grand_ua's avatar

grand_ua

1 points

I feel that I am starting to figure out how to approach learning and memorizing GTO, and the right kind of questions for this process. These are the questions that I came up with for the start:
https://www.runitonce.com/nlhe/balanced-turn-bluff-barreling-quiz-total-bluffing-frequency-with-all-no-sd-hands/

And I will follow with answers and more questions later on.

Nov. 24, 2022 | 10:57 p.m.

There are very many GTO experts and tools out there. But do they know how not to over-bluff or under-bluff?
I suggest you to test your GTO knowledge, GTO understanding and your choice of GTO learning and training tools, or even advertise your GTO coaching abilities by answering the next questions:

1 What are the average, minimum and maximum bluffing frequencies for turn barrel across all flop textures and turn runouts?
2 How can you explain deviations from average turn bluffing frequency, and from bluff to value ratio formula?
3 On what flop texture + runout combination GTO bluff barrels turn the most?
4 The least?

5 On what flop texture(s) GTO bluff barrels turn the most?
6 Why?
7 What is the bluffing frequency here?

8 On what flop texture(s) GTO bluff barrels turn the least?
9 Why?
10 What is the bluffing frequency here?

11 On what turn runout category GTO bluff barrels turn the most?
12 Why?
13 What is the bluffing frequency here?

14 On what turn runout category GTO bluff barrels turn the least?
15 Why?
16 What is the bluffing frequency here?

I plan to answer these questions for HU scenario node locked for 100% flop cbet with 1/3 pot in about a month, if other people do not answer by that time.

Nov. 24, 2022 | 10:51 p.m.

I have just realized that GTO Wizard is a name of online software. I have never heard about it before, but it looks interesting - checking it out, thank you.

Nov. 23, 2022 | 9:11 p.m.

"To learn balanced ranges. I recomend using somthing like Gto wizzard"

  • Do you mean something like GTO solvers?
    GTO solvers will show you that their strategy to bluff something like 23% of air + 76% of flush draws + 80% of OESD + 51% of gut shots on XYZ flop + x blank turn, or 51% of air + 79% of flush draws + 93% of OESD + 76% of gut shots on Q turn. This is a fake example with fake numbers but I hope that you get the point. All numbers will be not round and hard to remember, but you could use rounding feature in solvers to simplify that.
    The bigger problem is that those numbers will change a lot depending on a specific combination of a flop + turn runout. Also on flops and turns GTO will not bluff the same bluff-to-value ratio for a chosen bet size as per a simple formula, because there will be a different percentage of good river runouts for bluffing or value betting depending on flop + turn interaction with GTO ranges.

So how are you actually going to use that kind of information from a GTO solver as in above fake example in a real game?
You say: "you will not be able to replactate these ranges perfectly. Your not looking to copy them. You are trying to understand the why behind solver or AIs actions and then add it to your game." Yes, we cannot replicate. OK, we understand intuitively that on very wet boards and on boards where we have disadvantage we need to bluff less somehow. But what does that mean practically, how much less should we bluff?
Will you actually count your bluff and value combos when playing? Do you round GTO frequencies to be able to remember something? Will you bluff turn with some percentage of pure air and blockers like GTO does? Are you going to change percentage of draws and air that you bluff with, depending on how wet is the board and how good is board and turn runout for your range vs villain range? Do you use some sort of simplified heuristics for all that balance stuff, or do you do actual bluff to value formula calculation and combo counting while playing?

As for myself, I can tell that I do not do any of balance math at all. Specifically for the turn barrel line I mostly use standard lines that tell me that I am supposed to bluff barrel turn with all my gut shots+. But because it is too exploitable not to have any draws in a checking back range at all, I do mix draws between betting a checking depending on range interaction a little bit, though far from being close to GTO.
I mostly just end up over-bluffing when I know that population overfolds, or I bluff only very good draws when I know that population under-folds. That works pretty well actually in 6max up to mid stakes that I play.

But I play not only 6max, I play HU as well, and I do not feel that you can get away always bluff barreling all your gut shot+ draws on any board vs any villain, as per standard lines from 6max, or always over-bluffing in spots when population over-folds. Quite many times I run into very sticky players HU (some sort of aggressive fish usually) that did not fold much when people usually fold enough, and I did not have any good plan against them. If I did not bluff, they would bluff me out from most pots later on. If I bluffed, I would feel like spewing, because those villains did not fold much. I play those villains HU, because I know that they are fish, not good players, and that they are way off balance. But in fact I am off balance myself, because I do not have a good way to calculate or know balance, and I can be a fish myself if villain happens to play a style that is different from population, or does not do what I expect him to do by default.

