PokerEV Graphs

[Bavid Denyamine]

Anyone out there want to share their PokerEV graphs thanks to Phil's great program? (See thread: here.)


Here's my own:



Analysis: I run pretty good. -800BB difference between Total Showdown and Total Winnings = [img]/images/graemlins/frown.gif[/img]. +17BB/100 over 7k hands = [img]/images/graemlins/smile.gif[/img]

[Phil153]

Hey David (Bavid?),

At this stage I recommend being cautious intepreting results. PokerTracker has quite a few errors which I'm trying to map out, such as counting overbets wrongly and storing the same data differently on different sites. You don't see it in PokerTracker though because it simply queries total won - total bet. These problems (plus bugs, minor rake issues, etc) are the reasons I'm doing a limited pre beta.

The extreme nature of your graph leads me to believe it could be a program miscalculation (though I've certainly had verified runs this big, both hot and cold). I've also kind of neglected omaha high in my testing (I play PLO8 and holdem), mainly due to a lack of hand histories. If you could send me the hand histories I'll be able to verify the results and catch any errors.

[RoundTower]

cool idea for a program. I'm going to guess you can use this to find your winrate to within 3 ptBB after 10k hands or so. Is that too optimistic? Silent Acorns or someone please do the maths.

[Colonel Ingus]

Can someone point me to a good resource on "Skalansky bucks?" Or a good explanation would be good as well.

[Silent A]

[ QUOTE ]
cool idea for a program. I'm going to guess you can use this to find your winrate to within 3 ptBB after 10k hands or so. Is that too optimistic? Silent Acorns or someone please do the maths.

[/ QUOTE ]

Sorry, I have no idea where to start. It seems much too complicated.

For hands that go all-in before the river though you'll get a a winrate well within 1ptBB within 10K hands, easily (math insticts talking here).

I honestly don't think it's possible to use this kind of tool to evaluate your overall winrate. It will, however, give you a better idea of where your winnings come from and give you a rough indicator of what side of varience you've been on.

[RoundTower]

you can definitely evaluate your overall winrate with this more accurately than just by looking at your actual winrate and SD -- think about it. Just a question of how much more quickly.

[KhalynYohrk]

[ QUOTE ]
Can someone point me to a good resource on "Skalansky bucks?" Or a good explanation would be good as well.

[/ QUOTE ]

[Bavid Denyamine]

Sklansky Bucks = (% chance of your hand to win when $ goes in pot) * ($ that goes in)

The theory here is that if you are winning in terms of Sklansky Bucks, you'll win in on the long run because short run luck evens out. If you are constantly getting it in with the best of it, you'll have +Sklansky Bucks and +Real Bucks.

Assuming my graph is right, I'm +700BB Sklansky Bucks and +1200BB overall, which I'm assuming is about standard, but still seems like I'm running pretty well. You have to realize that spots where you're +EV to win, you'll be winning more real money than Sklansky Bucks on average.

[Troll_Inc]

[ QUOTE ]
cool idea for a program. I'm going to guess you can use this to find your winrate to within 3 ptBB after 10k hands or so. Is that too optimistic? Silent Acorns or someone please do the maths.

[/ QUOTE ]

I was going to write an article on it but still needed to do the maths. Cool that a program already does this..didn't know it existed because I was going to do it by hand.

There are several ways to approach the problem I think. You really need to make some assumptions on how you apply the concept of pot equity vs money going in (Sklansky Bucks).

But whatever assumptions you make you would have to essentially resample the data such that there was so much skew-kurtosis in the probability distribution.

I think the data will still be pretty widely distributed such that the variance will remain quite high and you won't be able to get a nice neat confidence interval. But my hope is the same as yours, that you can actually look at the winrate over shorter periods of time and have "more" confidence.

Lastly, there are definitely some factors that will remain elusive in this sort of analysis, i.e. so it won't be perfect.