For example, I solved a BTN vs BB spot 33bb eff, and the EV IP is 73.3 and the strategy is 70% cbet and 30% check.

But when I lock the solver to cbet 33% with whole range, the EV decrease to 72.5. So I loss 1.36% of the optimal strategy to simplify my strategy cbeting all hands.

I would like to know if this 1.36% loss of EV is justified to simplify the strategy and what would be the maximum amount of EV loss that I should use to apply this method.

Hi, since Wednesday PIO is not working for me, asking me to activate my license, and when I enter license getting this error.


It was working fine before that and I didn't changed anything in my settings.
Already asked in discord, and on mail but still didn't got respond.
Thank you.

OK so is the Maximum Exploitability Score the EV that a Maximally Exploitative Strategy can get against the current solution for the other player?

Is there a difference between a player's current solution and the Maximally Exploitative Strategy that is used to calculate that player's MES score?

I play mainly on PokerStars and wanted to try out PIOSolver but the site blocked me from accessing saying PIOSolver is blocked. Is there any way around this? Also I am looking for a PIO Solver tutor, any offers appreciated.

It's a judgement call. Simplifying your strategy also simplifies it for your opponents so it's hard to say if it's ever worth it. It's really a question for coaches/practical players not for solver programmers
It's weekend so it's slower than usual.
Let me copy-paste the instructions here:

MES is Max Exploitive Strategy and yes, it means the max EV you can get against current solution.

Of course.
To illustrate it imagine solving Rock Paper Scissors. Imagine that after 100 steps the solution is:

player one: 0.35 0.34 0.31 (frequencies for Rock Paper Scissors)
player two: 0.32 0.32 0.36

Max exploit for player one vs player two is 1 0 0 (always rock). Max exploit for player two vs player one is 0 1 0 (always paper). It doesn't mean those are good strategies but they do have the highest EV against current solution.
In a perfect equilibrium (exploitability of 0) the solutoins are MES'es vs each other and they also guarantee to have the highest possible EV against a perfect villain (someone who always plays MES vs out strategy).

PokerStars asked us to implement a check so it's not possible to run Pio along with their client. It's fine to run it when PokerStars is not running.

Thanks.

I remember you saying a while back that you can choose any amount of mixing among actions at equilibrium and obtain a Max Exploitative Strategy. So is it also correct that if Player 1 plays 0 1 0 (always paper) that would also be a valid max exploit strategy?

Are there multiple valid Max Exploit Strategies that can be used to "test" the current solution by calculating an MES score using those Max Exploit Strategy?

Yes.
Let me re-iterate some facts about Max Exploitive Strategy (MES):

1)MES is easy to compute unlike the equilibrium itself
2)In a perfecte equilibrium both strategies are MES vs each other.
3)Changing any mixed action in MES produces another MES
4)There exists at least one "pure" MES (without mixing); this is not true about equilibrium which often requires mixing

It's the only MES against current strategy of the opponent.

1)There exists at least one MES
2)If there is more than one MES, then:
-every one of them have the same EV
-they only differ at decision points where EV of 2 (or more) actions is the same

Such testing wouldn't make sense as MES by definition is MAX EXPLOITIVE that means it has the highest possible EV against current solution so it it in itself the ultimate test.

I have SB vs BB preflop sims that I've run with the preflop solver.

Would solver EV of fold in BB be equal to 0 to -1bb?

Does "current" amount in EV display include posted blinds in SB vs BB preflop sims?

Hi, guys.

Advise pls is it necessary to overclock rayzen 3970x to reach 0.8-0.9 bench?
And if I'm not overclock and will use standard settings, what bench it will be?

I mean, you can check yourself but from what I have just checked the fold there has EV of 0 as it always does.
Easy way to experiment is to build pure preflop trees as those are instant to build/solve.

I don't know because my 3970x setup still hasn't arrived. I will let you know once it's here.
Bench on 32core Epyc I've tested is 1.36 though and that is clocked at 2.4-2.5Ghz so I expect 3970x to be in the ballpark of 0.8-0.9 if it runs at 4Ghz.

Could u give some details about your order:
1. What Motherboard did u choose?
2. What RAM (mgz)?
3. Cooling system?
4. What case?

I don't really want to recommend anything before I get the computer and test it.
Right now we are having problems buying RAM so the config might change anyway.

Got a question. Never used any software for poker like piosolver ever.



But im going to get a new ssd and ram upgrade. My laptop is an old dell xps 15 9550
with i5-6300hq processor and 8gb ram and a 250gb 2.5 inch ssd that is running out of space. I use hem2 and starscaption/888caption etc.


Im going to upgrade to a 1tb nvme ssd and 16gb ram. Now i read ppl get 32gb for ram and fast processor for using pio and trees. Now is this completely new to me as i never used pio or anything like that.


But is my processor even fast enough to do something like this if i ever use piosolver? Im considering upgrading to 16gb ram but thought about 32gb ram but is that necessary?


I might use hem3 soon as well so i am wondering if i should go to 32gb ram instead. But is my processor even fast enough for piosolver if i decide to use it in the future? If not, i think i might just go 16gb ram.


Again this is a laptop.

