I bought a pro version. I had a specific question / bug report tonight, and even told punter that if he feels he is too busy these days, I wouldn't wanna harrass him too much right now and wait for the next update in a few days where that specific problem most likely would be fixed anyway. Nevertheless, he wrote several mails back and forth with me to make sure that the problem was dealt with immediately.

I think, with edge users, he / they feel more obligated to be even more helpful, and perhaps let them have access to some new features earlier than the rest, but am not too worried that if pro / basic, or even free users have suggestions that make sense, stuff will not be implemented because of our "lower status".

I am somewhat confuse with setting up the parameters. How do I make a player cbet OOP? Must I check "include first donkbet"? Totally confuse due to the language use here.

So what % are my estimated gains for i7 6 core/12threads running basic (4 threads) vs pro version (12 threads)?

Exactly, you have to check "include first donkbet" to be able to calculate OOP cbetting ranges.

Interested in joining a Skypegroup too, btw!

That's really great to see you want to that Unfortunately managing/moderating a private forum is a time commitment I can't take, at least right now. If you guys get something going though, let me know. I will make sure to check-in regularly to answer some questions etc.

It will always be the case with more information on the screen. Some people we showed it to love it. It's optional anyway, you don't have to use it (it's implemented in a way that the checkbox scales the squares by weight in every view)

Yeah, we didn't expect anyone wanting this anyway so we left one of the two we guessed would be more popular.

More variables will be added in coming update of the PioViewer.

You put me in a tough spot here: that's the discussion I am not interested in having on the public forum. On the other hand if I leave it unanswered someone may think I admit that there is a point in what you wrote, having in mind the assertive tone of your comment. That couldn't be further from the truth.
If I answer with technical details I would compromise the goal of not having this discussion in the first place so that's not going to happen. I can offer you some explanation privately though if you are interested and want to be assured bad things are not going to happen to honest people

Yeah, this is true. I assumed the donkbet will be universally understood as just bet first OOP but I forgot that it's usually not associated with things like c-betting OOP.
Thanks for pointing this out. We will try to come-up with better way of describing it (or maybe you could suggest one as well).

If you already have a basic version you can run a benchmark (open the solver, type bench). Pro version on your hardware should produce about 11 seconds.

If you don't have a basic version then I can offer you a good guess: I think it would be about 2x faster.

Hello,

We've just pushed the PioViewer update. More on our blog: http://piosolver.com/blogs/news

The new file can be found in your dropbox folder. Just copy it to where the previous PioViewer was and that's it. You can keep both for a while to make sure the new one works. If you somehow lost your dropbox folder link/password, don't worry, just drop us an email.
We will be sending an email notification probably tomorrow.

Let us know what you think!

^^

Hi ... why are those two numbers different, and if I aim for e.g. the "very good" solution of 0.5% - which number should I take for that, the upper or the lower one ?

The upper number is an absolute number, so in this example it's exploitable for 0.024$ which is 0.0436% of the pot of 55$.
The guidelines are for a number at the bottom (the percentage).

The solution on the screenshot is way better than what you would normally want (it was obtained by forgetting to stop the solver and falling asleep). You don't need to get anywhere that low to have almost perfect GTO in the tree.

cbetting is just a bet from the previous street aggressor (regardless of position). A donkbet is a bet from an OOP player than wasnt the previous street aggressor and is actually never associated with cbetting OOP.

Yeah, the solver is unaware of preflop action though (ranges and pot is all what is needed to describe the situation) that's why "donkbet" seemed natural to me. It is now (in the newest update available) changed to more descriptive: "Start the tree with OOP decision".

Can I modify strategy at some point and instead of Nash calculate best-response strategy for rest of the tree?
Let's say I want to diverge from Nash path at some point by modyfing player calling/betting range.

As of now it's not available although I have tested some of this functionality in dev version.
There are two ways to do that:

a)calculate MES (max exploitive strategy) for one player after changing a strategy in some nodes for the second player (say reduce river bluffs for one player and/or changing c-betting strategy) without touching other nodes of the second player.
b)change strategy in some nodes of 2nd player, lock those nodes, solve again (basically this way you get GTO with some hard coded assumptions about strategies).

While I can see both being useful I wonder which functionality you have in mind.
Those things are not at the very top of the to-do right now (building arbitrary trees is) but your feedback may help with priority choices along the way.

I meant b).

One of the tools available on the market can do that
I could consider to buy PioSolver but with this functionality.

Well, yeah
Your best option is to wait and see what is available. We have our schedule and this functionality while very nice isn't at the top of the to-do right now.
As to the other tools, it's good times for the customers as you can now sit and enjoy both teams fighting for the customers and introducing new features.

Let me just say that while I have great respect for anyone who can implement a solver like that there is a difference between claiming and delivering. We already have many people running the solver almost 24/7 and our results have many eyes on them so in that sense it remains to be seen how good the other team is and how good the produced results are (or if they are even correct, although I have no reason to doubt it)

It's becoming clear that such tools will be necessary to survive in modern Holdem cash games so your best option is to wait and see for a while and then choose the tool which is delivering what you want.

