Originally Posted by Stu_Ungar
One thing the artical pointed out was that AF was a stat which took a lot of hands before it became meaningful.
The example given was an opponent who hit trips on the flop and value bet every street. His AF would be extremely high and would take a fair few hands to settle down again. It could be very misleading to assume the AF tendances of an opponent based on AF with a low sample of hands.
This is true but doesn't triple value betting tell us something? Many people in this spot would slowplay and c/c a lot.
But you're right, it takes a little longer to converge but it's the information we have. If we see 9 in reality that usually means it's most likely between 6 and 12 (actually no because AF is a ratio, not linear, but basically the point is there's a range). So if we see a high AF, it means it's likely he's aggressive. If we see a low AF, it's likely he's passive. Sure we'll be wrong sometimes but it's just like any other spot in poker. When we don't have odds
we fold even if it's possible we would hit. The more hands we have, the more accurate the stat is and the more weight we can give it, and basically what you said is right, the vpip/pfr is going to converge the fastest and are thus the most useful stats to use on unknowns.