A Question on Sample Sizes

[JimRivett]

I have seen posts where an individual indicates that he has won so many bb per 100 hands (and other poker tracker stats), while playing say 2000 hands, and some responses tend to inform him that his sample size is perhaps too small.

Yet in other posts, individuals indicate poker tracker stats on their opponents while only having as little as 20 hands against them.

I suppose my question is when does the sample size become relevent? At what point do you think the sample size is reasonable to make the type of decisions to label your opponents?

I'm quite new to having tools such as poker tracker at my disposal, all of my play up until a few months ago was live, and I see that some of you use the stats on your opponents to build profiles of them.

Any input would be appreciated.

Jim

[Evan]

Stats like VPIP and PFR converge a lot faster than BB/100. Any stats you can get on opponents will help, obviously more hands are better. Common sense will be good enough to estimate reliability.