A Concrete Attempt To Model

Let's put aside the theoretical worries about how hard it is to model from observables, and the problem of a moving model. Let's look at something really simple, and really concrete, but pretty fundamental to opponent modeling and see how our ideas thus far will carry us.

Let's try to model the Preflop-hand range of an opponents. What are the various things that would affect their hand range?

* Some concrete valuation system of what is playable and how hands should be ranked
* Amount of money already in pot (i.e. blinds, other callers, etc)
* Amount of money needed to be added to the pot (i.e. someone raised)
* Number of players involved
* Opponents view of players involved
* Position

So some of these things end up being tied or related, but let's not worry about that.

Which of these items are most relevant (empirically)? Which of these items should be most relevant (theoretically)? Which of these items are hardest to account for?

I think coming up with a model for the opponent's view of other players is really tough, and should be stricken from a first pass decision. So let's assume our opponents do not differentiate between agents at the table; they have a view of how other players are playing, and this is what ends up determining what the cutoff range for playable hands is -- but we are consciously approximating this by assertion of the cutoff range.

Though the amount of money already in the pot vs the amount of money needed to be added to the pot is probably what good players consider before action, this can be simplified into the questions:

* Is this an unraised pot?
* Is this a raised pot?
* Is this a reraised pot? (Perhaps this is a question we should ask in a more refined model)
* Am I in the small blind?

Given our simplified model, what are the observables that affect/determine this model?

Ooops, time for bed...