villians w/ high Agr Freq & low Agr Factor, and vice versa

What does it say about a villain if he has a high Aggression Frequency but a low Aggression Factor? Ditto w/ the opposite -- a low Agr Freq and a high Agr Factor? I never quite know what to make of them or what to expect from them. What does it mean (not technically, strategically I mean)? How would you play against them?

I just dug around in my database and here are a two common examples (the accompanying VPIP & PFR #'s seems to correlate often w/ each type, btw):

Villain A: VPIP 35 / PFR 20 / Agr Freq 65 / Agr Factor 2.2.
Villain B: VPIP 15 / PFR 10 / Agr Freq 30 / Agr Factor 5.0.

Btw, for the newbies out there (like me!):
Arg Freq = (bets + raises) / (all bets + raises + calls + folds).
Agr Factor = (bets + raises) / call.

Originally Posted by rowhousepd

Btw, for the newbies out there (like me!):
Arg Freq = (bets + raises) / (all bets + raises + calls + folds).
Agr Factor = (bets + raises) / call.

i think you can find some stuff in there

Originally Posted by Imthenewfish

i think you can find some stuff in there

I appreciate making fun of stuff more than the average FTR'er, but stop posting dumb shit in this forum please.

It's easier to demonstrate this with a series of examples than try to explain the why to begin with.

Suppose you have 5 bets, 5 raises, 5 calls, and 5 folds.

Your AFq is 10/20 = 50%
Your AF is 10/5 = 2.0

Now let's add more folds and say it's 5 bets, 5 raises, 5 calls, and 10 folds.

Your AFq is 10/25 = 40%
Your AF is 10/5 = 2.0

Now let's add more bets and say it's 10 bets, 5 raises, 5 calls, and 10 folds.

Your AFq is 15/30 = 50%
Your AF is 15/5 = 3.0

So if you add bets/raises AND folds, AFq remains stable while AF goes up. Similarly, if you decrease bets/raises AND folds, AFq remains stable while AF goes down. The first describes tagg behavior fairly well, and the second describes loose/passive behavior pretty well, which is what you're seeing with the 15/10 and the 35/20 there.

A lag
b tag

I just added AFq to my stats and was trying to understand the difference in the two and found this old thread and was playing around with some of that math stuff.

Since AFq=(b+r)/(b+r+call+fold) we can say that (b+r+call+fold) is "all hands" (not including checking)

And AF=(b+r)/call or (b+r)=AF(call)

If we sub this into above equation then

AFq=AF(call)/(all hands) and do some of that algebra stuff then

AFq/AF=(call)/(all hands) which is the percent that villian calls. So if you have these two stats you can find out how often villian calls by just dividing AFq by AF

So if villian is
Villain A: VPIP 35 / PFR 20 / Agr Freq 65 / Agr Factor 2.2.
Villain B: VPIP 15 / PFR 10 / Agr Freq 30 / Agr Factor 5.0.

Villian A calls about 30% of the time-passive
Villian B calls about 6% of the time-aggro

True?

in before the rush to build a custom HUD stat:
(Aggression Frequecy)/(Aggression Factor)

interesting thread

Originally Posted by daven

in before the rush to build a custom HUD stat:
(Aggression Frequecy)/(Aggression Factor)

interesting thread

That would be the station frequency

since:
Arg Freq = (bets + raises) / (all bets + raises + calls + folds).
Agr Factor = (bets + raises) / call.

Arg Freq / Agr Factor = [(bets + raises) / (all bets + raises + calls + folds)] / [(bets + raises) / call]

which is the same as:
Arg Freq * (1 / Agr Factor) = [(bets + raises) / (all bets + raises + calls + folds)] * [call / (bets + raises)]

Underlined terms cancel each other out

so...
Arg Freq / Agr Factor = call / (all bets + raises + calls + folds)
Station Frequency lol

Wow, cool math! I might try out this stat.

first, sorry, i didn't read op properly first time through, otherwise i would have spotted the mistake earlier.

anyway, so reading through this thread and the maths and something didn't feel quite right based on how i use Agg freq during in-game. Note the impact this has on Agr freq/agr factor - you can't call when checked to...

