I think you missed my response above.

As stated above, duplicate flops are ignored.
I'm not seeing any glitch of flops being incorrectly ignored though.
Can you send such a .txt file to support?

We offer predefined subsets.
The following subsets are available: 15,19,23,33,37,44,55,66,80,89,111,129,141,163.
To load a subset, enter the desired size and click on "Import flops from file".
See the screenshot below.

This is intentional.
Flops are sorted from high value to low value.
So a 32K flop will be turned into K32.
As for suits, they are ordered sdch.
So AhAcAs will indeed be turned into AsAcAh.

Sorry, I did miss some of your replies. Thanks for your help. I was unaware that GTO+ would just automatically ignore identical flops. I presumed it would just parse each flop as written. I can see that changing the suit of one card makes no difference when players play ranges comprised of either suited or unsuited cards and treat no suit differently to any other.

The reason I was looking into this was an interest in breaking down flops into "different families" and examining how ranges play against that family grouping. For example, a HML family would be High Medium Low cards like K95 or Q84 or J92. Presumably, whilst rainbow flops and monotone flops will be considered identical by GTO+, will two-tone boards also be considered identical? Whilst I can see that Kh9d5d and Kh9h5d and Kd9h5h may well be considered identical, can the same be said for Ah9d5d and Ad9h5d and Ad9h5h? Surely the effect of having the Ace of flush draw on the board is much different than having an Ace on board with two other suited cards?

Before posting this as a question I created a text file with:
Ad9h5c
Ad9d5c
Ah9d5d
Ah9d5h
and GTO+ displayed all four, so clearly treats them all differently.


Just trying to drill down a bit in my understanding of what GTO+ needs and excludes flop-wise before I spend anymore time on this flop idea.

I'm interested in looking at how ranges perform against certain types of flops and thought that running GTO+ to see what it says and then using Play Against the Solution might provide a deeper understanding of the diffferences between flops, than using subsets. What do you think?

Anyway, thanks for your time and many thanks for your patience!

The second flop isn't identical to the other two. Which cards have the flushdraw matter.

Same thing, the second flop isn't identical to the other two.


To be clearer, replace the suits with XYZ instead of actual suits:
For example KX9Y5Y
Any board that follows this format is the same :

Kd9h5h
Kd9c5c
Kd9s5s
Kh9c5c
Kh9s5s
Kh9d5d
Ks9h5h
Ks9c5c
Ks9d5d
Kc9h5h
Kc9s5s
Kc9d5d

are all identical. They are all K of one suit + 9 and 5 of another suit.

You can filter the database when you are viewing it :


So you can still solve a big subset with all types of flops, and filter the ones you want to study as you need. The filters also apply when playing against the solution. (if the filters are not precise enough, you can right-click a flop in the database to disable/enable it)

Why have your videos for GTO+ been age restricted for so long on youtube, but not videos for other poker solvers? Do you think you can get the age restriction removed?

YouTube has changed its policies recently, and we felt we had to comply and set an age restriction.
They tend to close accounts for the slightest infractions nowadays, and we would like to avoid this.

hi I thought by doing a full save which takes quite a lot of space that it would be possible to see the avg GTO frequencies for turn and riv spots too.

such as sb vs bb , sb bets bb calls. Then on the turn sb bets bb raises - is it not possible to see some avg gto freq % for this node if we run a dozen flops? I know it's possible to see all these things on the flop though.

ty.

It's possible to see the average frequency over all turns for a single flop. For this, check the turn report for a tree and navigate to the decision that you're interested in. The average frequency is displayed below the table. We will probably also add this for aggregate reports over multiple flops, although it's important to note here that this data is significantly less relevant. The problem here is that these lines are reached with completely different ranges and frequencies, on a completely different board; so although it's possible to compare these values across different flops, these are not values that are suitable for being compared to one another.

I'm modeling a toy game where both players have AA-66, with starting pot 100 and starting stacks 100.
Flop 22223.

