[zooey]

actually, i dont think so.

When you're meauring things that are in the coin flip arena, i.e. +-1 (saw the flop or not, went to showdown or not, etc) The standard deviation is on the order of ~1.

Therefore things converge as ~1/sqrt(hands), so you're pretty sure to be in the ball park after a few hundred hands.

Its when you're measuring rare events (how many times did someone have AA vs my KK on a AKx flop?) or non binary events (the amount you win is usually bigger than the amount you lose) that convergence takes longer.

Ive written a post or two on measurement error, and Bruce Z has written some better ones, if you're curious and want to search through the archives.

Best,

Zooey