[GTO] PioSOLVER - postflop equilibrium solver for Holdem
11-15-2017
, 06:40 PM
Quote:
Same as with any other window in Windows, by clicking the middle button here:
. You said I can make it fullscreen, so I thought it can be FULLSCREEN (you are talking now about what's called Windowed screen).This middle button is presumed to be pressed of course.
Quote:
If it doesn't move maybe the whole PioViewer window is too small or resolution of your display doesn't allow for more space.
That is my view as it is.
My resolution is pretty standard (1920x1080).
I would happily move the bar to the right but I can not.
Last edited by Novembersdoom; 11-15-2017 at 06:47 PM.
11-16-2017
, 12:07 PM
11-17-2017
, 10:39 PM
Quote:
A few points:
1)It makes very little sense to use % of the pot when solving preflop trees. The reason is that it's hard to define what starting pot even is (in HU it's 0 and then there are forced bets called blinds). It's better to use bb/100 metric
2)The current preflop solver struggles if there are multiple bet sizes postflop, the reason is that it's already uses very little memory and thus is less precise. There is little to no reason to use multiple sizes postflop when solving preflop trees as they don't influence preflop ranges almost at all (as long as you have at least one reasonable size included).
The situation with multiple sizes is going to improve in the next release but the point about there being little reason to use them stands. It's better to use a bigger preflop subset instead for example or just get to the solution quicker.
1)It makes very little sense to use % of the pot when solving preflop trees. The reason is that it's hard to define what starting pot even is (in HU it's 0 and then there are forced bets called blinds). It's better to use bb/100 metric
2)The current preflop solver struggles if there are multiple bet sizes postflop, the reason is that it's already uses very little memory and thus is less precise. There is little to no reason to use multiple sizes postflop when solving preflop trees as they don't influence preflop ranges almost at all (as long as you have at least one reasonable size included).
The situation with multiple sizes is going to improve in the next release but the point about there being little reason to use them stands. It's better to use a bigger preflop subset instead for example or just get to the solution quicker.
2) That's quite interesting, I thought there could be some minor changes when there're more strategic options for both players, which could probably lead to some better equity realisation overall.
Thanks for a quick reply, appreciate it
Last edited by lkjhgfdsa; 11-17-2017 at 10:52 PM.
11-19-2017
, 08:51 PM
Quote:
There is little to no reason to use multiple sizes postflop when solving preflop trees as they don't influence preflop ranges almost at all (as long as you have at least one reasonable size included).
Originally Posted by punter11235
There is little to no reason to use multiple sizes postflop when solving preflop trees as they don't influence preflop ranges almost at all (as long as you have at least one reasonable size included).
I've just tried to compare multiple sizings vs single and found that at least in my samples, it does influence preflop ranges, against multiple sizings OOP player was defending about 5% less for example. Same flop subsets (54). I guess I have to try it with more flop subsets, but I just can't run 128gb atm, have to wait for a BIOS patch to be able to run my memory sticks on x399 chipset, even though it was on QVL
11-23-2017
, 03:10 AM
Quote:
So my understanding is that for each decision pio looks at how ranges interact with the board then finds the optimal decision based on the chosen bet/raise size(s) in relation to the eff stacks
Quote:
and it seems to generally make up for the lack of sizings by using more of a mixed strategy than I'm guessing it would need to with 2 or more sizes (am I missing something?)
My personal view on why there is so much mixing is that without it it's very difficult to ensure checking or betting range don't have holes in terms card removal/board coverage. In other terms: it's too easy to play against someone who always bets a good hand (for example).
Quote:
So in theory where does the incentive for multiple sizings come from?
Quote:
Im guessing, so if we're using anything other than optimal sizings would the "exploitable for x" ever get down to 0%?