So there is no such thing as "playing better than someone else" in poker then? Since you cannot guarantee an outcome you should discard it?

Of course not. Playing better than someone else is how we win at poker. That has nothing to do with the impossibility of 'solving' poker, as if it were an equation with a correct answer.

In 2095, Skynet creates the T3000. It can read your pulse rate from across a table, identify pupil dilation and microscopic hands shakes. It is physically indistinguishable from the usual degenerates that still play poker.

B&M MTTs are solved. Game over fellas.

On a more serious note, chess has only recently been "solved" and even that is questionable. Deep stack, big field MTT poker, in isolation against a game of chess, must have almost infinitely more variable plays.

The power of a bot is in it's ability to perform millions of calculations per second. So two things..

1) Every possible move in chess has a finite number of options. In poker that's not the case once you factor in bet sizing. Even over a sequence of 5 or 6 bets in one hand, the difference grows exponentially.

2) In chess you can see your opponents "cards". In poker, Jesus Ferguson excepted, not so much. This is massive from an AI perspective.

So yeah, I genuinely don't see MTTs solved in my lifetime. But then i'm 89 so hey. Or something.

Not sure you understand "solved" that much. Of course you can never predict the outcome of one hand of poker or one tournament. The target is always the "long run". So solving poker in a way that you don't lose in the long run its certainly possible if we had all the processing power we wanted.

I will use your example and say that if you had multiple copies of the bot that was playing perfectly they would all end up break even (draw) if there was no rake in the long run, exactly as chess.

GTO assumes best move (or mixed strategy) against best move.

Based on that, by symmetry all participants have equal expected value, which because of the rake is negative. The GTO move is to unregister the tournament before it starts. - MTTs solved QED, we don't need to solve the strictly dominated lines involving playing.

Of course, If you sit and play in a raked game at all you are have already shat all over the assumptions of GTO and are playing exploitatively.

I don't think MTTs are going to be solved as equilibrium in multiway pots is a shaky concept and even more so if there are ICM considerations (so it's not zero sum for players involved in a given pot).

It's possible someone comes up with very strong AI based on exploitive play but I don't think anyone is going to "solve it", at least not during my lifetime.

I think the only games in danger of being solved are all kind of HU poker. 3 handed might be approximated so closely that it's effectively solved for practical purposes. Maybe 4+ handed games will require some rule changes to make it more difficult (encourage more multiway pots) but again, equilibrium isn't the end of it in 3+ handed poker as it is in heads-up.

In other words, the question of wether poker can be solved or not is irrelevant to humans playing the game.

The thing that has a chance of adoption is play fixed limit pre-flop, then no-limit post flop.

Thread/

Poster is creator of the leading GTO software for those not aware.

Sent from my SM-G900F using Tapatalk

Is there any talk of heads up limit holdem specialists challenging Cepheus? Or is it accepted that it's better than any human now?

The word "solved" is egghead stuff. But practically, it looks like 2/3 solved to me, including cash gto and stack size gto, and some practical pot control stuffs, and money and bubble kind of stuffs.

This looks pretty good to me.

Who knows with computing power improvements and new ideas what the long term actually holds, but I can't imagine MTTs being in any form of trouble in the foreseeable future.

Ofc its gonna be solved, its just a matter of time - definitely within the next 20-30 years. People really underestimate the power that quantum computing and AI will have in the future.

^^^lol

Mike Sexton solved MTT confirmed GOAT

The problem with this is even if a perfect unexploitable strategy is solved for MTTs it will never be the optimal strategy in MTTs.

There will always be seats filled with players that don't care about playing that strategy; chasing draws with improper odds, defending unprofitable hands preflop to 3bets, etc. Playing with many other deviations from what would be this "GTO strategy". So as long as they take up seats exploitable deviations from the pros will be the more optimal counter strategy, than just sitting there playing "GTO".

Indeed, deviations from GTO will always be more profitable

This.

The problem is that the solution to multiway games just doesn't exist. Yes, you can find an equilibrium but that may turn to be a losing strategy if someone is deviating and you are not adjusting. While in HU poker you are guaranteed not to lose (and you will most likely win against someone not following something close to equilibrium) in multiway it's not true. It's even less true in games like MTTs where following the equilibrium (one of many possible) might be just terrible.

it will be solved when you can solve flipping a ****ing coin

It doesn't have to be solved to get killed.

Afaik, PLO bots on PS didn't play some 'perfect' GTO. They just played better than most human players of that field. This is all it really comes down to, imo.

Saying, the more complex any given poker variant is for human brain to solve/digest/apply concepts in-game, the easier it is for a bot to exploit humans at it.

Makes sense?

^Thanks Captain Obvious

I think GTO MTT will get solved (**) and the whole population of players is drifting toward the GTO solutions to some extent. A perfect exploiting strategy is bound to be better than GTO but when we do find the GTO solution it will probably look quite like current, very good players think of as exploiting play. We currently aren't really that good at exploitative strategy anyway, the GTO if it existed is likely to do as well as existing good players.

** close enough solutions, even without knowledge of all the other table stacks, in this game knowing just the players left alive and assuming the distribution of stacks elsewhere on other tables, as we humans have to try to do now.

I also think that if such a GTO bot existed today it would beat all levels easily and nearly as well as the very best exploiting players in today's game (***). The best exploitative strategy will be better but the plain vanilla GTO would do really well atm in most situations. Luckily GTO solutions don't yet exist.

*** playing at the micro levels today it is likely current good exploitative players can easily outperform GTO, the higher the stakes the more likely GTO is as good as present players.

For an example atm the HRC can do a really good job of raise, 3bet call or fold solutions for approx 20bb deep play. I use this tool a lot analyzing hands and usually have to reduce the aggression of the typical 'GTO' like solution to get a better, more exploitative result for particular opponents. The thing is though that these 'gto' like moves given for these solutions are very strong and better play than I see from most of my opponents in the current climate.

eg, for chip ev, 20bb deep SB vs BB, 8 handed with antes, HRC has a 2.5bb raise for the SB of 65% of hands, the BB response is to re-raise allin with the top 31.5% of hands and call the rest. The SB responds to the 3-bet by calling the BB allin with 26.3% (down to hands like J9s, 22, A4o)

I think the above style of play is very much better than I see at the moment at tables and it would pay off against the current type of players I see. You could get even better by adjusting for the actual play seen, as I try to do, but this as a default is really good, and really aggressive. In this solution the BB call path does no carry on through to post flop play but the current batch of solvers could do a good job at this.

These are for chipev but HRC can do a solution for icm with an adjustment for other tables in play and as I have mentioned this is only for one spot but playing this well at all the hand positions is so strong it would beat most of today's games.

Methinks this thread doth amuse the poker gods.