Do you you any abstraction model in your calculations?

The new update is great. The new variable added, "other", allows me to do exactly what I wanted to and the new way to display ev (not splitting up by action) makes it easier to read.

Thanks for contacting me privately, I now understand why this anti piracy method presents no harm to honest buyers of the software.

You missed my point. Yes my ability to see data becomes worse as more data appears on the screen but my point was that the value of the area used to represent one data set (range) affects my ability to see another data set (strategy). By having the cell split into 3 sections, the value of the range wouldn't affect my ability to see what strategy to use because the size of the area used to represent strategy remains constant. If the cell is split in three equal sections it doesn't matter if the range is 0.1% or 100%, it's still easy to view strategy. But in the case of the current implementation it becomes harder to view strategy depending on value.



This is after taking a rare line (OOP donking). It is very clear in this image what the range %s are (they are all clearly very low), but it is difficult to judge what the strategy and ev is because the area used to show strategy and ev is so small. If the cells were split into 3 sections (range, EV and strategy), it wouldn't be made any harder by the particular value of range.

That leads me onto another feature request: normalization of the visualization of range data. The hand with the highest range value would be multiplied by a number that brings it to 100% (so would take up the full cell) and all the other ranges are multiplied by this too. So it would no longer be hard to see the differences between the hand. Now it's hard to tell if one hand is twice as common as another if it takes up 2 pixels vs 1 pixel, but with this method it would take up the whole cell vs half a cell making it much easier on the eyes. This feature would be useful now and also would be useful if the cell is split into 3 sections.

Allow me to clarify my harsh criticism. I do appreciate the new feature very much. But in my view it is much worse than the alternative of splitting a hand cell into 3 sections. PioViewer was going in the right direction by allowing cells to be split in two, but instead of also allowing the user to split it into three cells in the new update it now does something totally different which I feel is very inferior.

I did find a small bug in how the data is displayed shown in the next image.



There's some kind of visual artifact that sometimes causes the very edge of cells to appear red for no reason. If you zoom in you can see the very edge (one pixel) of the JJ cell is red when it should be green. Also there is no black border completely surrounding many cells on the edge of the range such as J9s, JJ and A8o and this artifact is also present in the previous image in my post.

Alongside the exploitable for x%, it may be wise to add some kind of description to help users understand the level of accuracy attained. The accuracy rating could say very low (3%+), low (2%+), medium(1%+), high(0.6%+), very high (0.3%+), excellent (0%+). I say this because there's already been some confusion over what % accuracy should be targeted. There should also be a loading animation while the program is calculating its results. Like a rotating square or something that cycles through colours, some kind of animation that tells the user the program is doing something. Right now I have to read the text to see if the last message is "SOLVER: stopped" to figure out if it's running or not.

I just bought the Pro-version. Happy sofar.

Is it possible to select how many processor threads that the program is allowed to use?

Why it is not possible to use it at cloud computer? Is it due to licensing or technical reasons?

What advantage would you see with having it in the cloud ?

Fan noise, power cost, ability to run at 32 cores.

Starting an unofficial Skype group for those who purchased the software and want to share insights/strats. Add me on Skype: onlinepokes

I will not answer questions about how the solver internally works beyond necessary minimum. The results presented (EV's, frequencies, MES'es, Exploitability etc.) of course don't assume any abstraction which becomes clear if you try browsing some of them.

Yeah I can see that. It's just not that easy to make it usable. We will make another attempt at it for sure at one point.

We actually tried that 3 days ago but first attempt wasn't very successful. Putting it back to the to-do list.

Luckily the way it is coded we can keep trying to provide useful views or maybe even expose some of this functionality as configurable plug-ins. Another thing on the to-do.

We were thinking it's nice that the solver reports raw numbers and let the user decide what is good enough. Doing otherwise would convey a bias of the authors what constitutes a good solution. I've already shared my view on it which is a bit similar to the one presented by you.

Yeah, it's on the to-do and very useful, just not the priority as of now.

Yeah, you can also keep task manager open (as I always do) but I agree it would be nice to have some kind of animation for that.

For now it's only possible from scripts (set_threads n) but if you need that functionality for normal usage in GUI contact us. I think we can make you a quick fix.

Mix of both. Cloud version might be available in the future but it's somewhere on the bottom of the list of priorities as of now.

Currently when you copy hands EV into clipboard it is sorted from the highest to lowest. Can you sort it with the fixed order to be able compare hands at different flops?

Currently not possible from GUI. We may add this in the future.

Hi,

I am looking in upgrading my PC system. Till now I really did not had a good reason to do so and it was my intention to postpone it until my installed Windows is no longer supported.

I have found your system requirements on your site.

Can you add 2 sections to your site?
The first where you sum which kind of minimum components are required to get decent performance out of running 1 copy of PioSolver pro.
The second where you sum which kind of components are required to get close to everything out of running 2 copies of PioSolver pro on 1 system.

I do not really know that much about those things but I guess processor, RAM, HD, ... surely have to be considered in this?