Also it is not very comforting to know that if someone buys my hand history and finds all the many spots where I am off balance, he can own me HU. Hopefully it will not happen in mid stakes, but it can definitely happen if I start to play high stakes HU, because why not? In fact people do not even need to study my game - they just need to know population hot spots and have a prepared tactic how to exploit someone who tries to exploit those population hot spots.
And what and how can I counter that? -If I knew a balanced way, I could counter by being balanced when I feel that my exploitative line does not work, and I do not know how to exploit. But I do not know a balanced way, I only know how to over-bluff or under-bluff.

So my point is that both GTO and exploitative suck in their own way, GTO is hard to figure out and implement when playing, and pure exploitative makes us being owned in some situations, so it is better to know major frequencies from both, and switch them depending on what we learn about villains during play.

Maybe most people have an idea what to do to figure out exploitative lines and frequencies.
But how many people can figure out a balanced strategy, that is feasible for humans in real time, especially in high frequency spots? So far no one seems to be willing to advertise their services or tutorials about that.

Nov. 23, 2022 | 7:58 a.m.

Thank you for ideas how to approach it.
- I already know basic GTO fundamentals like bluff to value formulas, and I had a look into different solved solutions. But I fail to see how this sort of knowledge can be applied practically during play.

I have an access to solvers data that is combined by flop textures and runout types, and I can tell that GTO does not bluff simply using something like bluff-to-value formula on flops and turns. Bluff percentage before river changes a lot depending on how advantages turns and rivers will be for our range, and on our range and nut advantage.

Also to my knowledge only very few people even try to be balanced. Even most coaches do not really try, or they are not balanced by fact anyway.
I myself know only one player who did a focused effort to learn how to be balanced in frequent lines - his name is Doug Polk, so you most likely know that name too. Balance is such a hard problem for humans, that I would guess that Doug Polk 6 max play would be significantly less balanced than his HU anyway, and if we study his bluff frequencies by board textures and runout types, I would expect that they would not look so much balanced.
Do you know other names who play mostly balanced by chance?

If someone somehow obtained all your hand history and calculated stats from it, and studied your bluffing frequencies for hours, do you expect they would see balanced numbers there or not so much balanced? How about if they analyzed your bluffing by board textures and runout types?

Nov. 20, 2022 | 5:55 a.m.

I do not know about here, but Poker Detox specializes on mass database analysis (MDA) and has good video causes about that, but they are not cheap. Bestpokercoaching.com also has some information - look for Population Hot Spot and Standard Lines keywords. PokerStrategy.com also has some advice derived from MDA, but there is no way to see it all together, as far as I know.

Nov. 20, 2022 | 5:21 a.m.

Can you come up with practical and interesting questions about GTO that would be hard to answer using solvers by chance?
-I do not know how to market this thing yet, so I am looking for good ideas.

I myself know a lot of questions that would take days to answer using solvers, but I do not know which of those questions are practical and interesting to good players.

For me it was revealing to learn that on turn GTO does not bluff some constant percentage as the bet size and the bluff to value formula would dictate. Turn bluff percentage changes wildly based on factors such as a number of river cards that are good for us and for villain, our nut and range advantage or disadvantage. This fact alone makes it impossible for humans to replicate GTO bluffing frequencies.

Nov. 20, 2022 | 5:08 a.m.

I have created the first tool that allows to simplify, compare side by side, and as a result really understand GTO (or node locked solver strategies) by starting analysis from simplified top levels like categories of hands, flops, runouts and bet sizes and drilling down to more specific levels and individual cards.

It is the only tool that allows to compare side by side GTO frequencies and bet sizing across multiple flop textures and any other combination of factors of your choice.
Any kind of analysis should be possible, and any kind of solver data can be loaded potentially.

My next goals are to gage interest, find partners and understand how to make money, as well as guide adding new features by chatting with smart people and figuring out what kind of things and tricks are the most important to pick up from GTO, and what kind of GTO analysis is too difficult or impossible currently (balancing bluffs on all 3 streets comes to my mind here as one example).

Below you can see some screenshots with annotations. The screenshot data is from a modified HU GTO scenario with flop donk bet line removed and 100% flop cbet for 1/3 pot for in position player. Unique features are circled in yellow:
Screenshots

July 21, 2022 | 5:34 a.m.

Does anyone know any software (or reports) that allows to compare GTO bluffing frequencies for different hand categories on multiple flop textures, turn and river runout types?

Or can anyone recommend any people who can teach how to construct balanced ranges, how not to over- or under-bluff in frequent lines, especially in HU sessions, especially on turn and river?
And if yes, what is the process that those people use to achieve that?

July 8, 2022 | 9:48 p.m.

Does anyone know any software (or reports) that allows to compare GTO bluffing frequencies for different hand categories on multiple flop textures, turn and river runout types?

Or can anyone recommend any people who can teach how to construct balanced ranges, how not to over- or under-bluff in frequent lines, especially in HU sessions, especially on turn and river?
And if yes, what is the process that those people use to achieve that?

