It's an interesting question which I didn't investigate much. Most of the exploitability is on the rivers (did experiments where I resolved rivers to very good accuracy leaving flop/turn intact and was often able to get very big improvement). To answer this question reliably though you would need to do a lot of research. It's likely quite dependent on the exact algorithm used as well.

No. One may hope relative exploitability (in % of the pot) is similar throughout the tree but there is no guarantee for that even.

The concept of "exploitability in node" need a clear definition. You may calculate sum of mistakes in a node but then sum of mistakes across the whole tree is bigger (and it's guaranteed to not be less than) exploitability. Maybe if you can tell me why you want the answer to those questions I would be able to give you some useful insight.

There are no guarantees what so over about how exploitability is spread across the tree.

Commands which make it possible to run scripts are disabled so for example "wait_for_solver" one;. You can still get all the functionality though as evident by the same viewer working with basic/pro/edge version (and the viewer uses the very same interface which is exposed to you and isn't a privileged program in any way).

oh well, im thinking that if somehow the flops are better then turns which are better then rivers in terms of exploitability, then if a tree is solved for 3bbs and saved as small save, then, when revisiting the river, we are looking at a better solution after recalculations then the initial 3bb/100.

This is in fact true. I remember testing a solution solved to around 0.75% and it got to 0.25% after resolving rivers to default resolving accuracy. I haven't measured the effect exactly across many tree types but I suspect it holds in general.
This makes sense as well: if the solver spends 200ms for resolving the river you want to see it's not a noticeable delay but it's a lot of time in terms of the whole tree as usually there are hundred of thousands of rivers so during normal tree solving there is no way a single river can get that much solving time. As you will only look at very small subset of rivers when browsing the tree (even if you browse it extensively for human standards) it's still a very small fraction of all of them.

Hello, i'm trying to make new software for my purposes, what displays what far i'm playing from GTO.

But i dont know how can i send these commands from my software

I tried:

or easier:

but it writes only pftables.dat is missing

have you some other advice how to connect it with my soft?

TY

Windows has a concept of "current working directory" and you need to set it to one PioSOLVER is in so it can find pftables.dat file.
Alternatively you can copy that file from your Pio folder to the folder your script executes from (as that's default current working directory).

Feel free to pm me so I can add you on Skype, I am always happy to talk to other programmers and help out as much as I can

Ok, so here's what I did: I write a script file, with standard commands, like show_strategy, show_node, ... I put that to a file like script.txt. I go to cmd, and run like this: PioSOLVER-basic19.exe "script.txt" And it does nothing. It works with edge. Can you please help me with this?
Thank you!

Yes, basic version doesn't do scripting so commands which are specific to running scripts are not available. This means that neither "load_script" or feeding a script as a command line argument won't work with it. You can however run commands one by one which is easy if you are writing 3rd party software to work with the solver.

Hi punter,

Am having a problem with the scripting. When I run the script on 49 flops, PIO includes rake on the first flop and then does not on the other 48. I tried to play with the script a little by manually inserting

set_rake 0.05 30

after/before each new board and didn't help any.

Here is the link to my script:

https://www.dropbox.com/s/0zfkq58247...cript.txt?dl=0

Please help as scripting is pretty much useless to me if I can't do rake! If there is a simple text line I have to insert after each board just give me the syntax and I would be happy to do it myself. Thanks!

It should just work. What version of the viewer are you using? (go to "About", should be 1.9.2.5) and what version of the solver (look at the bottom of the window when it says: "connected to...", should be 1.9.2.

I am running a simple script with rake and it just works.

I am on 1.8.2.3. I have no idea how to update to your version. When I run pioupdater161 I just get that it's already current:

https://gyazo.com/c4dd7cbd86ca6362fd17e0de9194a74e

I go through the updating process on the video, here:

https://www.youtube.com/watch?v=1PEmsGaGLvk (point 1)

It's likely you have Pio in two different locations and are running the older one still.
If that doesn't help please drop us an email with your registration key and I will double check if everything is ok with your license.

You were exactly right about everything-switching to the latest version fixed the rake issue and you were also right that I had it installed in 2 locations! Thanks for the prompt reply!

