Quote:
Originally Posted by asif00
Thank you for a quick reply. I have both books and they are excellent. Can you give some advice on best way to extract heuristics as you did extremely well when I run pio on different flops? Were some of your heuristics "educated guesses" since no definitive way to disassemble the pio solution?
Thanks, asif. I think you can look at the POP books, the second one especially, as a template for how to derive heuristics from solutions. To break down a little further, I have a workflow that looks like this:
1. Identify a situation I find a tricky or where I have low confidence about the equilibrium solution.
2. Make some predictions. These are akin to the questions from the book: how does equity and nut advantage break down, roughly how often do I expect a given player to check/bet/raise, which hands do I expect to be pure strategies, etc.
3. Identify surprises. Run the simulation and evaluate your predictions. Note where you are off. Look over the results for other things that surprise you or that you can't explain.
4. Find explanations. This is the hardest part, of course. It requires a good understanding of concepts from POP 1, in particular indifference and equilibrium. If you're surprised that a player checks at a high frequency, take a look at which hands are pure checks, as those are the ones driving the strategy. Take a look at how the solver expects the opponent to respond to a bet, how it expects them to play the turn after the flop checks through, etc.
5. Formulate a hypothesis. We're essentially back at Step 2. Your answers to (4) are new predictions you can test. Once you have a tentative heuristic like "do more checking when you have a large equity disadvantage", you can design experiments to test them. Run some other simulations and see whether your hypothesis holds true. Revise accordingly, rinse and repeat.
Hope that helps!
Andrew