Pardon the ignorant question but ... :

(THE QUESTION IS, HOW DO WE DETERMINE THE ORIGINAL RAISER'S OPEN RANGE?)

Here is how I understand things, please correct anything that I am saying that is wrong.

GTO is a synonym for Nash Equilibrium.
N.E. is a set of strategies where each player knows the other player's exact strategies and all players are maximally exploiting each other at all times.

You end up with am equilibrium strategy pair.

So ... If I open a certain range for a certain amount, say a 20% range for 2.5x, what is the Big Blind's 3 bet strategy?
This depends on a number of things, ... MDF, and in part, what is the original raisers 4 bet strategy.
O.R.'s 4 bet strategy depends in part on the 3 bettor's 5 bet strategy.
So these players and their strategies are interdependent.

MY QUESTION IS, HOW DO WE DETERMINE THE ORIGINAL RAISER'S OPEN RANGE TO BEGIN WITH?

It seems like, if this is original range is wrong (or not optimal), then everything else is wrong.

When using a preflop solver your inputs are just oR sizing & rake then thru/ cfrm the solver plays out however many iterations of the game tree (including postflop) and then the RFI ranges are determined at equilibrium.

Sent from my HRY-LX1 using Tapatalk

Thanks. What is cfrm?

Every strategy in GTO depends on every other range and strategy on the same branch of the Game tree. The only way to calculate It is with a computer.

You can make some good guesses as to what variables affect each node the most. For RFI, things like how much you get collectively 3bet, how often you go to postflop and whether it's multiway or heads Up, and how often you steal the blinds are some of the most importante variables

Thanks.

counterfactual regret minimization

Sent from my HRY-LX1 using Tapatalk

You don't know the correct range to begin with, it's an iterative process. Here's an oversimplified version of how solvers work:

Start with two players, A & B, and completely random strategies.

Player A assumes player B's (random) strategy is fixed and adjusts to exploit it.
Now fix player A's new strategy and let player B exploit it.
Now fix player B's new strategy and let player A exploit it.
Repeat until equilibrium.

Thanks. That makes sense.

I love this simple explanation. Is manually setting a strategy that we think is good instead of using a random strategy cut down significantly on the computation time or is it pretty irrelevant?

Thanks!

I think it can be, and it's been talked about in some of Noam Brown's most recent game theory papers.

The problem is that it can be difficult to define a "good starting strategy" algorithmically.