I got past this bug. But now whenever I try to do a bit of formatting, edit the layout and then save it, I get the following message: Some features in your workbook might be lost if you save it as a CSV. And then it doesn't save any of my formatting, do you have any idea how to fix this?

Ok i think i understand. So Im not really fluent with coding or anytbing like that, but I am interested in having the ability to solve preflop as well as post. I know the pro version now supports 16 cores and presumably 32 threads.

The website mentions that people who buy edge get a custom compile tailored to hardware. Could you elaborate please?

very interested in upgrading asap

Pro version supports 16 threads. This means you won't be able to use hyperthreading on 16core CPU with it. The diffrence between using and not using hyperthreading is about 15% in case of Intel CPUs at least (probably less with that many cores).

Only edge version supports preflop solving (although you can read preflop saves with basic/pro versions as well).

We ship optimized compiles for modern CPUs along with the edge version and we make compilies for untypical hardware. In case of Intel CPUs there is around 15% speed gain. In case of Ryzens/TRs I don't know what it is, the version of the compiler we are using now doesn't support have optimized compiles for Ryzen specifically and I've heard that AVX is slower on AMDs than it is on Intel so the gain will likely be minimal.

Ok, i would like to buy edge. How do we proceed?

PM sent.

I wonder if the GTO software couldn't work with less combos. I mean AsKsQhJh on JsTs6h is the same as AhKhQsJs on JhTh6s. The isomorph template of ProPokertools has only 16432 combos instead of 270725 combos.

It couldn't.

Short explanation: it's the same reason we need 1326 combos in Holdem postflop while we need only 169 preflop.

Longer explanation: you say that:

This won't help. One could use preflop ranges on the flop (that is using 16432 combos) but once any action is taken like a bet or a call then combos which are previously strategically the same are not anymore. For example in Holdem:

AKs are the same preflop and you can use them for equity calculation when the flop is dealt. Once say Js Ts 2h is dealt then AsKs, AhKh and AdKd/AcKc are different. On rainbow boards all combos in AsKs are strategically different.

Unfortunately even in simple equity calculator you need all the combos. For example in PLOCalc you can filter (and then calculate equity vs this range at any point) preflop ranges depending on what you have on the flop/turn/river. For example you can say this:

Preflop play any AAxx, double pairs, double suited connectors/1gappers.
On the flop filter: toppair+, 8+out draws.
On the turn filter 2pair+, 10+ draws

You will arrive on the river with range which is not representable using 16432 combos or anything below 270725 in case of some boards. Even hands like AsAh3c2h and AsAh3c2d are strategically different on Ks 7h 2s board (even though they are kinda similar). There is a blocker effect so As Ah 3c 2h blocks more flushes in opponent's range.

All that means that 16432 representation is only useful for preflop ranges and we in fact use that for preflop calculations (that's why saves are small when you save preflop ranges). Othe than that it's only useful on the flop at the beginning of the play. You won't be able to represent real ranges once any action is taken beyond preflop.

BUMP

I am sorry I've missed it.
I don't. I don't even know what "workbook" is. It's an issue of .csv reader you are using and this is a specific problem related to that program not to Pio. Pio just outputs the most standard .csv files which every spreadsheet like program operates on (Excel, LibreOffice, Google Docs and tons of others). Why it doesn't save your formatting can only be answered by authors of that program so I encourage you to contact them.

I didn't mean to ask about multiway, just need clarification about method.

I'm trying to build preflop ranges for various 6 max spots where the pot is heads up on the flop by using aggregated reports and looking at hand EVs.

What is the correct process for determining a rough estimate of a defending range?

Example: IP open, BB defense-
-Assign IP an open range and BB a cold call range, leaving out the hands that BB will 3 bet.
-Review the report and look to defend the hands that have EVs equal to or higher than the cost to call, and add a few based on variables like opponent strength, reads, realization, etc.

Some have said that if the hand EV is above 0, it's better than folding, but I assume we have to account for additional risk. As the BB in 2 bet pots, we will need to win back 2 bb in EV if IP's open size is 3x, 1.5bb if 2.5x, etc. I also assume that we should build the trees with starting stacks for each different size used, even if it's the same spot.

For 3 bet pots, we will need to win back our additional risk (3 bet size minus 1 bb) in EV.

I have gotten some surprising results so far doing this for two spots. Am I doing anything wrong?

This "solving preflop by hand" method should work but it's very hard to determine the way to adjust. +EV hands are in fact better calls but if you adjust all of them they will become folds as the opponent is going to adjust as well. The way to do is to adjust the most clear hands EV wise, run the report again, adjust etc. This is way slower than normal preflop solving but if you can start with an educated guess about ranges you should arrive at pretty good approximation of the solution after a few iterations of the process.

As to how to calculate the EV: you have to take into account what is invested preflop. For example if the blinds are 5/10, BTN opens to 30, SB folds, BB calls the pot is 65 so only hands which have EV higher than 20 are calls.

