Hey Scylla! Loving your software. I have two different range configurations in Flopzilla and GTO+. Is there a way to combine the two different range files from say Flopzilla (newdefs2.txt) and GTO+ (newdefs3.txt) upon import? Or is the only option to replace current range configurations upon Importing?

E.g., say we have ranges a, b, c in flopzilla and ranges d,e,f in GTO+. Ideally I would like to combine these ranges so both flopzilla and GTO+ have ranges a,b,c,d,e,f.

It's possible to import the Flopzilla ranges into GTO+.
For this, place Flopzilla's newdefs2.txt file in the same directory as GTO.exe.

Then:
1) Start GTO+
2) Select "Settings->Import preflop ranges" in its menu
3) Select newdefs2.txt
4) Select "Import into sub-category"
5) Click on "IMPORT RANGES"

Flopzilla will offer the ability to import GTO+ ranges in its next update, but this is not available yet.

Ok, I will see what I can do for later releases.


If your EV is $2.60 and BB is $2, then your EV is 1.3BB. And if you were to play that exact spot 100 times in a row then this would make your EV 130BB. So an EV of $2.60 comes down to an EV of 130BB/100.

That being said, BB/100 is typically used for calculating your winrate. It's used to check how many blinds you win on average for every 100 hands you play. This value can then be compared with the winrate of winning players to see how you perform compared to them. So BB/100 is a method for estimating how many leaks there are in your game. It's possible to calculate BB/100 for postflop spots, but it doesn't have a practical application here that I know of, given that no long term data for winning player exists that you can compare your results to.

Ok, I have created an internal update that will auto-convert Mac paths to the C:/ format. Also, it will give an error in case the path is cut off before /drive_c/. And spaces in paths should no longer be an issue. I'll send you a PM with a link.

Great thank you!

Hi,

Im looking to buy a new laptop. I was wondering how well does gto+ perform on i5-8300H and if there is a real benefit in buying i7-8750H chip.

Are there other factors I should consider? 8mb ram is enough right?

Thanks

I would expect it to function perfectly well on the i5-8300H chip. The i7 one would be a bit faster, given that it has 6 cores as opposed to 4. From what I can tell, the price basically scales with the speed here. As for memory, I'd go with 16GB, given that Windows and other processes will take up about 5GB. 16GB seems to be more of a default nowadays for modern systems.

How much of a dealbreaker would that be? Im looking at Dell XPS 15 and 15gb Ram model costs almost double the 8GB one. Given That I dont play games, paying for GPU seems reduntant.

I had 8gb on my desktop and gto+ worked fine.

Thanks

Some people like to build really large and complex trees. If this is not the case for you then 8GB should be fine, especially if you felt it was sufficiently previously.

Hi Scylla - I recently purchased GTO+ and I noticed in the previous parts of this thread, I saw screenshots of 'database distributions' and aggregating flops. Are you able to tell me where I can locate this?

Much appreciated.

For this, create a database of the same tree, but different flops (please let me know if you need instructions for this). Solve the database, and, after that, to see the aggregated data for the database, click on the graph icon (see sceenshot). Clicking on the "Aggregate" tab will take you to the aggregated data. There's several output modes available (see screenshot). For those, click on the small output icons in the upper right.

Hey Scylla any plan on making that agreggated who shows the lines and frequencies on the flop include turns and rivers too?

Enviado de meu XT1033 usando o Tapatalk

Thanks for getting back to me so quickly Scylla. Yes are you able to give me instructions on this? Also, do you have instructional videos on this? I'm happy to just watch those.

Also, what do you mean by "creating a database of the same tree" - I'm confused by this. Apologies if these are stupid questions.

Cheers

Scylla - I just ran a solve for a HJvsBB spot on a T87 board:

These were the ranges I gave:

