If I only input the flop, I get the strategy and ranges on this street, does it shows a strategy for most turn cards?

Yes, for any possible turn cards, and then for the river as well.

Hi,

3 small requests.

First,
is it possible to somehow to allign 4 different windows when they are openend with 'open in new window' so they are displayed all 4 together on 24 inch monitor ?
This request because the option in Windows 'Show windows side by side' does not seem to work.
Essentially when the Windows option would work this would fix things.

Secondly,
Is it possible to reduce the names in right bottom screen so everything is still visable when we make it smaller?
ie
R3850 # -250,385
instead of Raise 3850: -250.385
c # -14.872
instead of Call : -14.872

Maybe an option to move the description in the grey area and only insertion of the numbers is even better.
Or an option to move the description to a button in the grey area and only insertion of the numbers.


Thirdly,
Is it possible to insert saved file-name in the header of the window after 'open in new window is clicked'?
Also is it somehow possible to include the specific board @ the node after 'open in new window is clicked'?
This because I just am now used to open a first instance to equate everything while I open with a 2nd instance all my different trees and after 4 different preflop ranges OR 4 different turns; it becomes a mess to remember which screens correspond to which preflop ranges/board combo's. With the saved file-name (if present) + board in each window-header this is no longer a mess.

Hi Emus,

Those are all very reasonable requests. I will put them on the to-do list but I can't guarantee any timeline as we have a bit different priorities now.
I am happy to see you use the solver that heavily. I didn't expect this kind of request use case (2 cases open separately etc.)

Lol, whoops, did not know the program does that. That is really convenient. Thx

Btw, when we save a build tree. Where does the file goes? I have save some trees and can load them but cant find the physical files on my computer.

Np, I always try to reach out with things that would make software for me personally more of a gem compared to coding the things myself (AHK like or everything from scratch). Meanwhile I try to take effort to only post things that would also mean for others an advantage.
Sometimes I see the coders interested for a while and there arrives a synergy.
Other times I start to realize the coders are not really interested anymore and I stop reaching out.

So this situation becomes rather predictive

This is a mistake I made: if there is a space in the directory name (like c:\my documents\...) it won't work and save the files to c:\my).
I already ship the corrected version now but it will be fixed for everybody in the next solver/viewer release.
If you need the fix now, contact us at piosolver@piosolver.com and I will make you a better compile.

I see your point. The problem is that our resources are rather limited and we need to set priorities. As of now they are elsewhere but I hope there will be some time in foreseeable future when we come back to your requests.

On really tiny thing: when I am at one point in the game tree, and click on another point, then whatever is in the 13x13grid at the moment (e.g. Strategy) does not get updated, i.e. I have to click on "Strategy" once more. Unless this is intentionally so for some reason, I think it would be smoother when clicking in the game tree updates the information in the matrix as well.

It is intentional. The reason is that some things take time to calculate (especially EV) and you may want to avoid waiting (just want to navigate to some further node).
That being said adding that as an option is I think very good idea and I put it on to-do.

Here's an interesting video about color and I think PioViewer could be improved by using these ideas.

I set the "Number of colors in EV gradient" to 2 and set EV0 to 255 0 0 and EV1 to 0 255 0. This is the result



The formula PioViewer uses to calculate the gradient is naive. Half way through the average value is 128 128 0. This is this shade of yellow



As you can see, this is quite a lot darker than the RGB values of 255 0 0 and 0 255 0. The middle of the gradient is very dark compared to the rest. If the method described in the video was used it would come to 180 180 0. It is derived in the following way:

sqrt( ((255², 0², 0²) + (0², 255², 0²) )/ 2 ) =
sqrt(( (65025, 0, 0) + (0, 65025, 0)) / 2) =
sqrt( (65025, 65025, 0) / 2) =
sqrt( (32512.5, 32512.5, 0) ) =
(180, 180, 0)

The resulting color looks like this


This color is much more in line with the brightness of the other colors and would make the gradient much easier to interpret.