It's a laptop so it will never be fun. That being said this CPU is quite decent so it's possible to solve simple postflop spots in reasonable time.
16GB is a lot for postflop trees. 32GB is an overkill on a laptop imo as even if you want to build a humongous tree bigger than 16GB you will run into limitation of your CPU being to slow to solve in reasonable time.

I notice that trees where flop is all different suits tend to be much larger than flop all of same suit and flops with a FD tend to create trees right in the middle of the two in terms of size. Is that to be expected? Why would suitedness of flop affect size of the tree?

Okay to confirm. I use this laptop mainly to play poker and use hem2 and possibly hem3 and also programs like starscaption/888caption etc.


I never used piosolver or any of those software. Im just wondering right now if i should upgrade my ram from 8gb to 16gb vs 32gb. So me having 16gb or 32gb will mean nothing relatively if i want to use piosolver on my laptop right? I was going to get 16gb ram earlier but then the holdem manager thread, they said its best to get 32gb ram if i could because its more intensive than hem2. I always thought 32gb ram is never needed unless you do extreme things.


So you are saying 16gb ram vs 32gb ram on my laptop machine won't mean much at all right? So don't consider this a factor when choosing which ram to upgrade to? Because anyone who uses piosolver and get computer software for it... is doing it on a desktop with at least 32gb ram minimum and that ryzen processor which is miles better than my i5-6300hq?


Thanks.

It's because Pio does optimizations based on isomorphism.
For example, if you have this flop:
As Ks 2h then 3d and 3c turns are strategically the same (assuming range preflop doesn't have biases towards a given suit which it shouldn't have).
The same goes for paired boards. For example Ts Th 2c - both Ks and Kh are strategically the same.
Unfortunately on a rainbow unpaired board every turn card is different strategically so there is no way to cut the tree.

I am really not a hardware expert nor do I know how much RAM those programs need.
32GB is obviously better than 16GB. The question is if it's worth the money and that's the question only you can answer. My intuition is that you will not need more than 16GB for about anything unless you somehow want to solve very big trees and run other programs at the same time but then your CPU would be a bottleneck anyway.

This is correct.

1)We have a lot of customers who use laptops, most of them have 16GB of RAM
2)We have a lot of customers who use desktop computers with 16GB of RAM and it's enough for even big postflop trees unless you go to extremes with bet/raise sizings
3)A lot of our customers only have 8GB of RAM and are still able to solve trees with 2-3 bet sizes at some points.

4)Desktop CPU, even something from middle of the range like 3700X is going to be several times faster than your laptop (or any other laptop). In case of 3700X it should be 3.5x-4x faster. There are CPUs on the market which are more than 10x faster than yours. I expect my new build to be more than 15x faster than your laptop for example.

5)You need a lot of RAM when using Pio for two things:
-very big postflop trees (multiple bet/raise sizes)
-preflop trees (here you need at least 64GB, 128GB is better)

Cant open Pioviewer, keep getting this error

https://imgur.com/dCPmf8c

Started working again without doing anything.

saw this and got so excited to answer and sound intelligent but, alas, punter beat me to the punch

It seems that PIO only saves notes in the "Additional info" section (under "Tree Information" tab) for full saves. Can you make it possible to save notes here regardless of save size for the tree.

In fact I seem to be having trouble getting the notes to save regardless of what kind of save size I choose...

Hi everyone,

I'm going to run a script for BU vs BB SRP

I made this config tree and would appreciate some opinions on it



The aim is to simplifying at first and don't make overcomplex strategies that are really hard to master (that's why I don't put flop overbets for ex)

I'm really not sure about the flop raise sizings (I think just one sizing would be much simpler to implement and run but I don't know wich one would be the best)

If it could be relevant at some point, I'm playing 100NL to 200NL @ european rooms

Thanks

Sounds like a bug. Before we investigate it can you please make sure you are on the newest version of the viewer/solver? It should be 1.10.24.6 for the viewer and 1.10.23 for the solver.
First, it's always better to share the whole config. You can get that by clicking "copy to clipboard" button and then pasting either here in [ code ] [ /code ] tags (without spaces inside) or to some service like pastebin and linking here. This way we could see the whole config.

It looks more or less correct though. It's hard to judge for me because I don't play anymore but are 30% bet sizes on the turn important? Another thing would be that if you are using small bet sizes like that maybe cap of 4 could simplify the tree a bit (so bet/raise/raise/raise = all-in).
Another small thing is that you may consider inputting raise sizes as multplies of the previous bet like 2.5x, 3.5x or something along the lines instead of % of the pot as most people think in those terms.

I am updated to the latest version.

Thanks for response. Well with my xps 15 9550, well i obviously can't get a new cpu since its a laptop. But having 16gb vs 32gb wouldn't be that much of a restriction in RAM right if i were to use piosolver in the future on the same laptop?


I dont know much about piosolver, but how much bet sizes can you get with 16gb ram vs 32gb ram on a laptop?


You say you have lot of customers who use pio on a laptop with 16gb ram. I always thought ppl who use pio had those really expensive machines that are desktops with that at least 32gb ram or more? So the majority of them actually might have 16gb ram and laptop only like me?