Quick poll about sample size

[Tosh]

Ok I have my opinion here , but I just wonder what the general idea is;

[Bob T.]

SQRT(17**2*1000)/1000 = .537

So after 100000 hands, you are 68 percent certain that your winrate is within a half big bet of what you have experienced.

I would guess that you could be confident in your winrate with a smaller sample, if for instance, you knew several other players, who played similarly to you, and alse were experiencing similar winrates to you. That way, we could combine samples, and after 50K hands, each we might feel more confident of our winrate. If there were a whole bunch of us, and we all combined our samples after 50k, then we would be fairly certain that our winrate was accurate.

Hey, that's kinda what we did here, isn't it [img]/images/graemlins/grin.gif[/img]?

[Shillx]

You can have 95% confidence in your winrate after any number of hands. Obviously the more hands you play, the more you can narrow it down. The formula for a 95% confidence interval is:

Mean +/- 1.96*s/root(n)

s: standard deviation/100 hands
n: # of hands played (in hundreds)

Brad

[Tosh]

One time bump.