MPT vs "traditional poker (2 Viewers)

[djbird]

Hello - new here. Played a lot of poker in the early 2000's when traditional poker was more reading people and "feel" with some math (pot odds, rough odds to make hands etc) mixed in.

Now I see a whole bunch about Modern Poker Theory and "solvers" that maximize play/return over time.

Curious, has anyone studied MPT, do you use it, how does it work for you?

 

[guyfleegman]

Welcome to the forum J.

Lifetime friend finally on PCF.

 

[Jimulacrum]

I studied this a bit when it was first coming up—Bill Chen's Mathematics of Poker primarily—before anyone had developed "solvers." That is to say, I understand roughly how it works and why it's a powerful strategy. It was interesting to read and learn about.

Since then, my feeling is that it absolutely ruins the game of NLHE.

In theory, it ruins the game because it equips players with an effectively unbeatable, computer-generated strategy. Obviously this is far worse for online poker than live poker; it effectively nukes online poker forever by enabling the game to be beaten by players directly using solver output or coding it into bots. Even if they're forbidden, it's at best a game of cat and mouse, and any player not doing it would be well-advised to avoid public online poker. Even private online games, you'd be extending a lot of trust.

In practice, it ruins the game in ways like the Tamayo scandal. It was already advisable to ban devices at the table prior to solvers. Now it's practically mandatory, and any cardroom that doesn't ban devices will have to worry about people using solvers to cheat live. (Yes, cheat. Gaining an advantage by using a device to do math for you at the table is cheating, outright, no question.)

Looking at the long run, if enough players use a solver strategy, it turns NLHE into a chance game. Everyone's paying rake to basically gamble on the outcomes, not on skill, because there is no longer any skill differential—except for people who aren't using solvers, who will get fleeced with no (theoretical) chance to win.

TLDR: NLHE is a dying game. Play Double Board Omaha.

 

[djbird]

I studied this a bit when it was first coming up—Bill Chen's Mathematics of Poker primarily—before anyone had developed "solvers." That is to say, I understand roughly how it works and why it's a powerful strategy. It was interesting to read and learn about.

Since then, my feeling is that it absolutely ruins the game of NLHE.

In theory, it ruins the game because it equips players with an effectively unbeatable, computer-generated strategy. Obviously this is far worse for online poker than live poker; it effectively nukes online poker forever by enabling the game to be beaten by players directly using solver output or coding it into bots. Even if they're forbidden, it's at best a game of cat and mouse, and any player not doing it would be well-advised to avoid public online poker. Even private online games, you'd be extending a lot of trust.

In practice, it ruins the game in ways like the Tamayo scandal. It was already advisable to ban devices at the table prior to solvers. Now it's practically mandatory, and any cardroom that doesn't ban devices will have to worry about people using solvers to cheat live. (Yes, cheat. Gaining an advantage by using a device to do math for you at the table is cheating, outright, no question.)

Looking at the long run, if enough players use a solver strategy, it turns NLHE into a chance game. Everyone's paying rake to basically gamble on the outcomes, not on skill, because there is no longer any skill differential—except for people who aren't using solvers, who will get fleeced with no (theoretical) chance to win.

TLDR: NLHE is a dying game. Play Double Board Omaha.

I kind of look at it like blackjack. If everyone plays the way your supposed to it works, but if one person doesn't the theory doesn't work.

It's also just not the style I'd like to play, so I might study it some to learn what others might be doing, but hoping to see if anyone uses it and their thoughts if it's working for them.

 

[Jimulacrum]

I kind of look at it like blackjack. If everyone plays the way your supposed to it works, but if one person doesn't the theory doesn't work.

How does one person at a blackjack table not following basic strategy prevent it from working for anyone else?

 

[djbird]

How does one person at a blackjack table not following basic strategy prevent it from working for anyone else?

It's actually a bit of a myth that someone who doesn't follow optimal play strategy hurts the table. In the short term it can, but over the long run, it really doesn;t.

However... the point of saying that was to illustrate how I view MPT. If everyone everywhere played MPT, in theory you would have a high likelihood of knowing what everyone had (within range) but if one player doesn't play that way, it would throw off the game.

