by Schro » Thu May 26, 2005 12:36 pm
So is it the concensus here that AF would be better calculated as
(Bets + Raises) / (Calls + Folds) ?
I've written this SQL query for how PT calculates it (only calls), but Calls + Folds does seem better.
However, it would seem to lower the numbers across the board that we're used to seeing. I'd have to get used to the new numbers and where the natural breaks are.
Here is my AF chart for NL taken from Party data (25/50/100 tables). I decided to break it down into 5 categories, each representing about 20% of the population. The 2nd column is the number of hands a row represents, the third is the percentage of hands a row represents, and the last column is just some names I came up with. Lastly, to come up with these numbers, I limited the dataset to only those folks who had seen the flop at least 20 times. (edit: sorry if below is bit hard to read)
AF
<1 415 20% Meek
1-1.44 438 21% Unsure
1.45-1.99 436 21% Average
2-2.9 436 21% Bold
3+ 394 19% Maniac
I've been using the Autorate rules from the articles on this site, which create a break at 1.2. From my data <1.2 is 26% of hands and >=1.2 is 74%. Of course this is all post-flop, and not broken down by street. However, 1.2 seems a bit low, as you're saying that 74% of the general poker population is aggressive. That said, I recognize that the 15 cap of autolimit rules in PT is the main limitation.
Seeing the above, I'm tempted to change the break from 1.2 to 1.45, which would categorize 40% of people as unaggressive and 60% as aggressive. 1.6 is another candidate as that is the 50% mark. However, I might redo the entire thing if it's determined that PokerTracker didn't use the best calcualtion of AF by not including Folds.
What do you guys think?
Cheers,
Schro