Originally Posted by guyman
Well, I looked a lot at open-source bots code, and assembled something that measures the relative strength of a hand on a flop. I don't want to elaborate too much, because I don't want to give away my "secret sauce"
But anyway, this is mainly aimed to understand how strong hero is. After knowing that, we teach actions and reactions according to position, knowledge on villan, flop wetness, etc.
We try to find the right balance between going deep enough to be able to win the micros, and not too deep, to make things not too complex.
I have to admit, I didnt have time to test your simulator but from your general 'description' I assume you have to deal with lots of "hand vs range/range vs range on a given board" situations on the fly.
I've been thinking about writing a "similar" software myself (just for fun, nothing serious), not exactly like yours, but it would have to do lots of range orientated calculations (constructing ranges based on stats, dividing it into strong/weak fits, calculating equities between them etc etc).
Actually I coded a simple hand evaluator (just a standard "hand hashing with precalculated lookup tables" approach) and a Monte Carlo type equity calculator but its slow as... I can see some potential places which I could probably speed up a bit, maybe even use multiple threads but I read somewhere that applications like Pokerstove or Equilab dont use multithreading and they are really fast after all.
So thats why I asked that question, maybe you have some hidden ninja tricks how to do that efficiently. Oh and that 'tip' with poker bots
was actually quite useful, I picked some good pieces of code from them