[Octopus]

1. It takes more than 30 hands for VP$IP to converge. (Haven't you ever gone 3 orbits without playing a hand?) That said, it does converge rather quickly. For what we are using it for, 100 hands is probably enough, and even 30 hands will give a reasonable impression. (Loose players NEVER go 3 orbits without playing a hand.)

2. PFR% will also converge fairly quickly, for our purposes at least.

3. Aggression factor, on the other hand, can take a VERY long time to converge, even into fairly broad bins. Even over 1000 hand samples, mine ranges from about 1.0 to about 5.0. For a tight player, you might have only a few hands which actually had post-flop actions (other than check and fold) for every 100 hands played. Even if it does converge, a high number doesn't tell you if the guy is good or just never saw a flop not worth betting.

If I could find something better, I would replace aggression factor in the auto-rate. Unfortunately, nothing really jumps out. (How about WTSD%?)