Still question about accuracy, in your graph in your range explorer,
now we know that a lot of this hand won't exist in equilibrium,
but it just clutters the graph, would it possible filter out these hands?
Like say, don't display hands with freq <5%?
Quote:
This can't be explained by 1-a reasoning.
This can be explained by 1-a reasoning.
I've constructed a very simple example,
using board 2c 2d 2h 3s 2s,
OOP range, AA,KK,QQ,88 and 0.5 66
IP range,JJ,TT,99,77
starting pot, 100, let OOP bet 50
you can also tried bet 33,still holds
so that's a two step process,
first, you find the bluff hands which takes a mixed stragety,
which means its ev_check=ev_bet,
then you remove the hands worse than it, as you can see, 77, doesn't count,
since it will lose to 88 no matter we bet or check, so we don't take him into consideration.
After removing 77, the remaining hands satisfy (1-a) constraint.
Also you can see that OOP 66 is bluffing 100%, since it's strictly ev_bet>ev_check,
it takes a pure stragety.
I also adds QQ to IP's range, as you can see, still satifised 1-a.
a=0.46*3/4=0.34
Also I want to ask a question about stack size,
since we're impossible to run infinity stack size sims,
Let's say I run a sim for stack 100,
how much ev loss would be if I play the hand same as stack 100 when we're deep, say stack 200?
the function of ev should be smooth right? Which I mean, it's similar to play a hand when stack is 100 vs when stack is 101,
also it's similar to play a hand when stack is 101 vs when stack is 102, so we can give some reasoning about this right?
but when the gap becomes bigger, I'm not sure how much ev loss would that be
Last edited by femtowin; 11-28-2019 at 05:21 AM.