Quality of card shuffling on online poker sites
Gordon
I play poker online and sometimes I wonder myself if the shuffle is a really
randomly done. I can make some demonstration based on opinions like "what
would be the interest of a site to make biased shuffling?" but the questioon
I ask myself is : would it be possible _to prove_ that the shuffling is good
and really random knowing that we see for each hand (texas holdem poker):
- a minimum of 2 cards (the one that are dealt to you)
- maybe 3 to 5 more cards (the community cards)
- maybe cards of other players
thank you.
G.
Stan Devia
If you keep a history of the first card dealt to you, you may be able
to use the Binomial distribution to test the fairness of the draw.
For example the probability of getting exactly 3 Ace of Spades from 40
draws, given that P(Ace of Spades)=1/52 after each shuffle, is given
by B(3,40;1/52)=0.0342 (i.e just over 3%). Anything less than, say, 5%
MAY indicate an non-random shuffle system.
However the bigger the sample the better. For example P(exactly 3 ace
of spades from 50 draws)=0.0560, indicating a fair shuffle.
lsch...@d.umn.edu
> Hi,
>
> I play poker online and sometimes I wonder myself if the shuffle is a really
> randomly done. I can make some demonstration based on opinions like "what
> would be the interest of a site to make biased shuffling?" but the questioon
> I ask myself is : would it be possible _to prove_ that the shuffling is good
> and really random knowing that we see for each hand (texas holdem poker):
You might find this article interesting. It details the flaws that
were found in an analysis of the Poker Planet shuffling algorithm and
a proof-of-concept exploit that the authors presented to the company
the mde the shuffling algorithm.
http://www.cigital.com/papers/download/developer_gambling.php
It's a good illustration on how even simple algorithms must be
carefully designed to ensure that they follow desired statistics.
-Lucas