CT11
11-03-2004, 05:54 PM
I've seen a lot of posts out there with people asking how big of a sample is needed, what they should expect their swings to be and if their win rate of XX proves anything.
Out of curiosity I decided to make a graph of a theoretical player with a true win rate of 2.00BB/100 and a standard deviation of 16BB/100. The graph shows where his BR ought to be 68,95, and 99 percent of the time of a sample size of 0-50k hands. I believe these assumptions to be reasonable for party .5/1.
What this should prove to people is that in the long run you do actually win and your current win rate may be rather far from your true win rate even for a rather large sample.
I just thought I'd post thins because I found it interesting.
http://img36.exs.cx/img36/9027/BR.gif
~CT
Out of curiosity I decided to make a graph of a theoretical player with a true win rate of 2.00BB/100 and a standard deviation of 16BB/100. The graph shows where his BR ought to be 68,95, and 99 percent of the time of a sample size of 0-50k hands. I believe these assumptions to be reasonable for party .5/1.
What this should prove to people is that in the long run you do actually win and your current win rate may be rather far from your true win rate even for a rather large sample.
I just thought I'd post thins because I found it interesting.
http://img36.exs.cx/img36/9027/BR.gif
~CT