At least that's my initial thought on it.

 

[Jimulacrum]

It's actually a bit of a myth that someone who doesn't follow optimal play strategy hurts the table. In the short term it can, but over the long run, it really doesn;t.

Okay, good, we agree on this. I've gotten into many an argument about the "You made everyone else lose" mentality you see in blackjack sometimes. Easily disproven with simple examples.*

Spoiler: *

Sidetrack: Dealer has a 5 showing. The whole rest of the table stands with counts of 12+. Action gets around to you. You are last to act with A7. You choose to double down (regardless of whether this is right or wrong). There are only two cards left in the shoe; one will make the dealer win, and the other will make him bust. Has your action helped or hurt the rest of the table?

However... the point of saying that was to illustrate how I view MPT. If everyone everywhere played MPT, in theory you would have a high likelihood of knowing what everyone had (within range) but if one player doesn't play that way, it would throw off the game.

At least that's my initial thought on it.

Actually, this is not the case.

The way game-theory optimization (GTO, the basis of modern poker theory) works is that it recommends a mix of actions that if carried out in a properly randomized way, it will make your strategy effectively impenetrable. That is, even if someone knows for sure what your entire strategy is, the ratio of call/fold/raise, bet sizing, and pot odds balances out just right, so that there is no counter-strategy.

GTO doesn't make you play perfectly or maximize your earnings; it just makes it so your opponents cannot exploit you. It's more of a defensive strategy than anything, but an extremely effective one. Exploitative play—identifying and beating on your opponent's weak spots—is how you maximize profits, but it opens you up to counter-exploitative strategies ("leveling"), and from there you may need an additional layer of strategy, and so on. With GTO, you don't think about your opponents. The strategy is standalone.

 

[ekricket]

Has your action helped or hurt the rest of the table?

In reality I don’t care. I’m not playing for the table to win. I make the best decision for me at the moment at the blackjack table. I would expect everyone else at the table to be doing the same.

 

[ekricket]

You are last to act with A7. You choose to double down (regardless of whether this is right or wrong).

Sorry for the derail, but I’d double down here myself. I want to get as much money on the table every time I feel I have an advantage over the dealer.

 

[djbird]

With GTO, you don't think about your opponents. The strategy is standalone.

Gotcha, I did not know that. I assumed there was always some level of traditional poker involved.

Having read a bit more about it, I don't think it's all it's cracked up to be. At least not for live in-person play.

 

[Jimulacrum]

In reality I don’t care. I’m not playing for the table to win. I make the best decision for me at the moment at the blackjack table. I would expect everyone else at the table to be doing the same.

Sorry for the derail, but I’d double down here myself. I want to get as much money on the table every time I feel I have an advantage over the dealer.

I agree 100%. My point with this example is to demonstrate that by making the correct play for yourself, you are not materially affecting the rest of the table in a probabilistic sense. Each card is still a random draw. They happen to be somewhat dependent draws in the case of blackjack, but even so, no one can say whether it would help or hurt the rest of the table because the cards can be in any order. Even if we extend this to a full shoe, the logic still holds.

But a lot of blackjack players get big mad if you draw a card here, IF it results in the dealer getting a favorable card. (If it results in you taking that card and the dealer busting, they don't recognize the contradiction compared to the times they get mad.) Mainly it's just people getting mad that they lost and trying to blame someone, but they try to peddle some really questionable logic about it, like that all the players should be "playing together to beat the dealer."

 

[Jimulacrum]

Gotcha, I did not know that. I assumed there was always some level of traditional poker involved.

Having read a bit more about it, I don't think it's all it's cracked up to be. At least not for live in-person play.

Yeah, for live play, it's more like something you'd study to have a strong default/defensive strategy to employ as best your primate brain can manage (which is extremely difficult because it's quite complicated). It's great as a thought exercise to help you understand how to play against ranges and how to perceive yourself as a range being played against, but it's not terribly practical to use as a whole unless you're breaking out the software right at the table.

