You can’t listen to or eavesdrop on a poker conversation these days without the mention of “GTO.”
That doesn’t mean GTO is something new. It’s not.
When it comes to poker, though, it’s often misused and misunderstood.
That’s why a tweet that included a poll about GTO from poker pro Oliver Busquet caught our attention.
If a perfect GTO bot played only live tournaments 25k entry or higher it would be
— Olivier Busquet (@olivierbusquet) February 21, 2018
What followed was a fascinating conversation.
But before we get to that …
What is GTO?
Let’s start by defining what GTO is and what it is not.
Game theory optimal solutions are specific mathematical definitions to a strategy. In other words, it’s the best, unbeatable solution to a game.
What you need to know is GTO doesn’t really exist in poker. Poker isn’t solved, and because there is no solution, GTO by its strictest definition can’t be applied to the game.
Defining GTO as it applies to poker just means taking the most unexploitable line. While simply described, there isn’t anything simple about it.
Dispelling common myths about GTO
There are some common myths about GTO and how it applies to poker that could wind up getting a player in trouble.
Two of the most common myths are:
- GTO strategy is a profitable strategy
- All the high-rollers use GTO
Let’s talk about those a little further.
GTO play does not guarantee a profit. GTO means you are using a strategy that is unbeatable. The problem is that a GTO approach doesn’t ensure you’ll win against certain opponents.
Technically there is no GTO in poker. When players reference GTO, they are most likely speaking of the best possible strategy.
GTO, as it applies to poker, is a strategy that is influenced by GTO.
GTO discussion
Busquet followed up with a request for opinions in addition to voting in the poll. He specifically asked for responses from players that are “very good.”
And the very good (and some not so good) players obliged.
Dylan Linde
Probably would have a higher win rate relative to the field in 25ks than say 3500s since the 25k fields are attempting to approximate GTO more and are less open to exploits. Whereas the weaker the field the more there is to make from proper deviation of GTO play to exploitative
— Dylan Linde (@DylanLinde) February 21, 2018
When Busquet asked what he voted for, Linde responded
“Among the elite”. As I think to be the best even in the more GTO minded fields you still play with fun players and exploitable weaker regs that deviation will be more profitable vs.
— Dylan Linde (@DylanLinde) February 21, 2018
Busquet was “quite shocked” by the answer and followed up with a response.
Far and away the best. I think the current “best” players make enormous mistakes and people give them way too much credit for trying to “exploit” or their exploits making up for their mistakes
— Olivier Busquet (@olivierbusquet) February 21, 2018
Which got Linde thinking.
Hmm with some further thought I am now also thinking about the limitations of solvers now. The machine could play GTO in multi way pots whereas the capability to solve those isn’t yet available to us. The machine would’ve answered this question properly on the 1st attempt
— Dylan Linde (@DylanLinde) February 21, 2018
Also the bot never loses confidence, gets tired, emotional or distracted.
— Olivier Busquet (@olivierbusquet) February 21, 2018
Dominik Nitsche
Obviously the best. Whether marginal or far and away is tough to say. I’d think some humans are quite good at poker at this point so the bot shouldn’t be winning too much
— Dominik Nitsche (@DominikNitsche) February 21, 2018
I find it particularly interesting that someone like you (obv elite) takes a nuanced view whereas many of the poll responses are polarized
— Olivier Busquet (@olivierbusquet) February 21, 2018
This is really common amongst the smartest/most successful people IMO. Inverse correlation between certainty and intelligence, esp on difficult/unknowable Qs like this
— Kintell Williamson (@JMC_PGH) February 22, 2018
Allen Cunningham
When players refer to GTO, I take it as using tactics influenced by game theory and NE strategies, such as (attempted) balancing. I haven’t seen any paper or study demonstrating there is a “perfect” ring poker strategy. I’d welcome a link if there is one.
— Allen Cunningham (@AllenCunningham) February 22, 2018
Ryan LaPlante
Among the elite, but not the biggest winner.
To be biggest winner youd have to properly adjust for explootative tendencies the recreational players have and how everyone adjusts to play vs them.
Could be instead marginally the best.
— Ryan Laplante
(@Protentialmn) February 21, 2018
Ryan Yu
To maximize theoretical win rate you’d be right. But no human can achieve maximum exploitive value either, and that just playing perfectly GTO would still be a greater overall net value than a human who might do certain things better than the bot.
— Ryan Yu (@ineedsheet) February 22, 2018
Olivier Busquet
A. I think players overestimate the value of their adjustments (credit) and underestimate the size of their mistakes (blame)
— Olivier Busquet (@olivierbusquet) February 21, 2018
Daniel Negreanu weighs in.
Both true. Only factor not being discussed thus far that I’ve seen is the value of exploiting via live tells.
— Daniel Negreanu (@RealKidPoker) February 21, 2018
I’m including live tells under the umbrella of exploits. I made my example live on purpose to include them.
— Olivier Busquet (@olivierbusquet) February 21, 2018
Are you struggling to understand GTO?
Does this clear things up? If not, Fedor Holz recently tweeted out a link to The Evolution of Trust.
The Evolution of Trust is an interactive guide to game theory and why and how we trust each other. It’s worth a look.
Trust us.
The post Discussion: How Would A Perfect GTO Poker Bot Perform At High Stakes? appeared first on .