Kooks and Computer Coders needed for experiment.

[lorinda]

Okay,

I believe that the distribution thing can be solved, but I need the help of kooks and computer coders.

The experiment runs like this.

Could the Kooks please post which things they would expect to be rigged, Im not talking "The number of Aces on the flop" or "The number of times a player is dealt Kings" because all but the toughest of conspiracy experts have seen enough evidence to the contrary of this...surely.

However, what we need is a list of distribution occurences to look for.
A couple might be Kings vs Aces dealt, set vs set and top two vs flush.
However, I would like the sceptics to provide examples.

What I hope someone can code (I hope it's not a very long exercise) is a poker program that shows us how often these occurences should happen, and then maybe someone could post how often they DO happen.

Unfortunately I have no coding skills, no tracking ability, and If it is ME that presents the test cases, then no credibility will be given to the case.

Any chance we could organise this?

Lori

[XlgJoe]

Lorinda

I really doubt that the larger sites are using crooked deals. But, if they were I would do it just as Terry mentioned in his post somewhere below. Just deal a true random shuffle. But on a random basis give the player that has lost the most money, the winning hand. The scond worst player the next best hand and so on.

The winning hand could be 32o and folded(or played by some maniac. The point is the distributions would be random, but the poor player would get some preferential treatment. And the deal wouldn't be manipulated every hand just some percentage from 10-25% of the time. I doubt you could determine that type of rigging.


Like I said I don't believe this is happening just saying how it could.

[Miah]

How would you account for the looseness of the game? Surely a very loose game is going to have a lot of set over set/flush over flush beats, as well as people sucking out their cards on the end.

I'm not trying to bring this to your attention as I believe you already are well aware of this factor. Rather I am asking how would you account for this in such an analysis?

[lorinda]

How would you account for the looseness of the game

I had totally overlooked this aspect, I'll get back to you when I have thought it through more carefully.

I am still sure it is a viable experiment, but I need to work out what I am talking about first /forums/images/icons/blush.gif

Lori

[Bubmack]

I think a good test would be a comparison of all-in bets and calls when the second best hand before the all in bet wins. Although I dont know what to compare it to since the hand that is a dog will have varying chances to win. Like a four flush is more likely to beat a set than KK is likely to beat AA.

I would still like to know the % though.

Bubs

[zooey]

Here's the main problem: These are all very easy to calculate and test for if you have data that is independant of strategy: i.e. you get to see everybody's hole cards and pretend that they went to the river every time.

But in real life play, where you get to only see shown down hands or possibly one player's and the shown down hands, you need to model both the probabilities (easy) and the strategy used to determine whether there's a show down. (very hard)

If the kooks would make a strong precise claim, i.e. set of set is happening with a frequency of 5 standard deviations greater than expected, then ok, you can make some very rough assumptions and "prove" that it is not in fact outside 5 S.D.s. But proving that eveything is right (to within the measurement error of your sample size) is dauntingly difficult.

(If you could model playing decisions accurately enough to calculate this, I'm guessing the work would be much more valuable as a poker player than as a validity test. )

And as I have yet to see a kook with a strong enough mathematical bent to appreciate my arguments here, I think your project is laudable but DOA. /forums/images/icons/frown.gif

BUT, if you want, I'll entertain you with any probability calc, like set over set, you need, they are fun little puzzles.

Just my opinion, btw, if the project goes, cool. /forums/images/icons/smile.gif

Best,

zooey

P.S. Great play in the last Zoo tourney, btw. Sure you caught some luck, but I really liked all your decisions that I saw.

[Bubmack]

</font><blockquote><font class="small">In reply to:</font><hr />
with a frequency of 5 standard deviations greater than expected

[/ QUOTE ]

Yikes - are we looking for statistical certaintity or a reason to just avoid a site. 3 STD are good enough for me.

Bubs

[zooey]

lol. it was just example. I doubt we'll ever see someone stick their neck out and make a testable claim.

The reason I picked 5 SDs is it would not be unusual to take a data set, and find something in it that's 3 S.D.s off. 3 SDs is like 1:1000, so if you test for 1000 things, (and maybe that's what your brain, pattern finder/creator extrodinaire, is doing as it skims a data set looking for something interesting) you would almost expect to find something wacky. Finding something 5 SDs off would be a lot more challenging.

But if one makes a claim independant of the data, and then we go get some random data and test it, then sure, a repeatable 3 SD claim is quite damning.

Best,

zooey

[crazy canuck]

The only thing pretty much left to test for are the independence of hole cards (since it has been shown that board cards are random). It really pisses me off people who really belive that a site is rigged don't test for this....it's not that hard.

Another theory says that everything is roughly random but good players get a few hands that they lose a lot of money on. This could be tested by weighing the occurance of a hand by the pot size...this would require a lot of numbe of hands tho.

