Well, nothing is set so it shows nothing. If you go to Tools->Configuration->Debugging/logging and enable log then you can see the whole communication from the viewer. You will see that when it builds the tree it first sets all the parameters for the tree and then build it. When you double click on PioSOLVER-free.exe you are getting a new instance of "fresh" solver so nothing is set there yet.

No, the viewer is just an external program which communicates with the solver using the public text interface. The same way you can code your program.

This is fraction of the pot so 500 * 0.002 = 1.

Yes but notice that a pot of 500$ is quite big for 5$/10$ game. It's natural that in branches of the tree where the pot is already very big the accuracy in bb/100 terms is not that good as in branches with smaller pot. The relative accuracy is similar/the same though. For example if it goes cbet/call; check/check then the pot would be something like 145 and the 0.002 of that is 0.29 so 2.9bb/100.

Also notice that the rake in postflop play is 15bb/100 per player (so 30bb/100 total) which puts the accuracy numbers into good perspective.

OK , I am able to figure what commands I need to do by looking at the console output from PioViewerFree . show_tree_params is not working in PioSOLVER-free though at the moment ( maybe because is free version ) .

I loaded the tree first in PioSOLVER-free.exe so I don't know why show_tree_params is not working except that in free version the command doesn't work :

load_tree D:\tree.cfr
load_tree ok!
show_tree_params
board: not set

As you can see , it says load_tree ok!

hey punter, i will run a bunch of postflop trees on a dedicated server and i would like to know beforehand from which overbet size up does the solver starts getting trouble converging (you said it at some point)? I mean i will use a 200% overbet and in some spots the add allin option which triggers with a stack under 300% of the pot so overall i will not have obvertbets bigger then 300%.

when looking at a combo is there any way to lock it in the bottom right window? could be handy when comparing different strategies in different windows.

I am noticing some things with the bet sizes .

In the console I see : r:0:b25:c:2d:b104:b262
But in PioViewer : 2d : BET 79 : RAISE 237

I am not sure where b262 is coming from .

Second example :

console : r:0:b25:c:2d:b104
Pioviewer : 2d : BET 79

So the console is adding the pot when it says b104 ?

Third example :
console : r:0:b25:b75:b175:c:2d:b479
Pioviewer : BET 304

So b479 = b175 plus 304 for some reason ?

Having trouble deciphering this .

When you build a preflop tree as 'pure preflop tree (no flops exits)', does it mean that it gets checked down posflop or does it calulate the postflop plays without saving the flops to look at?

Hi,

If I bought PioSolver Pro, can I upgrade to the preflop version (edge) for the difference in price?

Please PM or post and let me know what to do and how to do it.


Thanks,
Evan

This is a bit advanced and not really useful for most people so it would be good to move that to emails or Skype (pm me and I will send my sn).

In short: those are part of the state which you set and then they are used to build a tree. If you load a tree they are not set by that. There are good reasons for that design but I can see how it's not obvious at first. The best way to start experimenting is to replicate what the viewer does exactly (by reading log.txt and repeating the commands). This way you can get the hang of things and start writing your own queries.

I am happy to help you with some details on either Skype or via pms/emails not to clog this thread too much with programming related discussion.

I don't know the answer to this question sadly. This is of course imperfection of the algorithm. In general the bigger they are and the least used they are the more problems they cause.

Sounds good. Big overbets are used mainly on the river so I would limit myself to "add-allin" on the river and maybe some bigger bets (150% of the pot or something) on the turn/flop instead to see if they are chosen.

Two ways:
-you can hold left click and move your mouse away, then it won't change on hover
-Tools->Configuration->Behavior->[ ] require click to update zoom (at the top)

then they won't change on hover but will require a click to change.

The solver always reports cumulative amounts internally (how much a player invested so far including that bet). This is easier to reason about/code for but it's not what people are used to that's why the viewer translates the bet amounts.

In your example Raise 237 is the same as :b262 as 262 is a total amount (so 25 bet before and 237 now which ends up being 262). The solver internally doesn't see a difference between bet/raise or check/call either (as a check is call 0). I hope that makes sense

It assumes a check-down postflop, that's why it's instant to solve.

Yes and it's already taken care of via PM.

Hello, I am going to buy PioSolver Pro, but I first need to buy a new pc to be able to run it.
A friend of mine is helping me build the pc since I'm clueless. Something about piosolver using multiple threadings or whatnot, but basically there is a relatively new processor, "AMD RYZEN 7 1700 3.0GHz 8-Core Processor" and do you know if that will work with pio / will be better than an "Intel Core i7-7700K 4.2GHz Quad-Core Processor"?

