GTO is dead ? Poker is alive ??!!

    • asimos
      asimos
      Coach
      Coach
      Joined: 21.07.2011 Posts: 1,438
      This guy is a Greek Professor in MIT.

      In his web page:

      http://people.csail.mit.edu/costis/

      and in the section Complexity of Equilibria one can read in the 6th line:

      "...Finally, I recently showed that even arriving at an approximate Nash equilibrium can be computationally intractable."

      wow !!! :f_drink: :f_drink: :f_love:
  • 8 replies
    • YohanN7
      YohanN7
      Bronze
      Joined: 15.06.2009 Posts: 4,629
      Yes, that is good new, and the particularly good part is that two-player games were under consideration. The bad news is that, in spite of this, a bot playing approximate (for human standards, perfect) GTO in FL Hold'em has been developed (after the Greek guy published his paper), namely Cepheus.

      I don't know how much bearing Daskalakis' paper actually has on NL Hold'em. For that one would have to penetrate the paper and references therein, and they are decidedly nontrivial, requiring a certain academic level in the field (game theory). I might try to just ask him by e-mail. University professors tend to be most forthcoming and helpful when you show interest in their field.
    • RomikZb
      RomikZb
      Bronze
      Joined: 12.12.2009 Posts: 96
      Just look at this: http://simplepostflop.com/en/Preflop
    • YohanN7
      YohanN7
      Bronze
      Joined: 15.06.2009 Posts: 4,629
      I wouldn't put any trust in commercial claimed-to-be GTO strategies. If they knew how to really do it, they'd not sell it for a fistful of dollars.
    • asimos
      asimos
      Coach
      Coach
      Joined: 21.07.2011 Posts: 1,438
      Originally posted by RomikZb
      Just look at this: http://simplepostflop.com/en/Preflop
      Those calculations are possible to be done for given bet sizings I presume... but the game is no limit holdem.
    • RomikZb
      RomikZb
      Bronze
      Joined: 12.12.2009 Posts: 96
      It is possible to build limit Holdem tree in this programm
    • RomikZb
      RomikZb
      Bronze
      Joined: 12.12.2009 Posts: 96
      Originally posted by YohanN7
      I wouldn't put any trust in commercial claimed-to-be GTO strategies. If they knew how to really do it, they'd not sell it for a fistful of dollars.
      We really know how to do it. I am the developer of this program. So, we don't play poker, we just develop software that is why we just sell it.
    • YohanN7
      YohanN7
      Bronze
      Joined: 15.06.2009 Posts: 4,629
      Originally posted by RomikZb
      Originally posted by YohanN7
      I wouldn't put any trust in commercial claimed-to-be GTO strategies. If they knew how to really do it, they'd not sell it for a fistful of dollars.
      We really know how to do it. I am the developer of this program. So, we don't play poker, we just develop software that is why we just sell it.
      No, you most likely do not know how to do it. It took a team of scientists (real ones) at a well-known university many years to develop Cepheus. Also, your brief "explanation" of what you are doing at simplepostflop,
      The calculations are performed without abstractions, using exhaustive preflop to river tree with every possible runout,
      indicates that you fiddle with the truth, misusing the term GTO in order to sell.

      Whenever you manage to get your "GTO strategy" published in a peer-reviewed scientific journal (like the Cepheus team finally managed), then I might take you seriously provided you get a few citations in other scientific peer-reviewed journals.
    • GingerKid
      GingerKid
      Black
      Joined: 05.08.2007 Posts: 5,565
      Originally posted by RomikZb
      Originally posted by YohanN7
      I wouldn't put any trust in commercial claimed-to-be GTO strategies. If they knew how to really do it, they'd not sell it for a fistful of dollars.
      We really know how to do it. I am the developer of this program. So, we don't play poker, we just develop software that is why we just sell it.
      If you claim that you know how to do it, can you please explain how you did it? I am a developer as well, so you can go into details if you want. Or even better if you explained it already in your software, as I would expect it as a user of your software, to know what is software doing exactly. Or if you followed some algorithm, maybe you can provide reference.