 

[Jers28]

GTO isn’t the end all be all for NLHE imo. Or most of the top pros, for that matter. It’s a tool. Like any other tool, it absolutely has its value. But, if it’s the only tool you own, you will definitely be leaving profit on the table. There is no GTO that is ever going to tell me to play 2/5o, for example. But, mixing GTO with exploitive play, traditional reads, etc can be extremely valuable.

It also depends on stakes. If you try playing GTO in a small stakes game, best of luck. You’re on the button with A5s and six players limp in to the pot. You 3 bet to 10x and all six players call. When the dust settles, some of those limpers had trash that should have been folded, per usual like 10/4s. One guy limped with pocket kings and another with QKs. I don’t see how GTO is radically changing this situation. And I’m yet to see a small stakes game that doesn’t have its fair share of players who will call/limp with any two cards on nearly any size raise. Firing multiple bullets in position into 6 players, 4 of which should have folded preflop, only to find out that flop that was 10, 4, 4 had you crushed by the guy who limped with 10, 4. No solver is putting the middle position player on that hand. And he’s not betting. Just calling as he has been all night long with everything.

I’m in no way suggesting his play can’t be beat. Please, by all means play that trash all night long. I’m merely pointing out the hole in GTO is that for it to work, the player in the low jack has to be at least close to fundamentally sound. Just one example. Please don’t get concerned about the semantics of my example. We all have players in our games that won’t fold cards they should and won’t 3 or 4 bet hands they should, either. I don’t need GTO to beat those players and my argument is that GTO will cost me profits in those games.

I still find it to be an extremely valuable tool, provided you use it as just one of the tools in your bag.

 

[Taghkanic]

A GTO solver doesn't care if its opponent is playing a tight or loose or even a random strategy.

If two GTO bots play a million hands against each other, it is going to be a draw.

If a GTO bot plays a million hands against a typical 1/2 fish, or someone choosing bizarre hands and bet sizes, it is never going to lose.

It may win less than a very sharp pro who is trying to exploit the fish or random player to the max. But it will never lose even with extreme variance, because it is playing unexploitably.

Some big caveats are:

1) No human can play perfectly balanced GTO live, even an idiot savant who can memorize hundreds of tables, because there are just too many permutations. At best a human can try to approximate pure GTO by lumping similar but not identical situations together.

2) Things get complex even for computers in multiway pots with an almost infinite number of bet sizes. Current GTO models tend to limit the number of bet sizes available and focus on heads-up play.

3) Most pros studying this topic are not trying to play "perfect" GTO. They are trying to understand it better so as to see how others at the table are deviating from GTO, and use that info to better exploit their flaws. They are likely to try to play pure GTO against another top pro, or against total unknowns, until they see enough hands to understand what is sub-optimal in their play. Then adjust from GTO to profit more while protecting against being exploited themselves.

 

[Jimulacrum]

A GTO solver doesn't care if its opponent is playing a tight or loose or even a random strategy.

If two GTO bots play a million hands against each other, it is going to be a draw.

If a GTO bot plays a million hands against a typical 1/2 fish, or someone choosing bizarre hands and bet sizes, it is never going to lose.

It may win less than a very sharp pro who is trying to exploit the fish or random player to the max. But it will never lose even with extreme variance, because it is playing unexploitably.

I was going to say something like this in response to @Jers28's reply above yours. The example about people calling down with making GTO not work reads a bit like the classic, "These people don't respect my raises!"

Obviously GTO is an inferior strategy EV-wise against lolfish because it leaves a ton of money on the table, but it doesn't fail to work as a defensive strategy because of them. Long-term, it still protects you from being exploitable; it's just a waste against people who can't spell "exploit" and have massive weak spots you could attack instead.

 

[Jers28]

Did you read the whole post? I ramble, so I don’t blame you if you didn’t. Factoring in what level of game you’re playing was kind of my whole point. The majority of people don’t play high stakes. Even if they want to, it’s hard to find games beyond $2/$5 at the majority of casinos, never mind home games. Discussing hands after a game with poker buddies, I constantly hear explanations like “I just play based on feel” when I ask why they did xyz. Assuming that player is player proper GTO is pointless. I went out of my way to say that GTO is an extremely valuable tool. So is a hammer. Unless the job requires a wrench. FWIW, I sincerely hope people don’t respect my raises.