Thanks, Fats

How do I join the skype group? I can't find the name mentioned anywhere on the website.

While I don't have an exact banchmark for 7700k I do have one for Ryzen 1800 (it's like 1700 but clocked at higher frequency). Ryzen is going to be significantly faster from what I gathered.

There is a link in your registration email. If you are not our customer yet but would like to join the group please PM me. We don't monitor the group these days but it gets pretty good traffic and we need to remove old accounts on regular basis to keep it under the limit of 600 participants (this is hard limit imposed by Skype).

punter,

with "show_all_freqs global", does the output list all possible lines, even ones that have probability 0 of being reached? if not, how does it filter them? also, when a line says "nan" does that mean it has 0 chance of being played?

thanks.

Yes.

I think lines which are not in the tree (so in case of small saves) have nans displayed. It should be 0 (and not nan). Let me know if you see the case where it's not working as intended.

How far off would a good reg play from GTO? For example on one hand as the flop defender, I threw in what seemed like a reasonable strategy, and I was worse by 70 bb/100 to a high accuracy (.7 worse in a pot of 55 at 5/10). I wouldn't say I'm a good NL reg though, (PLO player).

That is just vs flop cbet though, I didn't input anything else. Roughly how far off from GTO would I be (considering 70bb/100 off flop) factoring in all the different lines, runouts, and betsizes? What about for a good reg?

This is certainly very surprising for a cbet strategy being as much worse. Feel free to share the configs (which you can get by clicking "copy to clipboard" button) either publicly or by PM. I would need a description of your "reasonable range" as well to re-run it

It depends what you mean by "off". There is big difference between being exploitable and playing a strategy which loses to a GTO player. While there is not way to measure it precisly my guess would be that every human player is exploitable for huge amount (like 70bb/100 or even more) but they certainly wouldn't lose that much to a GTO player who doesn't go for max exploit. I think GTO would still obliterate even the top pros (at least in game withou rake) but that would be closer to 15bb/100 or something in that ballpark.

Those are obviously wild guesses I may be way off on those so take them for what they are.

Sorry, I'm still confused about this, and not sure how to proceed. Thanks.

I think I got my decimals mixed up

how did you specify your strategy? asking because i'd like to play with that feature.

ty. small save was indeed the issue. what i did to fix it was load small save file, then build tree, then resolve. is this the correct method, or is there a more efficient way to do this? eg, to tell the solver to start with the saved flop solution, and fill in the parts that weren't saved?

It was the set strategy/lock node feature. The youtube faq's help a bit with it.

Hi Punter,

I got a strange error running a preflop tree. The tree completes and the strategy looks accurate but when I click on strategy and EV it says "PioViewer Error: Object reference not set to an instance of an object."

When I save the file and load the fold region disappears and is replaced with a call region, with all the EV's call EV's being negative. When I run the same sim on a small number of flops as a test things seem to be working properly, but when I run it on a large number this error occurs. Any idea whats causing this?

Here is the config file: http://pastebin.com/Pr6EJPtc

You can node lock on full trees but in case of small saves it's the best to just build the tree again, node lock before solving and solve then. It's in general recommended to proceed like that anyway (node-lock on fresh tree instead of already solved one). You just need to save the configs to be able to recreate the tree.

There is no way to do it right now but luckily it wouldn't provide much performance gain anyway.

First thing would be to make sure you are running the newest version of the solver (1.9.2) and the viewer, which you can download here:

https://www.dropbox.com/s/gesnfhgrbo...iewer.exe?dl=0 (download and substitute in your oflder).

This sounds really strange, are you sure you have enough space on your hard drive to save the files?
I will try to recreate, it would be useful to get exact flop subset which you are using which is causing the error.

I think this is the problem. I even started a new instance as I was getting this error on the last one I was on so not sure why disk space is an issue. I'll take it to the skype group as its more appropriate there.

Interested in piocloud, let me know if there's a better place to ask this. How is it figured? I'm looking at the sample for what I assume is the button raising 50%, and the big blind defends. It says:

"IPvsBB@100bb@50VPIP@30-105-235-1000@3.5per@14cap"

And their pdf description says it is modeled for 100bb vs population tendencies in regards to raise sizes.

I'm assuming 50 raise from the button was locked (range construction based on population), and then the BB response is calculated?