I think it's key to get the color gradient as accurate as possible because it is used to figure out EVs. It would also mean that you wouldn't need 5 colors to get a clear gradient from red to yellow to green (currently in PioViewer there are 5 different EV colors by default).

While on the subject of color... I am color blind (red-green, iirc) and the above red-green gradient is my absolute worst nightmare. 90% of it looks like the exact same color to me.

colorblind friendly contrast colors are always appreciated here

Yeah, fortunately it's easy to configure. Probably the best would be to choose 2 colors which are very different and which you can see better (maybe something like dark blue and light blue?).
I have no idea what would be the best here but we will try to get some feedback and maybe make some color-blind friendly settings. Maybe you would like to suggest what kind of gradient could work for you?

Could someone please explain what the view "Strategy and Ev" displays?
The cells are cut in two rows, second row sometimes has few colors, but they do not match the ev colors you get in the "Ev for Hero" view.

I remarked a small display problem:
When you look at the Root of the tree while a computation is running, each time you hit the "root", the percentages are updated (Bet or Check). But if you click on the "Bet" button to see the range for betting, the Bet percentage is not refreshed, only the number of combos for betting is.

In the 13x13 grid you can see strategy. The zoom view is divided horizontally. The top part is the strategy again while bottom half is EV with EV of actions. Remember that Hero means OOP and VIllain means IP. You can change those names to IP and OOP and new names are default in the newer version of the Viewer (available in your download folder).

If you click the big red rectangle with BET on it (and amount of combos underneath) the range will refresh. To refresh the strategy you need to click "Strategy' or "Strategy +EV".

Hey punter, when starting the tree from the turn I cant find a simple way to select specific suits. At the moment Im manually typing in the suits, for example Ac5c,Ac4c. Is there a simpler more efficient way to do this?

How would you use PioSolver to check if a hand should be in a range or not?
For example, suppose i want to check if 56s should be in the 3bet calling range.
Doing Ev computations for Hero's range without 56s shows a global profit of about 1.25% on most flops without a 6, but I guess there is nothing like a hand weakening your range? I mean that is not a good reason not to play it.

The right way would be to compare the ev of folding 56s in the CO (0) with the ev of betting it. But how PioSolver will give me the ev due to fold equity? Is there a way to estimate this and compare with the bad ev of the hand when called or 3bet?

the only way (right now) to check for this is to figure out whether the ev of 65s is greater than zero. so eg say you open to 2.5bb and get 3bet to 9bb and call, the pot now is 18 (in case of hu) -- if the average ev of 65s over all flops is greater than 6.5bb, 65s can (should) call the 3bet.

that means you can either run a 3bet spot with the ranges you assign (including 65s in the call3bet range) over a) all 22,100 unique flops and average the ev by multiplying with 1/22,100, or b) over all 1,755 strategically different flops and average evs by multiplying with the actual frequency of those flops and over all combos of 65s, or c) find a subset of flops (with corresponding frequencies) that has similiar properties to the full set of flops and run the solver over these.

Unfortunately it's not possible for now. Solving real life games on the flop was a priority so features for toy games are lacking.
If you would like to copy-paste ranges from different tools just let us know which ones. Maybe we can make the solver accept them.

Samooth answer explains it well. It's not in core functionality (we solver postflop). While preflop tools are great feature and I would love to see it sooner rather than later in PioSOLVER it's not a top priority as of now.

I've recorded my review of PioSolver that I'll be attaching below. If you're watching from your computer i'd recommend changing the playback speed (gear icon >> speed) to 1.25 or 1.5 to save you some time.

I want to thank Punter for taking time to chat with me on Skype about the program. He seems like a good guy and an incredibly smart person. I don't doubt that we're going to be happy to see a lot of the upcoming updates to PioSolver.

