Quote:
Originally Posted by PikachuDemolisher
There are a couple guys here id like to have read this with an open mind. Spadebidder and Josem and a couple others. I am no there to start trouble i just dont like being attacked instantly as im sure no one does. But these are my thoughts so please read this with an open mind and DISCUSS it if youd like.
Sure.
You've posted a lot of text, and I've got a little bit of a hangover this morning, so let me just cut to the key bits.
Quote:
I am making the assumption that everyone here agrees on the fact that
there is a skill element and not just chance.If that is a bad assumption
please let me know.
I think that's fair.
Quote:
I propose the following. If you agree that poker, specifically NLHE,
is a game with a skill component rather than chance only, then
HHs statistics alone is not enough to mine out possible
corruption. A game of chance alone I believe would have a different
approach to analyzing the data.
Well, it depends what time of 'corruption' you're looking for.
For example, the records of the hands were useful in proving that Pitbull Poker were dodgy. They were useful in proving the cheating at AP and UB.
I don't know if the original Planet Poker review needed hand histories - they just used the published algorithm of the site concerned I think.
There might also be insufficient evidence in the publicly available HHs for certain cheating detection. For example, there are currently a heap of people reviewing hand histories of 'stoxtrader' in the HSNL forum. I suspect that they'll find that there is insufficient evidence to prove collusion without the hole cards.
Quote:
So given this hypothesis I point to the Cigital study that shows that
of hands going to showdown, an equal amount of hands were won by the
best hand and an equal amount the best hand was folded before.
I will now make this assumption. This statistic is what is expected.
You have the statistics slightly wrong.
~75% of hands were won without showdown.
Of the remaining ~25%, half were won by the best 5-card poker hand out of all the hands dealt in. The other half, the best 5-card poker hand (as measured at showdown) had previously folded.
Quote:
*(However my initial gut reaction is that
possibly there is no way to determine what the expectation of this is.
Can anyone confirm or deny that?
Well, I think it might be theoretically possible, but that would require predicting how players behave. If you could model player decisions, you'd be able to then predict the results of player decisions. I don't think it is computationally feasible at the moment, but I suspect it might be theoretically possible in many years time.
Quote:
I beleive for this and 1 other reason
the study might be flawed or at least inconclusive.
I don't think the study is flawed because it wasn't possible to mathematically predict the results. I don't think that's even close to being a reason to consider it flawed.
If you have an experiment to test the colour of the sky, you don't need to mathematically prove anything. You just look at the sky and write down what colour it is.
The same thing is here. If you want to know how often players fold, you don't need to predict the outcome. You just look at the hand histories and figure out how often players fold. It's not rocket surgery
Quote:
1 they dont
examine the cards that didnt go to showdown and it is possible that
the best cards drove the players to make the choices that ended the hand
early.
But the survey did examine the cards that didn't go to showdown?
Quote:
Therefore there is a hidden aspect of the luck element tucked
away in this 75% of no showdown hands. We cannot examine that. This is
where my partial tree theory comes in.
I still don't understand this partial tree theory stuff.
Can you please show us a hypothetical complete 'tree' for one hand?
Quote:
And 2 this statistic may change
when better/worse players or even just differnet players are the sample.
Yep, probably it does.
Quote:
I still believe there is skill, I jsut dont think its quantifiable how
much is skill and how much is luck and would prefer to see a range
number here.)
I'd suggest you read the discussion about the survey:
http://forumserver.twoplustwo.com/28...needed-393876/
Quote:
Now back accepting that skill does exist. Lets consider an example.
A skilled player gets a good hand against a skilled player. Lets say AA.
He gets a read and decides to fold. We now must remove this hand from
the analyzing of hand histories because it never made it to showdown.
Why?
It wasn't removed from the Cigital survey; Such a hand isn't removed from
anyone's surveys.
Quote:
So how can you present the statistics for how this particular hand
performed over a large sample if not all instances are available to
analyze?
But it is available to analyse.
No one is excluding particular hand histories from their analysis. If they were, it'd be pretty flawed.
Quote:
The randomness of a softares deal is not limited to just
the hands that go to showdown, as we have seen this number can be
as low as 24%, and maybe even lower at times. So we are only seeing
24% plus 1 players hole cards as a representation of the overall deal.
Not 100% of the deal. I understand that you do not need 100% of hand
historis to study this. However of the X% of hand histories you do have,
how can you possibly analyze the WHOLE deal's ability to be random with
only 24% of players in a showdown + 1 players HCs? This is is what I call the partial tree.
Well, obviously you analyse what you have.
Sure, if you have a hand history that doesn't include hole cards, you obviously can't analyse hole cards.
Fortunately, if you play at PokerStars, they'll send you every single real money hand history you've ever played, including your hole cards. You can then perform whatever test you want on it.
On top of that, they can even provide an analysis of the distribution of your hole cards over the last month or so.