Now please take a moment to browse through new features in 1.9:

https://www.youtube.com/watch?v=zLGpcZavxeQ
https://www.youtube.com/watch?v=1PEmsGaGLvk

if you need to copy your ranges/configs from the old location you can just copy Ranges and Treebuilding folders from the old one to the new one and it will work

If you add rake to preflop sims, does it rake if there is no flop? Aka the times we steal blinds it rakes those winnings or not? Thanks

It doesn't.
The simplest way to test it is to go to a preflop branch where there is a fold, for example:

https://gyazo.com/8874da73d953360aaf838e3e856e65be

and verify that EV for both players add up to the full pot using those buttons:
https://gyazo.com/a298ea6ae906581d8ed00c373814189a

If on the other hand you go to a node where the rake is taken (for example all-in/call) you will see that EV doesn't add-up to the full pot (as the rake is taken).

Is there a smart way to use lock_node to lock my opponents' strategy for all nodes in the equilibrium?

I am doing some modeling where it would be useful for me to lock my opponents' whole strategy to the equilibrium strategy. I start by calculating the equilibrium strategy, then I would like to lock my opponent to it, then I will make some simplifications to my strategy with my opponent not changing his.

I do this to model a decent-playing but not 100% aware opponent. I'm trying to model a human who plays well, but is not paying attention when I make deviations from my equilibrium strategy. He keeps playing his, which allows me to simplify my strategy without getting insta-punished for it. For example, I might check my whole range on some turns where I should check most of my range, but not all of it.

Now, I could look at individual turn/river runouts and use node lock, but the way the model is set up (I scan a lot of turn/river runouts) it would be most practical for me to lock my opponent's strategy for all nodes before I make changes to my strategy.

No, but locking all nodes doesn't make any practical sense.

You don't need to lock your opponents strategy for that. It's enough to change yours and then without resolving click "calculate results" in tree building and calculation tab.

Again, there is no point in locking anything. Just change your strategy and look at new EVs

Here's an example to show what I mean:

3BP, I CBet flop 33% OOP with my whole range and check range on certain turn cards. My checking range is now stronger than in the equilibrium solution where I use all options. When my opponent stabs turn unaware of my changes, I can check-raise more than in the equilibrium solution. I want my opponent to play under the assumptions that I'm using all options on all turn cards, so I start with an equilibrium calculation using all options, then I lock myself to check range on certain turns.

I thought I had to run a resolve to see what my new maximally exploitative turn XR range should be for my modified turn checking range against my opponent's static turn strategy.

If I do need to resolve, I need to lock my opponent (if I want him to be unaware of the changes I've made to my strategy). If I don't need to resolve, I'm dandy. :-)

Hi, I watched your preflop tutorial and was wondering - If I want to solve Fixed Limit preflop, what do I put in for bet size% and raise size% for post flop betting?
Is it feasible to run limit holdem preflop calculations with simplifications like no donk betting and no preflop 4 betting?

Thanks

Unfortunately this is not possible right now. It will be possible in the next release (as it works in development version).

There is an option to insert fixed bet/raise sizes using "c" postfix like so: 100c, 200c etc.
Full preflop limit tree is here:

https://pastebin.com/dJPATZnY

To use it copy the text then open the viewer and go to Tools->paste treebuilding config
Look how postflop play is filled in and start your experiments from there (limit tree will need a lot of cutting to make it reasonable in terms of size as they are humongous in full form).

Thank you. That just saved me a great deal of time.

Hi, I have the basic version 1.9.2 which has a preflop tab but when I go to estimate size I get a message telling me to manually connect to the preflop solver. Do I have such a thing or do I need to update to edge? I didn't see preflop solver in my pio files.
Thanks

The preflop solver is only available in the edge version. It also requires very big hardware to run (at least 64GB of RAM for practical cases, you can do some simple ones under 32GB but it's very limited) and fast modern CPU. Most of our edge customers rent dedicated servers/cloud instances for the purpose.

Pio/basic version can read preflop saves though as well as solve pure preflop games (assuming checkdown postflop).

When running a script with x flops, how is possible to extract only A high flops? I want to see how pio is playing on these flops only and not in all 300

Are you working on 3way solver?