Minimal observed hands to identify a loose player
 

I wondered how many hands pokeroffice/tracker has to log of a specific player to tell for sure, if he is loose or not.

I've calculated a little chart for that. On the x-axis, there are the observed hands and on the y-axis, there is the observed percentage of seeing a player on the flop.
A loose player is defined as being more than 30% on the flop.
If the observed percentage is above the red or green line, the player is a loose player with a probability of 95% resp. 99%.
(The line wobbling is caused by low statistics using binomial distributions).
http://img516.imageshack.us/img516/2653/looseplvm1.png

Derived rules of thumb:
after 20 hands => min. 0.6
after 40 hands => min. 0.5
after 100 hands => min. 0.4

Re: Minimal observed hands to identify a loose player
 

This is a pretty slick analysis Python -- thanks.

I'm often frustrated when people say <X hands is a meaningless read. This gives some evidence that, for example, someone who's seen 1/2 the flops in your first orbit at a (10 handed) table has about a 95% chance of being a +30 VPIP. This helps validate my intuition.

Can you do the same thing with aggression?

Re: Minimal observed hands to identify a loose player
 

You can determine it in as little as one hand, depending on what he shows down & what circumstances led him to the river. To give an example, if I post my BB new to a table, a total unknown limps from UTG, and for whatever reason I see him show down K4o...he's loose until proven otherwise.

PT stats are pretty useless over tiny sample sizes. But traditional reads can & often do validate what the stats are suggesting.

Re: Minimal observed hands to identify a loose player
 

Python, its interesting you bring this up. Shape up poster's post in the SSSH forum actually goes into some details on defining new opponents in a short amount of time. If you can skip through some of the noise in the thread, you will find two excellent posts by Stellarwind and Leader. Any other micros players reading this, I think its worth your time to read the thread.

Re: Minimal observed hands to identify a loose player
 

I think you can get useful reads once a villain plays about 3 hands. Not folds, but PLAYS. If he limps in EP and shows down - that's a read. If he limps 3 out of 3, it's a read. If he plays 6 hands out of 10 in an orbit, file him under loose category - that's enough. What did he show down and how did he play?
If he raises 4 out of 5 - that's a maniac. 2-3 orbits are actually plenty, if you are paying attention.

PT stats are far away from converging, but you can surely make first impressions quickly and refine them as hands fly by.