Solution (dEV =0) has OOP check 100%, IP bets Pot 1/3 of his range (AA-KK, 66) and IP calls 37.5 %.
But when we nodelock the same strategy for IP, OOP calls now 54.4 %.
Why is OOP response different, when IP strategy is well defined and OOP is clairvoyant in both cases ?

forgot to mention : OOP is forced to check

I've been playing around with that, and I think I understand.

If you lock the IP strategy (betting only AA-KK and 66), then all the bluff catchers of OOP (QQ-77) become equal and call 50% of the time.
But if you were to lock that OOP calling range, unlock the IP strategy and rerun the sim, then IP would start betting more: QQ and JJ become profitable bets because you sometimes call with TT-77.

So essentially, when you lock IP strategy, you have multiple strategies that are all equal (you just need to call AA-KK and 50% of your bluff catchers), because you know that IP will not adapt his strategy depending on which bluffcatchers you chose (since he's locked).

But when IP is unlocked, if OOP starts choosing his bluffcatchers differently, the IP player can exploit that. Only the original strategy is unexploitable.


It seems that when you lock part of the strategy, the solver stops looking for an unexploitable strategy and just goes with the average one that gives the most EV, since it knows the lock prevents it from being exploited.

Yes, as rogorz stated, if IP's strategy is locked, then OOP can play QQ-77 in any way that he sees fit. Given that calling and folding have the same EV, it doesn't matter how they are played. When IP's strategy is unlocked, opportunities would arise for him against most QQ-77 strategies. In this case OOP will need to find a specific GTO strategy.

1: Drill a specific line in database mode, not only vs the current tree
2: A slider bar to we try to represent the frequencies (I know all mixed strategies have the exact same EV vs a GTO opponent, but this feature would make more intuitive to us to train to recognize wich hands are already mixed and wich takes a pure action in a specific spot)

Hello,

I've recently bought GTO+ and I'm using it to solve Spin & Go spots.

However, most of the time I have to deal with HU limped pots.

Considering cbetting the flop when SB, i have many choices such as 20%,25%,33% pot etc in advanced options.
However, it comes that many betsizes are under 1bb with thoses choices, and I would like GTO+ to consider "1bb=minimal betsize" in order to get good ranges with thoses betsizes

In many tutorials of piosolver, i saw people indicating that there is an option where a box can be checked with "Minimal betsize=1bb" to ensure that even if many betsizes are simulated the solver corrects itself by putting a good solution, whenever 20% pot, 25% pot for example are selected.

Is there any kind of similar option in GTO+ ? ty

1. This is somewhat tricky, given that you wouldn't actually be drilling a single spot, but a different spot for every flop. If "Basic" storage is used, and the hand is on the turn, then a recalc would be needed for every tree that is drilled.

2. We'll consider something along these lines for future releases.

An option like this has been added for v134.

Hi Scylla,

During a "cut-paste" operation I have just lost a very important folder where a bunch of .gto saves were inside. Unfortunately I didn't have windows backup enabled so I can't restore them that way. I tried to use a recovery software and it found a bunch of .rng files which are associated with GTO+ I assume, but I am not able to use them in this format (it found no .gto files). Is there a way that a .gto file can be reconstructed from this data or should I give up and accept this loss?

I'm not familiar with .rng files, so I don't think your gto files can be restored from this format.

hi Scylla, is there a way to nodelock every node for one player w/o manually doing it?

Currently attempting to start automating some of my analysis using python. I've been having some trouble interacting with GTO+ using pyautogui specifically, while all other programs I've tested so far interact as expected with the package. If I wanted to automate node locking specific strategies to compare to the GTO EV, is there a good place to start or am I 100% on my own?

Hello, what's the max number of cores/threads that gto+ supports?

I don't have experience with pyautogui, so I don't think that we can offer anything here.

We don't have a function specifically for that.
Do you mean that you want to lock all nodes on the flop for one player?