 

[Jimulacrum]

Assuming that player is player proper GTO is pointless.

I did read your whole post, and this is the part where we're diverging. GTO strategy does not rely on your opponents also playing GTO (or close to it). It's intentionally agnostic to your opponents' strategies. The whole point is that it neutralizes any strategy anyone might use, including "I just play based on feel."

If you find your opponents are playing GTO, the solution isn't to play GTO to counter them. It's to find another game.

 

[Jers28]

I did read your whole post, and this is the part where we're diverging. GTO strategy does not rely on your opponents also playing GTO (or close to it). It's intentionally agnostic to your opponents' strategies. The whole point is that it neutralizes any strategy anyone might use, including "I just play based on feel."

If you find your opponents are playing GTO, the solution isn't to play GTO to counter them. It's to find another game.

Fortunately, Mike Mizrachi, Daniel Negreanu, Phil Ivey and a ton of other pros would disagree.

Maybe I’ll try Star Wars…

 

[ekricket]

Solvers and theory didn’t come into popular use or mainstream poker awareness until the 2007-2010 timeframe. Pros were winning before that too. They may have evolved to use the tools, but they were playing pro poker long before these things became commonplace. It’s not what made them.

 

[Jimulacrum]

Fortunately, Mike Mizrachi, Daniel Negreanu, Phil Ivey and a ton of other pros would disagree.

Maybe I’ll try Star Wars…

View attachment 1553597

Unfortunately for you, I've got the pope and the king of England on my side, since we're just dropping names with no context.

 

[Jers28]

That’s enough. I no longer care. You do you brother.

 

[ekricket]

sorry, got op mixed up with an responder.

 

[Jimulacrum]

That’s enough. I no longer care. You do you brother.

In these clips, Negreanu says essentially the same things that @Taghkanic and I have said.

 

[Jers28]

I don’t think we are far off. Daniel says essentially it’s a tool to be used. I never said it didn’t have merit. Just that bs real life humans, you can’t play optimal and max profits, which is exactly what Daniel says. This is pointless. My bad for chiming in. You win. I’m a moron.

 

[Jimulacrum]

Just that bs real life humans, you can’t play optimal and max profits, which is exactly what Daniel says.

I feel like we more or less agree in spirit but not in semantics.

Pardon the friction. Maybe I'm a moron too. It's Friday, man. Truce?

 

[Tommy]

What about CTO

 

[Jimulacrum]

What about CTO

 

[ekricket]

What about CTO

That’s in this thread

https://www.pokerchipforum.com/threads/why-does-this-keep-happening-to-me.135554/

 

[Taghkanic]

I was going to say something like this in response to @Jers28's reply above yours. The example about people calling down with making GTO not work reads a bit like the classic, "These people don't respect my raises!"

Yes, saying that is a fundamental misunderstanding of what game theory is.

 

[Eriks]

I don’t think we are far off. Daniel says essentially it’s a tool to be used. I never said it didn’t have merit. Just that bs real life humans, you can’t play optimal and max profits, which is exactly what Daniel says. This is pointless. My bad for chiming in. You win. I’m a moron.

I’ll give it a go, here is my understanding of GTO:

GTO doesn’t mean optimal in terms of always making max profit. Playing perfect GTO would simply make you unexploitable and nothing more - no matter if you’re playing the T4s guy, Daniel Negreanu, Phil Ivey or Kevin playing randomly, you won’t lose (theoretically). But you won’t make the max either.

Anyone who does not play perfect GTO (which is every human) is exploitable in some way though. So in order to make max profit one would deviate from GTO to exploit any given player’s in-game tendencies.

Naturally, players do this regardless of whether they’re knowledgeable in GTO or not, just by feel e.g. ”this guy never bluffs big on the river, so I can fold my top pair”. But having a theoretically sound understanding of the GTO baseline and why and how it makes sense to deviate from it in any given situation is probably helpful in order to get as close to max profit as possible.

And since GTO play is unexploitable it makes sense to try to play as close to it as one possibly can when facing unknown opponents. As soon as their tendencies are known we can start to deviate.