Poker dealing algorithm

  

Poker dealing algorithm


Back to forum


Don Groves    (2010-03-31)
Poker dealing algorithm

Hello, Thibault -- It seems to me that unlikely card combinations are dealt here more often than probability would dictate. This may just be a feeling on my part that has no basis, but I'd like to ask which algorithm you use to deal the cards and how well has it been tested.


Thibault de Vassal    (2010-03-31 09:09:20)
Poker dealing algorithm

In terms of probabilities, everything that is unlikely to happen is likely to happen one day :)

Many hands have been generated so far, so I'm not astonished if some kind of repetitions occured already.

I created the random-like algorithm and I doubt there's something wrong in its "random" part. If needed I can prove that no card is dealt out of the algorithm, but I'll have to change it then. Anyway, if you wants to make statistics, you're welcome :)


Don Groves    (2010-03-31 09:19:17)
Poker dealing algorithm

Yes, everything will happen eventually, but some unlikely repetitions have already happened many times. At least two royal flushes for example. Also, there seem to be too many cases of the flop being all of the same suit. I have seen this several times in only the past few days.

I don't understand what you mean by "prove that no card is dealt out of the algorithm, but I'll have to change it..."


Thibault de Vassal    (2010-03-31 10:01:23)
Poker dealing algorithm

I mean that there is no card generated 'randomly' [without any control] like on many poker sites that can e.g. give better cards to super players and say "it was just luck"). I can prove that all cards come from the same algorithm, but of course if I reveal it, I'll have to change at least some parameters of the algorithm for the next games.

Anyway I can make a study and generate billions of hands (would it be enough, I'm not even sure) automatically to see accurate statistics but really, I would be surprised to see any problem.


Svante Carl von Erichsen    (2010-03-31 19:42:53)
Poker dealing algorithm

"..., but of course if I reveal it, I'll have to change at least some parameters of the algorithm for the next games."

I don't get this. There is a well known algorithm for shuffling cards, the Fischer-Yates-shuffle. All you need is an unbiased random number generator. There is no need for secrecy.

If you use anything else, it is most likely wrong.


Thibault de Vassal    (2010-03-31 20:24:34)
Poker dealing algorithm

It seems that I like to do everything by myself :)

The main difference (if I understood what I just read) is that this algorithm (e.g. Durstenfeld's algorithm) uses a different randomly generated number for each hand, so it is theorically possible to change the cards because there is no mathematical way to prove the cheating. With my algorithm, cards are pseudo-randomly generated but it is also possible to verify every card of every hand. And after all, even random functions are pseudo-random. (I hope that pseudo means the same in french & english)

Anyway, I'm curious and if someone knows a statistical way to know if an algorithm is good or not, I'll apply it to my method.


Don Groves    (2010-04-01 02:04:45)
Poker dealing algorithm

Google gives many results for testing random number generators. Why not just use Fischer-Yates? Here is a site that gives that algorithm in Java Script: http://sedition.com/perl/javascript-fy.html

And, yes, "psuedo" means the same in English. A true random number "generator" is not possible, of course.


Ilmars Cirulis    (2010-04-01 11:35:44)
Poker dealing algorithm

it is possible to use noise from "outer space"

as it's done in random.org


Don Groves    (2010-04-02 08:46:45)
Poker dealing algorithm

True. But then it's not an algorithm either ;-)