3-bet modelling for a HUD

Some interesting discussion about classifying and modelling 3-bet calling range depending on different criteria (for HEM/PT3/PT4).

Lately I've been trying to decide how to classify opponents' 3-bet call range when I don't have enough hands in my database to do so. Does anyone have a rule of thumb that they use, perhaps based on opponents' other stats. Maybe a percentage of their opening range differentiated by player type? e.g. Nit, Tag, Lag, passive fish, aggressive fish (maniac)

"A true lag player will 3 bet call a higher percentage of their opening range than a nit or tag.

Anyway, the best way to expand my main point is probably to lay out the action in a bit more black and white:

TAG calling the 3bet

TAG - Open raises
LAG - 3 bets
TAG - Often folds, or shoves/4bets with the goods. Not so much calling.

LAG calling the 3bet

LAG - Open raises
TAG - 3 bets
LAG - Will see that as frequently strong, and will call often to look to beat the hand post flop, or the player post flop. The best lags will very rarely raise-fold, and they will have a healthy degree of consideration/balancing on the timing/frequency of 4-bets.

"If you are serious about wanting to model it via some form of custom hud stat, then you have a pretty big task on your hand.

Firstly we need your LAG/TAG/etc definitions from your HUD

Then we need to take into account table size, position (both open raiser position and 3-bet call position), effective stacks, presence of other players in pot, and stakes.

We may also want to try to tweak it further using attempt to steal/defend stats

Then we need to take into account that a pure percentage based range may not be accurate. What a lag raises (and calls with) is often going to be a wide mix of not necessarily true range model percentages. Some sort of bespoke sub set ranges or definitions are probably required.

Then after all that we have an issue of reliance. ie relying that the type defined by a small data set is going to extrapolate out to a specific ranged behaviour, before we have sufficient stats to record that specific behaviour itself.

Personally if I am honest I don't think the reliability is going to be accurate enough to warrant the attempt. But if you have the time and the inclination why not I guess. What are you building it in PT3/4 or HEM?"

"Let's take "outright profitable" to mean either they fold, or they call weak, or they 4bet with worse than you.

At it's most simplest, trying to look at those factors purely in isolation, you can:
1) extrapolate out tightness (vpip) to tightness (fold to 3bet).
2) looseness (vpip) to calling weak (range-wise)
3) lagginess (vpip+pfr) to 4betting worse. Factor in your own image/stats a little.

Obv we're working only with the most basic stats here because we're working from our very limited ~30 hand sample only, and we're making pretty big generalisations for that reason. So there are going to be many weaknesses/exceptions, and other things already mentioned to account for (ie to try to factor in on-the-fly)

One of the on-the-fly things mentioned being effective stacks. With regards to lags, if their remaining effective stack is less than 5 times the call cost (diff between raise and 3bet size), which is generally rare from their side (but still), then you can expect quite a few more folds for sure, and if it is 5-10 times the call cost you can expect a few more folds.

In summary you can roughly determine your "outright profitable" 3bet line for players with little history by 3-betting appropriate hands against the ones with low vpip or high vpip, and awkward sized effective stacks. Try to take into account as many of the other "on-the-fly" factors as possible."