There is no reason why Mason or I should have been expected to carefully vette an article by a well known poker author about a subject we are not that familiar with. Especially if we knew we would be reprinting it in a thread populated by experts and any errors could be discovered through logic and knowledge. If our magazine was concerning medical breakthroughs that couldn't be refuted through mere thought we would be more diligent.

What iteration do i run postflop ICM calculations to Piotr??

Is 0.1 too much? Thanks.

If pio was churning out results that could be so easily refuted and shown as bad when the engine and similar programs are the basis for the game plan for pretty much every elite player then that would be a spectacular scoop.

It would mean near every high level thinker In the game had missed something that could be refuted by the 1-alpha minimum defence frequency concept in a 2 minute video or short article. The 1-alpha concept has been seen as a horrendously bad idea for many years now.

It would mean that every high stakes crusher could be easily exploited and beaten by basic Harrington on Holden era poker play and thinking

Would you also print an article questioning the pokerstars dealing algorithm because a player who wins live loses online and can demonstrate the rig against him as he downswung the one time he withdrew? Can any claim make to print?

This article surely does not pass the most basic sniff test, and you'd have saved yourself and the author a lot of embarrassment and loss of reputation had you done a scintilla of research

It's reasonable but I think your judgement failed you on two points here:

1)You and Mason are both more familiar with the subject than you think you are. I've read a lot of your books when I was first learning about poker theory (16 years ago!) and my impression was always that you are both careful about claims and your standards for math/logical thinking are high.

2)The author being expert on the subject. He is not. I am not sure how comes you think they know more about it than you and Mason do. I have a lot of respect for you based on stuff you published 2 decades ago even though it's already printed and you didn't get a chance to update it based on modern tools. It's just interesting to me that your assessment is that the author is more of an expert than you are.

There is no harm done. We are way too big a player to be hurt by a random ranting youtuber these days. It's just interesting to me that someone so out of touch goes around publishing books/making videos without even bothering to ask around or contact us asking for help using the software. Maybe it motivates me write a bit more about poker theory. I used to think it's the task better left to others but for occasions like this it would be nice to have something on the record I can link to.

May I suggest that you might be putting too much weight on academic titles? They don't really mean much. It's a different game with different goals they play

Literally spit coffee out reading this. Well played!

This isnt really important, but i dont know anything about programming and maybe it only takes like 5 min, any chance we could get a dark mode pio : >

Final comments on Pio

Normz

Given the level of criticism I received from the Pio article I published, I felt compelled to run Pio much longer, for roughly 2 hours, seeking an accuracy of .1%. The result never got to that level of accuracy, but the results I got verified my complaints as well as Pio’s accuracy. All of my complaints (which were based on running the program for 100 seconds on a super fast computer), disappeared after two hours. Pio bet the pot with KQo roughly 100% of the time, which is what I contended was correct, which the two-hour run confirmed. Pio was not folding with 99 at all in response to a pot-size bet by the IP, which I claimed was correct which the longer run verified. On the river, while Pio was throwing away TT66 some of the time on the river and calling with 6622 some of the time, that didn’t matter because the IP was not betting any intermediate hands like 7766 or 8866 where that calling strategy might make a difference.

I still believe that in actual play, it is better to call more with quality hands and less with ragged hands, because your opponent will likely not be following Pio’s strategy.

I want to apologize to the creators of Pio for creating the rukus, but in defense of myself, I want to remind everyone that I followed Pio’s suggestion of using a .5% accuracy. So we were in effect, both correct. I was correct that at the accuracy Pio recommends, the strategy was non-optimal, but I also now agree that if you run Pio Solver long enough, it appears that Pio is generating a GTO strategy.

Is it possible to get frequencies for all the boards in a subset at a certain node using the aggregated reports? For example, if I browse in BB vs BN to [xx][xx] and on the river, will I get overall betting frequencies for the BB here averaged across the turns and rivers? Thanks!

All the colors in PioViewer can be either configured on the viewer's side or the viewer uses Windows theme colors. This means that trying Windows dark mode may give you what you want.

Yes. You will also get (if you choose to) frequencies/EVs for all specific boards.

Hi, thanks for spending a bit more time with the solver and verifying the results.

I encourage you to think a bit more about it. It's not only that it doesn't make a difference, calling those weaker hands is actually objectively a better strategy! I've explained the reason for this in my previous posts. It's interesting and one of the things about poker while not exactly discovered by solvers but certainly made more well known and understood thanks to them.

There must be some kind of misunderstanding there. We don't have a default timeout nor default accuracy. Maybe someone who set up the software for you copied their private settings as well which made you believe those are the defaults. I am also not sure about your hardware and what you consider a very fast computer. Even though this is a really big tree (because of double bet/donkbet/raise sizes everywhere) it takes 5 minutes on my machine (which admittedly is very very fast) to achieve pretty decent accuracy of <0.5%. If you don't use a timeout and actually get to that accuracy the most glaring errors will already be gone leaving only small imperfections in the solution.

I think we are all spoiled at this point expecting instant precise results for even very big trees. Not so long ago it was an achievement to even calculate all-in equity for range vs range in less than 10 seconds let alone solve anything. The tree you were solving has almost 2.5M decision nodes and at every one of them Pio needs to calculate strategy for 200 or 500 hands (from initial ranges). Finding equilibirum requires multiple (usually 200-500 for every player so around 400-1000 total) iterations of playing out all possible hand x hand x board combinations (around a billion per iteration), calculating EVs for all hands in all decision nodes and adjusting strategies there. It's a lot of computation even for modern computers.

