Project - Compiled stats for Short Handed Games

[vmacosta]

This is a great idea and I will offer stats if I qualify.

To all the haters,
CmI described his methodology and layed out the results. Of course its flawed but unless you are gonna help or do something better why get in the way?

[sethypooh21]

Just looking through those threads, guess who said that as of 18 months ago:
[ QUOTE ]
These are my stats. I can barely beat .5/1.

[/ QUOTE ]

someone's come a long way baby

[CallMeIshmael]

[ QUOTE ]
What did you use here?

[/ QUOTE ]

I think for the pictures posted, I used mean 15 std 2.5 for winners, and mean 22 std 4 for losers

[ QUOTE ]
Did you take this from the binomial distribution?

[/ QUOTE ]

Nah, I didnt hardwire the std as a function of the mean of the stat, since it allowed more flexibility.

The actual code looked like this (edited slightly for readability):

statMean + statStdDev*sqrt(perHandConst/player(loop).numHands)*randn(1)


statMean is self explanatory

statStdDev is the std of the stat, which I think I set at 100 for the above, which is quite a bit higher than all of the stats that fit into a [0,100] type range are going to have (as you bring that down, the cost of including smaller DBs decreases)

perHandConst is a constant to adjust for the number of times per hand the stat is used. Which is 1/(prob of acting on stat in hand) (ie. VPIP would be 1, but stuff like river bet would be much bigger)

randn(1) is a random variable from a normal dist with u = 0, and st = 1.

[ QUOTE ]
Also, how did you write the program? Some kind of stat program or was it writen in a programing langauge.

[/ QUOTE ]

Matlab


[ QUOTE ]
What I meant to impart was that if you want to look at statistics to become a better player, you're better off looking at the stats of a clearly winning player as a whole and trying to derive why that players stats are the way they are rather then the average of many players.

[/ QUOTE ]


Dont get me wrong, i definitely agree with this, for the most part. But, for more beginning players, it seems like they might often think "well, Im different on a bunch of stats. Some by a bit, some by more than a bit. I wonder what that means?" And, they might not be able to figure it out.

I really do agree with most (maybe all, cant remember) of what seth has said in this thread. But learning how stats go together is upper level thought. And some newer players might be better off thinking "I show down way more than most 2p2ers, maybe I should post a few hands where I showdown with what I call marginal hands, and get feedback" than they are with thinking about how their WTSD fits in with other stats.

[Leader]

[ QUOTE ]
To all the haters,
CmI described his methodology and layed out the results. Of course its flawed but unless you are gonna help

[/ QUOTE ]

I am helping. If the methodlogy flawed to the point that the results don't really say something meanful then there's a problem. You can't just say, "the methodology has been laid out caveat emptor," because 95% of the people on this forum don't have near the knowledge of stats necessary to understand the effect the methodology has on the results.

[ QUOTE ]
why get in the way?

[/ QUOTE ]

There's no edict from me stopping this. You'll note the general lack of interest before I voiced my concerns. IIRC this idea has come up before to and obv. It didn't take off.

[Leader]

[ QUOTE ]
Just looking through those threads, guess who said that as of 18 months ago:
[ QUOTE ]
These are my stats. I can barely beat .5/1.

[/ QUOTE ]

someone's come a long way baby

[/ QUOTE ]

Agro/suicidal bankroll strategy. IIRC at one point he sat with his whole roll, like 8k, at 30/60.

[Leader]

I can't really argue with any of the technical choices you made. We're still have some disagreement on the theory, but this seems throughly outlined at this point.

Interestingly enough, I was at a colloquium today on using Bayesian analysis in predicting future performance of MLB players for use in fantasy sports. I spent quite a bit of time talking to the speaker, who happens to be working with me on another project, after the talk about how his method might be applicable to poker. Bayesian analysis from the little I've picked up works by taking past information and updating it with current information in an effort to better predict the future. So you could say, for example, that a players past play is the first 80% of the hands we have on them, and their current play is the most recent 20%. This would help eliminate the classic problem with poker stats of having stats on someone he has changed their play. We also talk some about using hierarchical structures but I only understand that superficially.

[tagWAG]

[ QUOTE ]
These will have to do you for now (read the threads all the way through).

http://forumserver.twoplustwo.com/showfl...rue#Post7070792

http://forumserver.twoplustwo.com/showfl...rue#Post6301267

http://forumserver.twoplustwo.com/showfl...rue#Post5473357

http://forumserver.twoplustwo.com/showfl...rue#Post4987345

http://forumserver.twoplustwo.com/showfl...rue#Post4762309

[/ QUOTE ]

Leader, thanks for these. Very helpful

[vmacosta]

[ QUOTE ]
[ QUOTE ]
These will have to do you for now (read the threads all the way through).

http://forumserver.twoplustwo.com/showfl...rue#Post7070792

http://forumserver.twoplustwo.com/showfl...rue#Post6301267

http://forumserver.twoplustwo.com/showfl...rue#Post5473357

http://forumserver.twoplustwo.com/showfl...rue#Post4987345

http://forumserver.twoplustwo.com/showfl...rue#Post4762309

[/ QUOTE ]

Leader, thanks for these. Very helpful

[/ QUOTE ]