View Single Post
  #12  
Old 07-10-2005, 12:02 PM
aargh57 aargh57 is offline
Junior Member
 
Join Date: Jun 2004
Posts: 12
Default Re: Calling all Sample Size Nazi\'s

[ QUOTE ]
After n hands, a 95% confidence interval for your win rate in BB/100 is roughly results +- 300/squareroot(n). So, if you win at a rate of 3 BB/100 for 10k hands, your 95% confidence interval would be 0-6 BB/100. To shrink the interval by a factor of 2, you need to increase the number of hands by a factor of 4.

[/ QUOTE ]

So then say, for example, I have a 2.0 BB/100 after 40,000 hands my 95% confidence interval would be:

2.0 +/- 300/(squareroot of 40000) =
2.0 +/- 300/200 =
2.0 +/- 1.5 =
between 0.5 and 3.5

In other words I am 95% sure that my "true win rate" would lie between 0.5 and 3.5.

I just wanted to see if I'm calculating this right.

BTW, I am assuming that this formula is in a book somewhere. If so, which one?
Reply With Quote