July 8, 2022 | 6:35 a.m.

MDA is use case # 3 in my question, but I will cover all 3 as related.

I am not affiliated with Poker Detox, and I do not have a full insight, but I had some communication.

Also I had a look into the software that Poker Detox uses, and I tried to adjust it for either of the use cases 1-3 above. And I discussed my findings with the developer of that software. Poker Detox did not develop that software, but I will call it PD Training Software for references below.

So these are my conclusions so far:

Use case # 1:
You cannot do # 1 use case in PD Training Software, without paying to Poker Detox or some other human coach, as they have to analyze each of your hands manually (Well, they have someone to analyze for free for now actually, but I do not know the level of those people. And I am pretty sure that free is not forever, and that you cannot ask them to use PokerSnowie, PIO or other tools of your choice). You cannot work on your leaks in a poker re-player using your own and free hand history independently, basically.
More importantly, even if you pay to Poker Detox (e.g. via CFP program), or to any other coach, PD Training Software was not designed for leak busting, as their re-player always assumes that every uploaded hand has been played correctly.
All in all, that does not fit #1.

Use case # 2:
You can use PD Training Software to copy a game of the best regs (sort of), but you cannot use the most of your own hand history for that purpose either, as observed hand history is not supported. So you are forced to work with their affiliated pros only, even though you may have a ton of great hands to learn from in your own hand history. And those pros are not necessary the best experts in the player pool where you play.
All in all, that means that #2 is not supported either.

Use case # 3:
Now about converting stats into hands for an interactive re-player.
PD Training Software only input is via hand history text format.
It is not exactly easy or practical to convert population stats into hand history format.
One way to do that, is to play correctly and upload a real hand history.
Second way would be to find a similar hand and make changes to it.
To play correctly you have to convert population stats into a strategy and memorize that strategy. But if you do all of that, I am pretty sure it is not that fast to do, and you do not need training software for that line anymore. So you have to pay to Poker Detox again, provided they have enough content.
Finding example hands and making correct edits to them, and testing are not that practical or simple either in my opinion. I do not know about you, but I would not do it that way.
No wonder that no one except Poker Detox seems to have MDA based hands on PD Training Software platform, and even Poker Detox does not have that many.
PD may have a couple of hundreds of MDA based hands, while I have dozens of thousands in my personal training database already for comparison.

Things may change in future, but I would not expect changes for any of the 3 use cases above any time soon, as the developer of PD Training Software did not want to do either of them per my request

I may not know something, but that is what I know so far.

Oct. 15, 2020 | 10:56 p.m.

Do you mind to elaborate which ones are wrong and why please?

Oct. 15, 2020 | 8:41 p.m.

Thank you for your ideas BigFiszh. My response is below:
"against an optimal defending / playing opponent" - no human can ever qualify to play completely optimal, and this concern is pretty much completely irrelevant anywhere below very high stakes. Humans do tons of mistakes, so such an assumption makes this part incorrect also: "There's "one" solution that is "always correct", regardless of circumstances, that's the GTO solution. A GTO solution IS per definition the maximum exploitative strategy".
If you have taken all the information and ranges that you know about a specific player or player type and used it to node lock in PIO Solver, than yes, you would get the maximum exploitative strategy from a GTO solver, but it would result in many more solutions than one.

"very pool-specific strategies" - this is also a quite wild assumption. Humans are actually quite consistent in how they like to make mistakes across multiple pools and poker rooms (even though different player types prefer different types of mistakes). I can refer you to PokerDetox posts regarding MDA, and my own stats confirm their findings. You may find articles about population Hot Spots quite beneficial also.

GTO is good to know, but it is not the most efficient approach to real life poker.

Oct. 13, 2020 | 8:41 p.m.

Is anyone interested in utilizing their own free content like hand history and stats for exploitative strategy and leak busting training software?

Idea in details:

Personally I am interested in all of the next 3 use cases:
1. Leak busting: Interactive re-player software where I import my own hands that I played incorrectly (using PokerSnowie blunders for instance). I should have to press buttons just like in a real game, and be able to provide a correct decision for a hand that I played incorrectly.
2. Learn by copying strategy from the best regs based on my observed hand history. This is similar as in # 1, but it should be based off my observed hand history.
3. Learn the maximum exploitative strategy based strictly on a mass database analysis of a combination of line, board and player type. We will need an interactive re-player here also, but it should be driven by stats collected for board texture, player type and line combination.

My impression so far is that training software for any these use cases is not available on the market yet.
I would like to know if my impression is correct, or there is anything available already.
If nothing is available, I would like to understand why that is the case, and find partners to build it (I am a software developer).
I also would like to understand if anyone else is interested in training software for any of those 3 use cases, or it is just me.

Oct. 12, 2020 | 6:14 p.m.

Load more
Runitonce.com uses cookies to give you the best experience. Learn more about our Cookie Policy