This is a very nice CPU but not the fastest one (2.4Ghz is a laptop version so it doesn't overheat/consumes less energy but it's slower than normal versions). It still gives very decent performance for the solver.

cap 3 with a donkbet is 13GB per tree, without a donkbet it's 7.4GB.
Flop trees are way way bigger than turn/river ones. Again, most of the memory will be optimized away over time once more urgent things are done.

Again.. turn solutions are small even with 5 bet sizes or w/e.
Flop solutions are big. Usually 0.5% is very good for a flop solution, 0.2% of the pot is close to perfect.

Edit:
Figured what i was doing wrong

Is there a skype group available for this program? I have been spending quite a lot of time using it and would not mind sharing some of my results.

Hi,
There is quite an active Skype group now. It seems I can't PM you for some reason (new account?). Please send an email to piosolver@piosolver.com and I will add you manually.

I am not sure if I can recieve PMs until 50 posts....

I am curious as to why your program is so quick to solve for GTO flop the flop, where as simplepost flop states it takes about an hour to solve.

And lastly, would you give coaching in this program, how to use the script and how I could go about organizing results, and schedule numerous flop solutions overnight. etc..

This program is really great, I play live mostly and already feel this program has helped immensely.

trying to connect and I get the error try running as a stand alone application no idea what this means thanks for the help. Using free version btw

Well I we would need to know what kind of error is that.
Maybe drop us an email at piosolver@piosolver.com with a screenshot or some description what is happening?

I was testing something like this.
On the river, pot is 100, stacks are 100 - 100. bet size is 100%.

OOP player has KK 50% of times, JJ 30% and 33 20%. IP has AA 20%, QQ 30%, 22 50%.
the software says that OOP should check 100% of times, IP bets 100% AA and 20% 22 so the value bluff ratio is exactly (2:1) and OOP defends 50%.

the problem here is OOP's KK is mixed strategy, he should lead it some % of time cause when he checks IP bets only 30% of his hands and when he leads IP should defend 50%. so can you check it whats wrong?

also the weird thing is that when IP bets OOP defend 36/50 of his kings 8/30 oh his jacks and 5/20 of his 33, the IP's QQ still prefer to check his queens, however the good logic would be that oop should call top 50% of his range so 100% of KK and fold anything else but maybe it doesn't matter in this case...

For me to able to comment on it I need to know what the board is.

okay sorry I thought the idea was understandable, take 6s 6c 6d 7s 9h

it says bet 36.5% for me. Maybe you forgot to allow for OOP bet by checking the checkbox for OOP player.

Again, check "include OOP ..." on the river and re-run the tree.

It in fact doesn't matter.
I want to comment though on "good logic". That logic only applies for simplified toy games where you don't block any part of the opponent range. In real Holdem situations the blockers are very important and just calling top % of the range is very exploitable (the opponent knowing that can choose his bluff so they block your bluff catchers). That's the reason why river strategies for more realistic spots are often very mixed in bluffing/bluff catching department - if you are skewed too much in direction of particular kicker for example it would be too easy to choose the bluffs to counter that.

I asked because the solution is different for me (36.5% donk bet, not 100% check) and I wanted to make sure we are talking about the same spot. I think you just didn't allow for OOP bet.

okay thanks, it was my mistake when I open the program the option "Include oop bet" was not visible and I didn't notice the scrollbar on the right to make it visible

I have one more problem. In my PC it works fine but when I download and launched PioViewer.exe on my leptop, there is a little loading and nothing happens. The GUI doesn't appear, my friend has the same problem on leptop and what might be the problem?

It's almost surely a Net Framework version issue. You can install the newest one from Microsoft website here: http://www.microsoft.com/en-us/downl....aspx?id=42642.
Let me know if that doesn't solve the issue.

yeap it works thanks a lot.
p.s can you explain what exactly "All-in threshold of the initial effective stack" means and how it works?

When the tree is constructed using the inputted parameters the bets which causes given % of initial stack to be invested in the pot (counting all the chips invested for all the bets so far) are rounded to an all-in.
The idea is that you don't want bets in the tree which leaves very small amount behind. It could be called "pot-committed threshold" as well (so the player never invests that % of their stack just to fold to a raise).

thanks, One more question:
can't I choose the exact combination of the hand ? for example if I want only flush draws of some range on the turn, if not it would be great if you add ..

This will be added in the next release (working in dev already).
For now you can type it in but I realize it's a huge pain.

Hey punter I was wondering if there was anyway the solver could be used to create solid strategies that are easier for humans to implement? Obviously they wouldn't be as strong as the GTO strategies or even close, but it just seems too hard to implement these GTO strategies into ones game. Something like this was mentioned in the simple postflop thread so I figured I'd ask here, but I'm not sure how viable it is to implement something like this into the code, if its viable at all. Basically as of now I can only use the solver as a guideline for how I want to construct my ranges then I would need to go through and iron them out in something like CardrunnersEV in a manner that a human can actually remember and implement.

I don't know the answer to this question but my intuition is as follows:
-it's probably possible to make the strategies pure on the flop without losing much EV
-it's probably impossible to make strategies pure on turn/river without losing much EV

This will require more testing and experiments though. With the amount of urgent features on our to-do (several of which coming in the next release) it's very unlikly to be a priority in coming 2-3 months.
One thing which will help though is range analysis feature which will give you some basic breakdown about ranges for particular actions. One could hope then that constructing a range with similar attributes but easier to remember wouldn't lose much EV. It will also be possible to test that hypothesis.

Small announcement for our free-version users (or would be users):

The PioViewer requires Net Framework 4.5 to run. Unfortunately we made a mistake and the automatic check doesn't work. Please download the newest version from Microsoft here:

http://www.microsoft.com/en-us/downl....aspx?id=42642

if it doesn't work for you. If it does start then you don't need to download anything.

Hi Punter,

Don't want to advertise your competitors, but have you seen the article about using a particular tool to generate 'minimally exploitative strategies'?

Is that something you will consider implementing?

Oh, I'd like to get in on the Skype group too please.

Yes, I have seen it.
We have that working in dev and it will be available in the next release.
The concept is certainly theoretically interesting but it has limited practical value. The thing is that the the name "minimally exploitative strategy' is a good one which is why that strategy is exactly what you don't want to play. If you solve assuming a fixed strategy in given point then the resulting strategy is such that it compensate for mistake made in all future nodes. If you then follow the solution yourself you are playing with an assumption that your opponent - having made a mistake is then perfectly aware of it and plays perfectly from that point on.
This assumption can't be further from the truth in practice. If the opponents were aware of their leaks and able to compensate perfectly they would adjust the leak in the first place.

That being said the concept is theoretically interesting and it is easy to implement (basically it's 2 ifs added, one in solving function, another in max exploit function) so we added it. It will be available in next release coming soon.

If a turn scenario is ready in 30 sec, how many minutes for a flop scenario to get solved?