Quote:
Originally Posted by inneedformoney
Hi,
Is it possible to round equilibrium solver hand weight to 0 or 100%? This is mind blowing when it suggest you to bet AhQh for 42% and etc, why cant make it only 0 or 100%, a lot more practical use
Making sense of the ranges is indeed one of the upcoming challenges. I have some ideas on how to do this and we'll just have to wait and see if they work out in practice.
For now though the solver is already really handy for filling in remaining play in your trees (which is one of the main reasons why I really wanted a solver algorithm for CardRunnersEV).
If you want to enter a certain spot, but don't want to spend too much time filling in the play for all of the remaining actions then you can just lock whichever play you want to apply and let the solver fill in the rest of the tree.
For example, in the screenshot below where SB bets >=tp or a flushdraw and villain raises with >=2pair and 50% of his gutshots you can just lock those actions and let the solver finish the tree for you.
It should make postflop analysis quite a bit easier.
Another thing you can already do is check how much of a leak a certain action is by checking how much the EV will change if you unlock the action and let the solver figure out equilibrium play for that action instead.
Tip
Should you want to do this then it's important to always look at the EV for the first decision in the equilibrium tree and see how that EV is affected. So in the screenshot above you would always need to look at the effect on SB's first decision,
even if it's BB's decision that you've (un)locked.
BB's decision isn't what you should be looking at here, given that if BB's EV goes up, SB will simply check more often. The EV for BB's decision may be much higher, but the decision will be reached less often due to SB checking more. For this reason, it's the overall EV that you need to look at, and this is easiest to obtain from the very first decision.
Flop play
Although the speed for 2bet trees is already starting to get there, 3bet trees may still be a bit too slow for some people's tastes. However, if all that you're interested in is the EV for the first decision then it appears that once the dEV goes below a few % of the pot the EV doesn't really change that much anymore.
For example, in the example file flop_2bet.stx solving down to 5% will result in an EV of 13.83, while solving down to 0.3% will result in an EV of 13.98. It's a 1% difference.
The frequencies may still change, but the EV of the first decision seems to converge pretty quickly. It's definitely not ideal, but this compromise should already allow you to get pretty decent results within a few minutes for 3bet trees.
Last edited by scylla; 09-09-2015 at 03:46 PM.