I absolutely think that the potential to earn your money back+ on the purchase of this program if you use it wisely - and I would highly recommend it to any players who were serious about improving their understanding of poker. I'd just add the caution that you can't be lazy about your approach and cannot assume the responses you are getting are a truly optimal strategy against all players.
I can't stress enough that, IMO, the value of this program is more in learning concepts from digging into how the ranges are constructed rather than from implementing the strategies exactly in real games - but that value is big.

without further ado..

nive review, SiQ. one point that caught my attention is @6:35, scripting PioSolver. you mention you're not "savvy enough to figure out how to do that" -- I felt the same until i wanted to run a spot on 10 different flops and actually looked into the documentation. the good news is, it's extremely easy to do! you can find a good overview under http://piosolver.com/pages/text-interface-scripting, especially the third sample script is helpful.

in order to script PioSolver, do the following:

- open an empty txt-file
- open PioViewer and set up OOP and IP ranges
- in the range selector, you can click on the Copy/Paste tab and choose Copy in PioSolver format (this will give you the range with just weights)

to set up the spot, script it like this:
- first line in the txt-file, type "set_range hero" and the paste OOP range after that
- new line, "set_range villain" and paste IP range after that
- "set_board" and type in the board after that, in the same format as you would in the viewer
- "set_bet_sizes" (same as in the viewer, cumulative etc)
- "set_pot" 0 0 X with X being the pot size
- "set_donk_bet" 0/1 (1 meaning yes)

to run the spot, type:
- "build_tree"
- "go 300 seconds" (or how long you want to run it)
- "wait_for_solver"

to save the solution, type:
- "dump_tree C:\PioSolver\test.cfr"

- to run a new spot after that, you can either redo all the previous steps or, if you want to run the same spot on just a different board, you can type:
- "set_board Xh Ys Zd"
- "rebuild_tree" (this will rebuild the tree as in your last setting, but consider the new board)
- "go 300 seconds"
- "wait_for_solver"
- "dump_tree C:\PioSolver\test2.cfr"

and so on.. example script with incomplete ranges that runs a 3bet spot on two flops:

_____________________________
set_range hero 1 1 1 0 0.1 ....
set_range villain 1 0 1 1 0.5 ...
set_board Ah Ks Qh
set_bet_sizes 10 30 90
set_pot 0 0 20
set_donk_bet 1
build_tree
go 600 seconds
wait_for_solver
dump_tree C:\PioSolver\AKQ.cfr

set_board 9h 8h 7h
rebuild_tree
go 600 seconds
wait_for_solver
dump_tree C:\PioSolver\987.cfr

exit
_____________________________

save the txt-file in the PioViewer folder. open PioSolver and type "load_script nameofyourscript.txt" and hit enter, it'll run the script (the one above will take 20 mins), save the solutions and exit the app. you can now run as many different spots as you like in an automated fashion, but be sure that your cpu isn't overheating.

edit: also be sure you have enough space on your hard-drive

samooth,
that's awesome. thank you.

Would be happy to pay $250 for a version that could solve games from the turn on with 2 or more different bet sizes available for both players at all decision points.

Is this likely to be available?

It will be available in the next update. Our goal is to allow for building arbitrary trees but we will probably start with something a bit more limited (still allowing you to insert most cases with 2-3 bet sizes though).
We haven't decided yet which version is going to get what though. Your best option is to wait and see once it's released.

Thnx for the scripting tutorial.

What is the syntax for the ID of the
show_node ID
show_children ID
show_strategy ID , commands?

If you use the GUI you can see the line description just above Board. It's string of letters like:

r0bc8hbcJh.

It is update every time you navigate the tree and is constructed in the following way:
-r always means root or the beginning of the tree
-r0 is always first decision point in the tree
-after that "b" means raise or bet and "c" means check or call; the board cards are written as 8c/Jh/As etc.

If you want to show the first decision node you do this:
show_node r0

If you want to show a decision node in response to a continuation bet (in a full tree with OOP bet on the flop) you type:

show_node r0cb

If you would like to experiment with those (although this is mainly for programmers, you will have trouble making use of this information without additional tools) you can enable logging (tools->configuration->enable logging). This will produce log.txt file which contains whole communication between the viewer and the solver. The viewer asks about information using node IDs.

The format will change in the next release. It will be very similar but there will be a colon between actions/cards and there will be bet amounts (to handle more than one). It will be something like:

r:0:b100:c:8h etc.