Some stat discussion

[Leader]

OP,
http://forumserver.twoplustwo.com/showfl...=0#Post12464774

[ QUOTE ]
he's not that aggressive

[/ QUOTE ]

It's impossible to know this without knowing WtSD and how many hands this is over. AF is useless over less then 5K (conservatively could be 10-20K) because of how slowly it converges. Without WtSD AF has no context. 1.14AF and 34 WtSD is passive. Same AF with 50 WtSD and 40 VPiP is agro.

[rzk]

[ QUOTE ]
[ QUOTE ]
he's not that aggressive

[/ QUOTE ]

It's impossible to know this without knowing WtSD and how many hands this is over. AF is useless over less then 5K (conservatively could be 10-20K) because of how slowly it converges. Without WtSD AF has no context. 1.14AF and 34 WtSD is passive. Same AF with 50 WtSD and 40 VPiP is agro.

[/ QUOTE ]

i've estimated that the standard deviation for flop AF is approximately 0.6 - 0.7 for a 100 hands sample, 0.2 for a 1000 hand sample, and 0.06 - 0.07 for a 10K hands sample. the standard deviation for the total AF should be around sqrt(2) or sqrt(3) smaller, so around 0.4 for 100 hands etc.

as for the effect of wtsd on aggression, i actually wrote up a little calculation on that for my 100th post. unfortunately i can't find it for some reason as the search only gives me my 200 latest posts. the bottom line was that the AF for a wtsd 45 guy should be adjusted by about 10% compared to a wtsd 35 guy.

[Leader]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
he's not that aggressive

[/ QUOTE ]

It's impossible to know this without knowing WtSD and how many hands this is over. AF is useless over less then 5K (conservatively could be 10-20K) because of how slowly it converges. Without WtSD AF has no context. 1.14AF and 34 WtSD is passive. Same AF with 50 WtSD and 40 VPiP is agro.

[/ QUOTE ]

i've estimated that the standard deviation for flop AF is approximately 0.6 - 0.7 for a 100 hands sample, 0.2 for a 1000 hand sample, and 0.06 - 0.07 for a 10K hands sample. the standard deviation for the total AF should be around sqrt(2) or sqrt(3) smaller, so around 0.4 for 100 hands etc.

[/ QUOTE ]

How exactly? I have plenty of statistics experience so don't spare the details. As far as I knew this would have to be arrived at through some fairly complex simulation.

[ QUOTE ]
as for the effect of wtsd on aggression, i actually wrote up a little calculation on that for my 100th post. unfortunately i can't find it for some reason as the search only gives me my 200 latest posts. the bottom line was that the AF for a wtsd 45 guy should be adjusted by about 10% compared to a wtsd 35 guy.

[/ QUOTE ]

http://forumserver.twoplustwo.com/showfl...ue#Post11035889

Frankly, that thread is confusing. I don't follow the logic. Plus you don't say how exactly you came to the underlying assumptions.

[rzk]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
he's not that aggressive

[/ QUOTE ]

It's impossible to know this without knowing WtSD and how many hands this is over. AF is useless over less then 5K (conservatively could be 10-20K) because of how slowly it converges. Without WtSD AF has no context. 1.14AF and 34 WtSD is passive. Same AF with 50 WtSD and 40 VPiP is agro.

[/ QUOTE ]

i've estimated that the standard deviation for flop AF is approximately 0.6 - 0.7 for a 100 hands sample, 0.2 for a 1000 hand sample, and 0.06 - 0.07 for a 10K hands sample. the standard deviation for the total AF should be around sqrt(2) or sqrt(3) smaller, so around 0.4 for 100 hands etc.

[/ QUOTE ]

How exactly? I have plenty of statistics experience so don't spare the details. As far as I knew this would have to be arrived at through some fairly complex simulation.

