Sample size requirement to estimate sustainable BB/100 & 80k hand stat
 

Hello fellow 2+2ers,

I was wondering how large of a hand sample size will I need to accurately estimate my BB/100 in the long run? Is 200-300k sufficient? or 400k +?

How much BB/100 can a top tier player achieve in the long run in the 3/6 limits? Is 1 bb/100 achievable over a large sample size(ie. 400k+)?

Here are my stats so far for 3/6 and 2/4.

http://img150.imageshack.us/img150/3126/80k1fp0.th.png
http://img523.imageshack.us/img523/1791/80k2pe1.th.png
http://img504.imageshack.us/img504/3365/80k3li6.th.png

I am aware that I am too tight in the BB and I have been working on that the past month and defending more.

Please feel free to evaluate my stats and any comments will be much appreciated. Thanks in advance.

Re: Sample size requirement to estimate sustainable BB/100 & 80k hand stat
 

There are many threads (in other forums at least) that deal with sample size, confidence intervals, etc.

10k hands --> ~ +/- 3BB/100. The size of the interval is proportional to the square root of sample size. 80k hands --> +/- ~ 1BB/100. For a better estimate you'll need your standard deviation. In poker tracker, under the sessions tab, click more detail, and post your SD/100 (it should be between 16BB and 19BB).

Re: Sample size requirement to estimate sustainable BB/100 & 80k hand
 

thx for your reply DrVan. Can you please direct me to some threads on this information. I couldn't find any threads in this section when I ran a search.

My SD/100 is $109.89 -- 19.055 BB

Re: Sample size requirement to estimate sustainable BB/100 & 80k hand
 

Search for Pocker Tracker and standart deviation.

But this is pure math. In real life thing go different. See high and Mid-Stakes forum for a threat of Baronzeus. He posted his chart of about 500k hands. During the first 200k hands he dropped like 1500 BB…

Yet he seems to be a very good and winning players

Re: Sample size requirement to estimate sustainable BB/100 & 80k hand
 

[ QUOTE ]
The size of the interval is proportional to the square root of sample size.

[/ QUOTE ]
I got this wrong. It should be proportional to the inverse of the square root of sample size.

Without a priori (knowledge you had about your WR prior to playing) the central limit theorem can be used to find a confidence interval. The standard deviation of the mean is inversely proportional to the square root of sample size. If it's 19.055 for 100 hands, for 80k hands it will be:

19.055*sqrt(100)/sqrt(80000) ~ 0.674

95% confidence is +/- 2SD.

So your interval would be your current WR +/- 1.35BB