Range 1 OOP



Range 2 IP



I ran sizings up to 10x pot (1000%) and these were the results

33% 2.98EV 3.30EV for Betting/2.52EV for Checking: 59.8% Frequency
50% 3.00EV 3.42EV for Betting/2.57EV for Checking: 50.6% Frequency
75% 3.00EV 3.52EV for Betting/2.63EV for Checking: 42.0% Frequency
100% 3.02EV 3.63EV for Betting/2.68EV for Checking: 35.5% Frequency
125% 3.02EV 3.68EV for Betting/2.73EV for Checking: 30.9% Frequency
150% 3.03EV 3.73EV for Betting/2.78EV for Checking: 26.7% Frequency
175% 3.05EV 3.80EV for Betting/2.82EV for Checking: 23.4% Frequency
200% 3.06EV 3.87EV for Betting/2.83EV for Checking: 22.2% Frequency
225% 3.08EV 3.93EV for Betting/2.87EV for Checking: 19.9% Frequency
250% 3.09EV 3.98EV for Betting/2.90EV for Checking: 17.6% Frequency
300% 3.12EV 4.06EV for Betting/2.97EV for Checking: 13.9% Frequency
350% 3.14EV 4.13EV for Betting/3.02EV for Checking: 11.4% Frequency
400% 3.17EV 4.20EV for Betting/3.06EV for Checking: 9.12% Frequency
500% 3.20EV 4.31EV for Betting/3.13EV for Checking: 5.80% Frequency
600% 3.23EV 4.48EV for Betting/3.18EV for Checking: 3.75% Frequency
700% 3.24EV 4.49EV for Betting/3.21EV for Checking: 2.46% Frequency
800% 3.25EV 4.54EV for Betting/3.23EV for Checking: 1.62% Frequency
900% 3.26EV 4.77EV for Betting/3.24EV for Checking: 0.99% Frequency
1000% 3.26EV 4.86EV for Betting/3.26EV for Checking: 0.48% Frequency


Conclusion: 9x pot and 10x pot are the highest EV sizings - can you confirm this?

We currently do not display aggregate frequencies beyond the flop. Aggregate data is mostly useful for the very first decision of a tree. In this spot, each tree in the database will have the same range, and the data can therefore be added. As you go deeper down a tree, the ranges for a certain spot in each tree will be different, and can’t really be added anymore. You’re basically adding apples and pears.

In particular, when you go as far as the turn, or even river, then this issue becomes even more serious.
Here, even within a single tree, it’s not really possible to add data for the different turn lines anymore.
For example, if you bet 0% of a 2h, but 100% on a 2c, then the aggregate data for these two lines will be to bet 50%.
Although this addition is technically possible, it is not particularly useful, and can even steer our users towards the wrong conclusions.

It’s for this reason that, at the very least for the moment, we are restricting aggregate data to the flop.
We can consider expanding it anyhow for later releases, but right at this moment, it has been left out for this reason.

Do you mean that you have run 19 different trees, where the only difference between them is the first bet size for OOP? It's not entirely unthinkable that really large bet sizes perform really well. I've personally run into spots where this was also the case. It's rather against conventional wisdom, but there are indeed spots where large bet sizes seem to lead to a surprisingly good EV performance. One possible thing to look out for is if perhaps you've set up the tree builder in such a way that villain's bet sizes become similarly large, thus negating the possible negative effect of hero picking a bet size that does not fit the situation.

To build a database by using a subset:
1) Build a tree with the tree builder
2) Go to "Run solver->Advanced->Activate database mode"
3) Under "Import flops from file" enter the desired size of subset (available subsets are 15,19,23,33,37,44,55,66,80,89,111,129,141,163)
4) Click "Import flops from file"

By doing this you can, for example, quickly create a database of the same tree for 44 different flops.
These flops will be a close representation of all 1755 possible flops.




If you would like a video demonstration, then please watch the second video here: www.gtoplus.com/videos.
This video does not yet describe how to use subsets though, given that it was created before this feature was added.
However, the principle is the same.



It's only possible to add the data for trees if they are actually the same. So they need the same ranges, pot, stack sizes, bets, etc. The only difference between them should be which flop is used. If you use the subset feature then all of this will be taken care of for you automatically though.

Yes I ran 19 different trees. I didn't touch Villain's bet sizings I left those blank. So is this accurate? It is pretty much a range check OTF?

I would imagine so.
I'd need to see the tree builder to check for myself, but it does indeed happen that very large bets outperform regular ones.
10x being best would be a new record for me though.
In most cases the best performance will be more towards regular sizes, with there being little benefit to any particular size.
In this case what does occur is that the board is very draw-sensitive, with OOP having a more draw-dependent range, and IP having a range with more made hands.
So there's probably a reason why this spot stands out.

Thanks Scylla, could you please check for yourself .

Very cool solution! GTO is absolutely insane.

If you want me to check, then please mail a savefile to support.

okay will do. Thanks!

What's the email address and do I just save file?

Ok, from what I can tell we have received your mail and replied.

Im a bit confused over the effective stacks setting. If we both have 100bb in a 3bet pot of 20bb what number do I use? I thought I had to use 100bb but it doesn't seem to be taking into account the money in the pot already and removing it from our stacks so I need to be using the post flop effective stacks? cheers