[BBill]

I think this would be an interesting experiment. The thought it brings to mind for me concerns:
But in real life play, where you get to only see shown down hands or possibly one player's and the shown down hands, you need to model both the probabilities (easy) and the strategy used to determine whether there's a show down. (very hard)

Have these statistics / distribution occurences ever been perfomed in live games ? I am no expert in this but I am assuming that statistics of distribution for live play have been determined through mathmatical expectation of what should happen with a deck of 52 cards dealt to 10 players. We accept these results as we certainly can not prove they are false unless we analyze 1,000,000+ hands dealt live.
The nice thing about playing online is that the precise results can be collected with minimal error.

bbill

[GrannyMae]

but I need to work out what I am talking about first

if i worried about stoopid sh*t like this, i would have only 4 posts.

http://e4u.consoleradar.com/crazy/146.gif

[polarbear]

Here's something you can test:

The probability of flopping a set if you have a pocket pair, and there was a flop. You only need to look at your own hands.

Sure, there's the small effect of the deck being ace rich if everyone folds and you have AA, etc, but the results should be fairly accurate.

For some reason it seems like there's a better chance of flopping a set online than in live play, why, I have no idea.

[mongeron]

</font><blockquote><font class="small">In reply to:</font><hr />
I think a good test would be a comparison of all-in bets and calls when the second best hand before the all in bet wins.

[/ QUOTE ]

Actually a this kind of test has been made, I read yesterday about it at RGP. The analysis was done by Steve Brecher, and the subject of the mail starts like this: "PokerStars tournament all-in pot equities v".

According to that post, the all-in situations went as they should.

- mongeron

[Miah]

You play twice as many hands per hour online as you do in a casino, therefore you flop sets aproximately 2x more than you would expect to at a B&amp;M.

Also, if you are like me and play 22 UTG while chanting to yourself "implied odds implied odds" this could increase it as well....

[HavanaBanana]

Some hands are more likely to win from the getgo, if they did random deals,with a marked account getting fewer winning hands, with the assigned player to lose more than his fair share this would show up in the starting hands he would get.

Take a suspected marked account, bring 9 2+2 friends, play 2c-4c holdem with noone ever folding.
the marked account should lose in 8 hours or play right?

or, if you are REALLY sure it is fixed so the winners will become losers, then make new account , play from a new computer, buy in at 30-60 and play your little heart out to the river!
Even better, make an initial depost and lose it on purpose, then redeposit and play 30-60 you cant lose!

Don't come to me to borrow money later though.

You are the Weakest Link.... Goodbye.

[XlgJoe]

As i stated I don't believe the major sites are fixed.

All I was trying to say was that I don't believe doing a tracking of board flops/flush over flush/etc will indicate anything. To many poker programs keeping stats to get away with manipulating the deal.

Most people believe Propoker.com(not positive on site name) is rigged. Could this study prove them rigged? It would be nice if it could, but I doubt it will be able.

[Carl_William]

Hi Lori,

You posted:

"Okay,

I believe that the distribution thing can be solved, but I need the help of kooks and computer coders.

The experiment runs like this.

Could the Kooks please post which things they would expect to be rigged, Im not talking "The number of Aces on the flop" or "The number of times a player is dealt Kings" because all but the toughest of conspiracy experts have seen enough evidence to the contrary of this...surely. (and so on).... "

Dear Lori,

What you are talking about is about almost “The Impossible Dream – The Lady of Mancha.” It is possible – maybe, but not probable for many reasons. The ROI (Return on Investment “time &amp; brains”) would be essentially nil except maybe for a very few exceptional poker players – how they would use this to make a profit is subject to who-knows-what. It would be very difficult to get a fair and accurate set (population) of data to analyze. Granted, it would be fun to know all of this stuff, but what is analyzed is similar to a “house-of-cards.” A “puff of air” or a “vibration of the table” or a treak of the casino software program and everything changes. This task could be accomplished by a set of expert programmers within an Internet Casino Site if they wanted to pay for it – but would they share the unbiased results of the study with others – I don’t think so…. Some comments….
(Lori -- you know all of the below:)
For a flop game and using a data tracker, each stage of the flop, turn, river, and total set of visible board cards would be easy to analyze for a suitable large number of trials (deals). Using counting techniques, the number of all pertinent card sets could be compared with theory and determined if they are within acceptable tolerances. For example (for five board cards): for 10 times Combin(52,5) deals there should be approximately 10 times 3744 full houses plus-or-minus some acceptable statistical deviation. Analyses like this are easily done. Analyzing the private down-cards in a holdem game for an online Internet casino is no easy chore….

Can access to all the private cards in an Internet casino holdem game be collected? Assuming yes, then it would be easy to analyze the data and determine if all pertinent things are within the theoretical acceptable tolerances – no big deal. But….

I don’t know how any outsider could ever get good data for the Internet casino player’s private cards – this includes all hands that are dealt for the deal for all deals recorded.

I hope I'm not boring the Zoo population (visitors &amp; those who live there). But I will make more comments in the near future which I feel might be interesting to some....

Carl

[Terry]

ProPoker is "different." If you watch the games for 20 minutes and then buy in, you need ... ??? ... something, but I don't know what.