Rigged: The AA test

[Four Dogs]

I know we're all sick and tired of the on-line rigged debate. I've been trolling this site for 3 years and it just never seems to end, and the recent Absolute fiasco hasn't helped matters, never mind that what happen there actually had more to do with inexcusably loose security measures rather than actual RIGGING.

The logic behind the belief that the sites rig the game is ALWAYS the same and NEVER backed up by any facts. The belief is that the sites somehow profit from increased rakes by pitting big hands against eachother.

"Now why is it rigged...well like many people here have said, its because of the amount of rake they get...now alot of idiots dont know the bigger the pot, the more they rake...they also dont realise if there are MORE MONSTER HANDS ie, AA, AK, KK, there are going to be more all ins, the more all ins, the more you deposit once again..."

This quote is from a recent thread, but really, it's just one of hundreds and there's nothing unique about it. What surprises me is that this belief persists despite the fact that due to programs like Poker Tracker, these inane assertions are easily verified or dispelled.

So lets test it. Let's keep it simple though. I'd like any of you who actually track their HH with some sort of software like PT or Poker Office to post the number of ring hands you've tracked (no rake in tourneys right?) along with the number of times you've been dealt AA. Also, to sort the cheaters from the honest we should break the numbers apart by site. If the argument holds water then we should all be seeing hands like AA or KK more often then expected. I think the chances of being dealt AA or any pair is like 220:1 or about 4.5%

I'll start.

Site: Full Tilt
Number of ring hands: 16,901 (I've actually played about 3x that number but I lost my DB about a year ago)
Dealt AA: 68 (a little low)
Expected: 77
I am dealt AA: 1/248.5 hands = .402%

Site: Poker Stars
NRH: 5323
Dealt AA: 36 (a little high)
Expected: 24
I am dealt AA: 1/147.86 hands = .676%

Total
NRH: 22454
Dealt AA: 106 (a statistical bullseye)
Expected: 102
I am dealt AA: 1/211.83 hands = .4545%

I put alot more info in here than we need.
Just put the hands and times dealt AA and I'll do the math. I'll try to keep a running tally going, but I don't plan on being around much in the next few days so anyone may feel free to update it.

Cheers and Merry Christmas. :icon_sant

 

[Tammy]

This is fantastic. I'm going to sticky this so that is doesn't get lost in the shuffle. Come on guys, let's see those stats!

 

[Tygran]

Just AA data, and just pokerstars data but here you go.

6312 hands
dealt AA: 29 times
Expected: 28.6 times


looks pretty good to me

 

[Tygran]

I think the chances of being dealt AA or any pair is like 220:1 or about 4.5%

Minor correction, your odds are correct but the decimal is off a point in the percentage .

Should be ~ .45% chance




Great post, keep it going people!

 

[Boltneck]

What a brilliant idea - it won't shut the conspiracy theorists up, but at least they'll have to think of a new angle / excuse!

I've got 15,997 hands available at present. I created a new database when I made a fresh start at poker a couple of months ago. I have however got ALL my old hand histories archived (perhaps 50,000 more) so when I have time, I'll create a new db and load them also. For the moment, the results from my current db are:-

AA - 69 times (win 82.61%)
Expected - 72 times
Pretty close to expected

KK - 57 times (win 63.16%)
Expected - 72 times
Rather low, but not suspiciously so

The above is on the iPoker network.

Not sure what the win % should be expected (and it's probably not relevant to this thread) but my aces seems rather higher than I would of expected.

 

[switch0723]

you can work out the percentage to be sure

hands dealt: 12972

AA: 57

 

[tenbob]

Site: Pokerstars
Number of ring hands: 72,635
Dealt AA: 314
Expected: 330
I am dealt AA: 1/231.2 hands

 

[ChuckTs]

Site: Pokerstars
Number of ring hands: 33,252
Dealt AA: 158
Expected: 150
I am dealt AA: 1/210.5 hands
Won: 87.97%

 

[aliengenius]

Here is a similar thread.

 

[F Paulsson]

Site: Pokerstars
Number of ring hands: 116,766
Dealt AA: 535
Expected: 528
I am dealt AA: 1/218.3 hands

 

[Four Dogs]

Minor correction, your odds are correct but the decimal is off a point in the percentage .

Should be ~ .45% chance




Great post, keep it going people!

Oops thx for the fix.

 

[Four Dogs]

you can work out the percentage to be sure

hands dealt: 12972

AA: 57

Switch, is this Full Tilt?

 

[SeanyJ]

Site: Pokerstars
Number of ring hands: 31,216
Dealt AA: 137
Expected: 141
I am dealt AA: 1/227.85 hands

 

[mendozaline]

It all depends on how you design the experiment

.....to post the number of ring hands you've tracked (no rake in tourneys right?) along with the number of times you've been dealt AA. Also, to sort the cheaters from the honest we should break the numbers apart by site. If the argument holds water then we should all be seeing hands like AA or KK more often then expected.

I think you are making two mistakes here. One, I don't think it's a given that rigging would apply to ring games but not tourneys. Two, I don't believe that rigging would be to get more rake for the casino. In a way, though, that's just two different ways of saying the same thing.

For these two reasons, I think it would be a mistake to spend too much time studying AA and KK and QQ statistics. That's a little too obvious. Maybe JJ and AKo since they already have a love/hate relationship attached to them.