We hope we can make it significantly faster in the near future but it's still far away from providing instant feedback and maybe that's for the better for practical players.

Is there a way to copy and paste ranges that were made from tools -&gt; preflop charts to actual sim? When I loaded the .txt files in ‘Ranges’ folder nothing loads.

Thanks

Hi , can the CFR solution file be oppened unencrypted in text file, or make agregated report for all the solution, not just for 1 node?
Thanks

Our saves are not encrypted. They contain parts of binary dump of the tree.
It's not possible to open them as .txt because:

1)They contain only minimal information needed to recreate the solution, most of it is recalculated on the fly when you browse

2)Dumping all the data would result in files 5x-10x bigger than a full tree in memory. That means gigabytes or tens of gigabytes of data per tree

3)It would be very slow to work on this kind of data. It's much better to use solver's API to inquiry about the tree. See here:
https://cdn.shopify.com/s/files/1/07...68544248083387

Preflop charts files contain more than one range.
You can open them in "Menu -> Tools -> Preflop Charts Editor" and copy ranges from there.
You can copy range either by right-clicking grid and selecting "Copy Raise 70" or by clicking on the range icon.

That worked, thank you!

I need a third activation for my piosolver pro. Is it possible?

Thanks

Inviato dal mio SM-A750FN utilizzando Tapatalk

Why are some boxes only partially coloured?

I don't know how to start a new subject so i'll just ask a question in this reply if you don't mind:

I want to use node locking in PIO but if I click: 'set strategy and lock node' PIO says: 'setting and locking strategy doesn't make sense on incomplete tree. Do you want run plugin anyway?'

That means it wants to know the turn and river right? But that's not what I want.
If I build a new tree with the turn and river already known I think I only get the advice what to do on the river, but I want to know what to do on the flop if villain has a certain strategy. Can somebody please help??

Thank you in advance!

Yes, I've answered your email. I am not pasting the link here as they get removed anywya.

I can't see the screenshot but I guess it's because checkbox "square size proportional to weight" is checked.
Please see here:
https://www.youtube.com/watch?v=5Ozjfnatwms (question number 9)
Well, the problem is that once you change the strategy in the tree it's no longer possible to calculate new EVs (or solve it) without having the complete tree.
It's not so much that it wants to know them it's just that without them in memory it won't be able to solve for new strategy.

The problem is that a small save that you are using doesn't have turns/rivers in it to save memory. You can still see them when you browse because they are recalculated on the fly. To use node-locking you need the full tree. There are two ways to go about it:

1)Build a new tree, lock, solve
2)Load small save, press ctrl+b
Type:
rebuild_forgotten_streets
[hit enter]
Now lock and solve

1) Is recommended. 2) is not recommended and may backfire badly if the full tree was bigger than your RAM (for example if someone else has solved for you on a computer with more RAM)

Thank you punter, but it doesn't work, I don't understand. I made a new tree.
Can you maybe tell me exactly which actions I have to do at what moment and in what order to let my opponent call more combinations than PIO wants to in a certain tree?
Sorry!

Curious how some of you guys study frequencies - do you just study via reviewing hands and then rely on flow to make decisions in game? Or do you have some sort of flashcard set that you use (quizlet, etc.)

Thanks!

Is it possible to get lightweight version of PIO, so that I could use some browsing tools on third computer? I have edge license.

hi, I have the Edge version. I have a question: let's imagine that I want to study a specific situation (BTNvsBB) and have created a 12GB postflop tree (I have 64GB) and want to use a script for all 1750 possible flops to then create an excel table. My problem is: since each tree will be about 12GB I don't think I have enough memory (I have a 1TB HD) to save everything on the PC and then create a report. Is there a solution to my problem or a way around it? (for example create the report without saving the simulations?) thanks.

You can only lock a specific strategy. Here is how to do it:

1)Build a tree
2)Navigate to the decision node you would like to lock
3)Choose the strategy you would like to lock
4)check "set strategy and lock"
5)solve the tree

If you have a small save you can do one of the two:
1)Load a small save
2)Build a new tree using the config from it, so "build tree" button in the treebuilding tab
3)Follow from point 2 on in the previous instructions

or:

1)Load a small save
2)ctrl+b to bring up arbitrary solver command
3)type:
rebuild_forgotten_streets
[hit enter]

4)Navigate to the decision node, lock, solve

You can move the license without any limits so you only need a third one if you need it on 3 computers at the same time.
Alternatively you can buy another activation so you have 3.

It's recommended to use small saves. Those are 50x-150x smaller than full trees so that should fit easily. You can reports on them. The only problematic one will be on the river as that will take a very long time to complete and is not recommended. As long as you need ones on flop/turn they should be very quick (actually faster than with full saves)

Hi , i need to send the strategy in text format in .txt file from Piosolver cmd, i tried whith
show_strategy_pp r:0 , but it only write it on the console,is there any comand do send the strategy on text file ?
Thanks

Thank you very much for the support Punter but I don't understand what you mean by step 2. English is not my own language and I'm terrible with computers. I learnt in a video on the internet that you have to:
- build a tree, then
- click 'browser' and click the action, one step before the action you want to change
- click 'Tree' - 'Set strategy and lock node' from the menu
- check 'fixed' under the action (in this case call)
- go to the range next to the action and adapt the combos
- OK
- check 'lock all hands'
- click 'Set strategy and close'

But nothing changes I think.

I only understand clicking certain actions so I'm quite a challenge I'm afraid...