[/ QUOTE ]

yes, i think remember you saying that you are doing a ph.d. in stats, so you'll probably easily find holes in my calculation [img]/images/graemlins/smile.gif[/img] but anyway, here's how i reasoned. this is of course an approximation, but let's say the probability of a call on the flop is 30% and the probability of a bet or raise is also 30% (more precise numbers could probably be obtained from pt or pahud but they will do for our approximation, changing them by 10% would result in a change of around 10% in our final result). In 100 flops seen, the error (i.e. standard deviation) in these frequencies will be around 5%. so the error of the ratio will be around sqrt((5/30)^2+(5/30)^2)*(30/30)=5/30*sqrt(2)=0.24. 100 flops seen means approximately 250 hands for a vpip 40 guy. so in 100 hands the sd is 0.24*sqrt(250/100)~0.4. then i just increased it in an ad hoc way to 0.6 to account for the effects of various approximations (like the fact that frequencies of calls and bets/raises are not independent variables, for samples with frequencies of calls higher than average the frequencies of bets/raises are somewhat lower than average, which should increase the sd of the ratio slightly).

of course a better, but slightly longer way of doing is is to examine data from pt for small samples of hands and see how AF fluctuates. or better yet, write an sql query.

as for my 100th post calculation, what assumptions are you wondering about? if you mean the relationship between wtsd and fold to bet, i got it simply by looking at pahud stats various players. these variables are strongly correlated.

[NinaWilliams]

I hate when threads degenerate into debates about interpretations of poker tracker stats. Basically my read is that villain is clearly on the passive side, my sample size is about 300-400 hands.

[vmacosta]

[ QUOTE ]
I love it when smart ppl help me figure out how to interpret poker tracker stats.

[/ QUOTE ]

FYP

[Leader]

[ QUOTE ]
the probability of a bet or raise is also 30%

[/ QUOTE ]

I don't think you can do that. Bets and raises are distinct random variables. For example, the SD of two binomial with mean .15 is about 9% greater then the SD of one with mean .01 and one with mean .29 after 10K hands.

[ QUOTE ]
sqrt((5/30)^2+(5/30)^2)*(30/30)

[/ QUOTE ]

I see where the numbers come from but I'm not sure why your doing it. I guess you're saying that the SD of one binomial RV divided by another is the sqrt of (variance of the first plus the second) times the ratio of one to the other. Conceptually I have some concerns about this. Could you link me to something that talks about that?

[ QUOTE ]
of course a better, but slightly longer way of doing is is to examine data from pt for small samples of hands and see how AF fluctuates. or better yet, write an sql query.

[/ QUOTE ]

Well different AF are going to converge at different rates so this would be kind of hard. It's not impossible or statistically unsound though.

[ QUOTE ]
as for my 100th post calculation, what assumptions are you wondering about? if you mean the relationship between wtsd and fold to bet, i got it simply by looking at pahud stats various players. these variables are strongly correlated.

[/ QUOTE ]

While I believe you looked at it and saw a good correlation, these things have some complex dynamics. It's not as if fold to bet=1-WtSD.

PS. While I have some of the traits of a PhD student (I've done grad level research; I'm taking a grad level class on financial time series), I'm only applying to be a PhD student.

[Hobbs.]

[ QUOTE ]
I'm only applying to be a PhD student.

[/ QUOTE ]
don't do it, it's a trap!

[Leader]

[ QUOTE ]
[ QUOTE ]
I'm only applying to be a PhD student.

[/ QUOTE ]
don't do it, it's a trap!

[/ QUOTE ]

Student loans + no idea what to do with my degree + uncertain poker future = yay more school. Plus I like research and want the hell out of New England. I can always bail after a masters too if it's not for me. All in all, there have got to be worse fates.

[Hobbs.]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
I'm only applying to be a PhD student.

[/ QUOTE ]
don't do it, it's a trap!

[/ QUOTE ]

Student loans + no idea what to do with my degree + uncertain poker future = yay more school. Plus I like research and want the hell out of New England. I can always bail after a masters too if it's not for me. All in all, there have got to be worse fates.

[/ QUOTE ]
I'm just joking. I bailed with a masters this last june