No, in my opinion it would make much more sense studying other card combinations. Pick one maybe that would normally be considered a hand you would fold, or something like that. See how often they win vs how often they should win.

Or another good one might be to study the community cards against a control say of the 21st to 25th cards in a million shuffles by a shuffling machine. As a corollary to this, what were the winning cards, and what is their expected frequency against actual frequency. This would be a good test, because it would force the issue. One or other of the sides would have to admit they were wrong. I would certainly admit I was wrong if this showed no significant difference.

 

[Boltneck]

This is a "wouldn't it be nice if......" not a suggestion, because it's probably impractical. A quick bit of (rough) mental arithmatic tells me that we have about 350,000 hands posted so far. I'm sure there are many others who have PT / PO who haven't posted. Just think what we could do if we could collate all of those stats. into a single PT db (I've got all my HH's archived - as I'm sure do most others).

I suspect someone (AG would be odds on favourite) is about to post a link to someone who already has done this, but if it doesn't already exist, it would be awesome!

 

[F Paulsson]

What kind of information would you want from a big database? I mean, I've 120k from Stars alone this year, and few statistical anomalies require a large sample than that to find, really. I guess what I'm saying is that if there's something you'd want to test Stars ring games for that can be detectable in PT, I can probably do it with my DB alone.

/FP

 

[Boltneck]

What I was most interested in was a comparison of the number of times when 2 players had the same top hand ie AA vs AA, KK vs KK, QQ vs QQ or AK vs AK (either suited or not in the last example).

It would need a few lines of PL/SQL to extract it because it's not going to be available with any of the existing PT stats.

I would stress that I am not a subscriber to the rigged theory, my motives are to rule out as many of the common arguements as possible, hence looking for ways that rake could be increased that would not be so easily detectable.

 

[switch0723]

Switch, is this Full Tilt?

full tilteroo

 

[zachvac]

Actually I've never seen anyone argue this, it's too easy to disprove. Instead, people argue that AA and KK coincide more, so that AA still comes out as much as it should, and KK comes out as much as it should, but given there's an AA out there, KK comes up proportinally too much and vice versa. So even if (when) we come up with a perfect proportion of AA, people will still continue to believe that when we get AA, KK comes up more often than it should. This is not a testable hypothesis because we don't take every AA hand to showdown and especially when we set, it's possible someone is laying down KK. Although I guess what we could do is take all the hands in which we have AA where the board doesn't hit an A. We're assuming no one would fold KK to an under flop when we're in it with aces. Of course I guess that wouldn't work because that's assuming we never fold our aces, which isn't true at all.

So scratch that last idea. Basically, we are trying to come up with ways to prove a negative, which is a logical fallacy. The burden of proof is on those who wish to prove that the site is rigged. But I'll play along with this.

Pokerstars:
39,420 hands
AA 172 times

178 expected (damn pokerstars, ripping me off)

Full tilt:
3,798 hands
AA 21 times.

17 expected (now why have I only played 3798 at full tilt and 39420 at pokerstars? Perhaps a change in site is in order )

 

[Four Dogs]

Sorry for taking so much time to get back to this. Hollidays and all. I hope you all understand that this is just for fun and that IF a site wanted to fiddle with their shuffling algorithm, there are other, less obvious ways they could and would do it. IMO, the action hand argument is the most widespread, probably because it's the easiest to understand.

Dispelling the notion that big rake generating hands like AA or KK are more prevalent should go a long way toward silencing all but the most hard core conpiracy theorists. Take away their biggest weapon and they might actually have to learn something about the issue to continue the debate, something they are unlikely to be qualified or inclined to do.

Very few of us have a DB big enough on its own to prove or disprove anything about the frequency of a given hand but many who don't understand this might think that their library of 20,000 hands is a sufficiently large enough sample to do just that. I'm no statistician, but it seems to me that any such assertion is meaningless with anything less than 10 times that number. If anyone knows something about statistics and standard deviations I'd be happy to hear from them.

But there's a better more satisfying reason for conducting this experiment than using it to defuse ignorant assumptions based on scant evidence. Even if there is no actual calculated ill intent on the part of the poker sites, it would be interesting to see just how good their RNG's and shuffling algorithms are. with just short of 400,000 hands we are deviating from our expected result by just a little over .5%. Unfortunately, only the poker stars sampling is by itself large enough to have merit, thanks to big contributers like FP and TB. If we focus on that site alone the gap is MUCH smaller, about .1%. I wouldn't mind seeing some more stats from FTP as it is the probably the most played site after PokerStars. At this time it's deviation is high by about 4.35% but I suspect that given it's reputation, with a sufficienlty large sample it would approach that of PokerStars.

Wow, that's the most I've written in like a year. Time for some Data.

 

[Four Dogs]

The results so far.

 

[Debi]

Pokerstars:
12,359 hands
AA - 52 times
56 expected

Fulltilt - 1724 hands
AA - 7 times
7.8 expected

 

[Four Dogs]

Ah, very nice. Thx Dax.
Ooh! That one hurt our cause a bit. lol

 

[Debi]

Ah, very nice. Thx Dax.
Ooh! That one hurt our cause a bit. lol

I knew it was rigged against me lol.

 

[royalburrito24]

This experiment seems like it would be very fun to participate in. However, I do not have PT or PO. I would like to purchase PT when I begin to actually show a big enough profit in poker. Sigh.