I hope that helps although I am not 100% sure if that's what you meant when you asked the question.

Someone bought a license yesterday but provided incorrect email address (probably a typo). As we have no way to contact them I am hoping this post brings attention to that. Please drop us an email to support@piosolver.com with some details of the order (for example name on the card used so we can identify it) and correct email address and we will send you the license.

I have a really strange bug right now. So I ran a script, and it saved like x amount of flops. Now when I look at one of the flops I calculated it loads the tree fine in the browser screen, however the one in the 'postflop tree building and calculations' is totally off. It resembles (might be) one of the last trees I ran yesterday. Now I saved all the trees as a small file, but thought if I wanted to re-run the tree it would still be possible if I juist rebuild it. However the configuration settings are totally off, and I don't know how to get them back to the original, other than setting every thing manual I guess. So is it possible what i'm asking or is this also a limitation of small saved trees? Basically I just need the original configure settings back (so ranges, spr & betsizes)

Edit: I found a .txt file in my saves, guess that's it.

Hey punter, loving the preflop solver and ability to use all my cores now! I was wondering, inside the preflop solver, why do you have preflop subsets as high as 400? It requires over 1000 gigabytes of ram O.O

Also, what do you recommend is a good number of preflop subsets in order to get a good idea of preflop strategy?]

One last thing, I was curious if there was anyway to have the solver determine optimal opening ranges for each position in full ring? It seems that the solver only really looks at each spot as if it was headsup?

Its not a big deal obviously, as I can just input a standard opening range and solve starting with that player raising, but I was just curious.

Hi, guys! Is there any updetes after standart 1.9.2? It's been a year already.

I saw new PLO tool, but maybe there some pre-realeased stuff for NL holdem too?)

The configs loaded to the viewer are stored in .txt files so it's important to always save script results to one fresh folder and to not have more scripts/configs there because it may confuse the viewer. This is only a visual problem though. The results are of course still correct, loading the config is meant to help in recognizing what the tree associated with the save is.

1)Some simpler trees wouldn't take that much RAM, you can limit a lo of options, especially postflop to make them smaller

2)Some users have machines with 1TB of RAM or more


We don't do multiway solving as of now. There are two reasons for it:

1)It's impossible without heavy abstractions
2)Multiway equilibrium is shaky which means that you can sill lose if you follow it.

1) and 2) make multiway solutions not as practical but still very difficult to get right. We appeciate there is practical value in preflop multiway equilibrium approximation but we are not focusing on it right now.

With business model of selling stand alone software for fixed price you first need to invest a lot of resources creating the software without being paid anything. Then you sell it and update it for a while. Then you have to start doing different things or you will be left with nothing once new sales dry up. It's not feasible to ship free updates forever without subscription based model which we wanted to avoid.

For the last year we did a lot of work in both Holdem and PLO. Some of it is getting released right now. Some of it will be in next free update for our main product. Some of it will be available in Pio 2.0 in the future.

I hope that makes sense. We feel we shipped a lot of free updates. Usual way those things work in the industry is that you sell ready product (for example a game) and then you limit yourself to bugfixes for that and start working on new things immediately.

Keep going exactly the way you have been, please. NOT doing the "subscription" thing, but still updating, you guys deserve a medal

Any thoughts on taking advantage of the gpu/opencl to assist the cpu in speeding up solving?

Our current algorithm and in general most known solving algorithms are not very well suited for GPUs. I am sure it's possible to code it efficiently (or at least make the GPU help with significant part of the calculations) but it's a bit of research project and not something with guaranteed success.

Damn, I guess ill wait on the quad SLI setup

We are optimistic about optimizing speed in the future

hi guys,
does anyone have please an empty Excel file to share that you guys use to save all the results from PIOsolver?
thanks a lot

Just to know how to build it

Hey,

I'm spontaneously selling my Windows laptop (using a Mac as the main laptop and just had the Windows laptop for PIO purpose) today and will be getting a new one fairly soon. Regarding this I found the following instructions in your FAQ: "5. Can I move it to different/newer computer?

In general yes although that has to be reasonable (like not every week or so). To do that open the solver and after seeing a console with a welcome string, copyright notice and a date of the compilation type "deactivate" (without quotes) in there and hit enter. Then close it and open it again to see if it worked (it should ask for the registration key again). After that you can just activate it on another computer."

Since returning from Barcelona I somehow wasn't able to connect to PIO again for some reason but I also haven't found the time to solve this issue yet. Therefore I can't really access the program and I can't enter/find the "deactivate" console. I'll be selling the laptop in 1 hour so I don't have any time left to fix those issues. I of course would still like to use the software on my future new laptop again though. What's my move?

Just email us with your registration key and in case of pro/edge license approximate time of the activation you want to reset. It's not a problem. There is always a way to get Pio license back if it's stuck or lost somehow.