Originally Posted by rowhousepd

Arg Freq = (bets + raises) / (all bets + raises + calls + folds).

um, time to re-examine some math - seems that op's definition of Aggr Freq is incorrect ? can someone confirm?
I think it should be:
Arg Freq = (bets + raises) / (all bets + raises + calls + checks + folds)

Originally Posted by holdemmanager

Agg Pct Aggression percentage can be anything from 0-100 and is based on an aggressive action on each street. So if I bet the turn and the river but check the flop I would have 66% Agg Pct because I made 2 out of 3 aggressive actions.

This could be a difference between HEM and PT3. Directly from PT3, here is the formula they use for Flop AFq:

((cnt_f_bet + cnt_f_raise) / (cnt_f_call + cnt_f_fold + cnt_f_bet + cnt_f_raise)) * 100

So is this correct or not?

If Daven's formula is correct then it is calling freq because "checks" is included.

@tyrn-Isn't "station freq" the same as "calling freq"? Not sure what you're saying

Actually in HEM it seems more complicated than that:
- the aggression factor is simply the total number of postflop (bets+raises)/calls
- the aggression frequency is the number of postflop streets where a player took at least one aggressive action/total number of streets seen.

As a sidenote, a problem with the HEM definition of the aggression frequency is that it does not take into account the fact that a player "sees" streets when he is all-in before the river and thus is unable to take an aggressive action. Some guy attempted to correct this with a custom stat here.

Nothing prevents you to define a custom stat for calling frequency though. It's easy enough to do. See here all the formulas and code of the HEM stats. You can write your own in the customstats.txt file in the HEM /reports folder (create it if it does not exist).

Interesting. I've noticed before that PT3 AFq counts multiple actions per street. For example, if you bet and then call a raise, you're AFq is 50%: 1 bet / (1 bet + 1 call).

I'll write the code for the "calling frequency" stat for HEM, but what do we want? Streets where the final action was to call? Streets where at least one call was made?

The PT3 formulas are simple, if you want them to act like AFq, i.e. doesn't include checks and will count multiple actions per street separately.

Flop: (cnt_f_call / (cnt_f_call + cnt_f_fold + cnt_f_bet + cnt_f_raise)) * 100
Turn: (cnt_t_call / (cnt_t_call + cnt_t_fold + cnt_t_bet + cnt_t_raise)) * 100
River: (cnt_r_call / (cnt_r_call + cnt_r_fold + cnt_r_bet + cnt_r_raise)) * 100

Total: ((cnt_f_call + cnt_t_call + cnt_r_call) / (cnt_f_call + cnt_f_fold + cnt_f_bet + cnt_f_raise + cnt_t_call + cnt_t_fold + cnt_t_bet + cnt_t_raise + cnt_r_call + cnt_r_fold + cnt_r_bet + cnt_r_raise)) * 100

Something is disturbing with counting multiple actions per street in the PT3 AFq.

For example, when we bet, opp raises all in and we call, then we are only 50% aggressive on that street?

Also if they don't include checks at all, for example, let's say over 100 streets, the actions went like this:
30 checks, 30 bets, 20 raises, 20 check-raises. We get:

HEM Agg%: 70%
PT3 AFq: 100%

Originally Posted by daviddem

Something is disturbing with counting multiple actions per street in the PT3 AFq.

For example, when we bet, opp raises all in and we call, then we are only 50% aggressive on that street?

Also if they don't include checks at all, for example, let's say over 100 streets, the actions went like this:
30 checks, 30 bets, 20 raises, 20 check-raises. We get:

HEM Agg%: 70%
PT3 AFq: 100%

That's how it works, unfortunately. I often check my popup stats to check a villain's "Bet %" and "Raise %" on each street to get a better idea of how aggressive they are, because those are more straightforward values.

mmmh, yeah... taken to the extreme, the PT3 stat seems downright wrong. Take a passive player playing in totally passive games where 90% of postflop streets are checked through. He only bets 10% of his streets. He doesn't need to call or fold because nobody else ever bets. And this guy has a 100% PT3 AFq??... no way!

Anyway, if the HEM people want a HEM style calling frequency, I'll write it but need to know: at least one call or final action=call? It makes a difference. For example in a 4-way pot where the action goes P1 bets, P2 calls, P3 calls, P4 raises, P1 calls, P2 folds, P3 calls

if at least one call, P1, P2 and P3 all get 1 count for the street
if final action=call, P1 and P3 get 1 count but P2 gets 0.

Added this thread to the digest.