Slashdot Mirror


Researchers "Solve" Texas Hold'Em, Create Perfect Robotic Player

Jason Koebler writes The best limit Texas Hold'Em poker player in the world is a robot. Given enough hands, it will never, ever lose, regardless of what its opponent does or which cards it is dealt. Researchers at the University of Alberta essentially "brute forced" the game of limit poker, in which there are roughly 3 x 10^14 possible decisions. Cepheus runs through a massive table of all of these possible permutations of the game—the table itself is 11 terabytes of data—and decides what the best move is, regardless of opponent.

7 of 340 comments (clear)

  1. Perfect? Really? by igny · · Score: 4, Interesting

    Wouldn't another robot which knows of all possible decisions of this particular robot be better that this "Perfect Robotic Player"?

    --
    In theory there is no difference between theory and practice. In practice there is. - Yogi Berra
    1. Re: Perfect? Really? by Anonymous Coward · · Score: 2, Interesting

      It's called playing game theory optimally. You can make decisions that are profitable OVER TIME no matter what your opponent does. An important caveat is that when opponents make mistakes, the game theory optimal approach may be LESS profitable because you assume your opponent is playing correctly.

      Most importantly, no one is even close to solving no limit -- where you are allowed to vary your bet size. That changes everything.

    2. Re:Perfect? Really? by mythosaz · · Score: 3, Interesting

      "Tells," for short, are hardly the end-all be-all of poker play.

      Further, a well written poker robot would bluff a certain percentage of hands.

      My only objection with the computer picking the best play strategy for 3 x 10^14 possible decisions is that in exactly the same situation, the right move would be, often, to make different choices, especially based on the makeup of the table, how your previous hands played, how the table reacted to those hands, etc. "Regardless of what its opponent does" leaves me skeptical.

      I have no doubt that this can beat the field and win at poker. Bots have been doing that forever.

      ...but the right play in a limit game may be to 3-bet bluff or check-raise opponents based entirely on who they are or how they play. The entire idea that this bot is unbeatable, and a different bot with the same 3 x 10^14 decision tree and a different play-style (LAG/TAG) couldn't out-maneuver this bot is absurd. It's pure hubris by the team.

    3. Re:Perfect? Really? by mythosaz · · Score: 2, Interesting

      The best you can do with another robot is to tie.

      Poppycock. Pure nonsense.

      Another bot with the same 3 x 10^14 possible decisions could beat it over time based on play-style alone.

      What's your decision tree say to do here when you're check-raised? Oh, it says fold? I've discovered that? Guess what happens now...

  2. Re:Bets by hodet · · Score: 4, Interesting

    In limit poker there is more often a correct play. The odds would dictate, in a large pot to call that last bet because it is only a fraction of the pot. As long as your pot odds are better than your card odds it is correct to call, even if you only have one or two outs. In no limit where you can adjust the size of your bet, the correct bet is to give your opponents worse pot odds then their card odds. No bot can ever master no limit, it's not a card game at all. it's a people game played with cards.

  3. Re:I guess that means ... by runningduck · · Score: 4, Interesting

    What happens if someone else creates an identicly perfect robotic player and joins the table? If, as the researches claim, winning limit Texas Hold'em is a directly solvable problem then anybody else who tries to solve the problem will come up with the exact same solution.

    If these two robots played each other wouldn't the winner be determined by pure luck?

    --
    -rd
  4. Comment removed by account_deleted · · Score: 3, Interesting

    Comment removed based on user account deletion