If we're just interested in determining randomness, we don't have to restrict ourselves to Hold'em. In Stud, your first 3 cards are random, in Omaha it's 4. In draw, it's 5 but there aren't many sites that spread it and most poker tracking software doesn't include it anyway.

Four Dogs said:

That would be a good test, but who's to say how often AA is expected to win? We know how pocket aces should fare vs a random hand; something like 80% (I'll run it through Poker Stove when I get home) but once a player has the chance to make a decision, bet, fold, raise, check, the hand is no longer random. In holdem, the only thing that is (or should be) random is your own hole cards and those of the players around you. But we can't see theirs, and if we do then their randomness has been tainted.

However! If we're just interested in determining randomness, we don't have to restrict ourselves to Hold'em. In Stud, your first 3 cards are random, in Omaha it's 4. In draw, it's 5 but there aren't many sites that spread it and most poker tracking software doesn't include it anyway.

Click to expand...

Would it be possible to see how often pocket A's are folded? I would imagine it would be a small enough number to not signifigantly effect a sample. After that I would think we would have to make a reasonable assumption. The assumption being that the variation in play would balance out in the sample.(i.e. From the player who goes allin everytime he has A's and wins small amounts but increases winning percentage to the player who slowplays and brings down his winning percentage but wins larger amounts when he wins.) If we can make that assumption I think we could reasonably say how often the A's should win.

This thread is a very valuable one. And while the subject matter is one that causes tension as people's belief systems are challenged there's no need for insults or rudeness, plenty of other forums for that. There's a fair bit in this thread that's close to the line.

Anyone, either pro or anti r1gged gets abusive and I'll get busy with the deleting and the warnings.

Keep up the good discussion folks and much kudos to FD for actually providing some evidence to try and test what should be testable positions(still waiting for some empirical testing of a rigged theory, any theory).

Love and kisses

Rex
(a moderator imo)

Irexes said:

This thread is a very valuable one. And while the subject matter is one that causes tension as people's belief systems are challenged there's no need for insults or rudeness, plenty of other forums for that. There's a fair bit in this thread that's close to the line.

Anyone, either pro or anti r1gged gets abusive and I'll get busy with the deleting and the warnings.

Keep up the good discussion folks and much kudos to FD for actually providing some evidence to try and test what should be testable positions(still waiting for some empirical testing of a rigged theory, any theory).

Love and kisses

Rex
(a moderator imo)

Click to expand...

Your intolerance of intolerance will not be tolerated.

oooooooooo rexy just got playonworded (plowed)... machete deathmatch will ensue imo.

I now believe that online poker is rigged. I was recently playing in a 180 SNG and had a huge chip lead with about 25 people to go. I had been playing very well the entire tournament and then started playing like a complete idiot. I managed to finish 17th and ITM due to the massive chip lead I had built over the entire tourney. Obviously, the site was sending me subconscious signals to disrupt my play.

P.S. In case anyone didn't catch it this was sarcasm. Most of the story is true though.

jstnorv07 said:

I now believe that online poker is rigged. I was recently playing in a 180 SNG and had a huge chip lead with about 25 people to go. I had been playing very well the entire tournament and then started playing like a complete idiot. I managed to finish 17th and ITM due to the massive chip lead I had built over the entire tourney. Obviously, the site was sending me subconscious signals to disrupt my play.

P.S. In case anyone didn't catch it this was sarcasm. Most of the story is true though.

Click to expand...

Got it. Had me goin' there for a second though.

Awesome Thread

Just started using this forum... love this thread.

I'm pretty tired of listening to people whine about getting beat. I was playing in a live game the other night, and people complained about getting beat by better play... no one said that the live game was rigged...

Welcome to the forum chloerox. Glad you like it. Now if I could just get people to start contibuting more data.

Although a different theory altogether, I am beginning to think that the river hits lesser hands more often in Bodog that in live games (bad beats). Not sure how to get stats on this theory. (My feelings are probably skewed by me getting beat on the river, but it also appears that others I watch also get beat by the worst hand if it goes to the river.)