The feeling didn’t last. Shortly after our match, Prof. Bowling sent me an email, debriefing my performance against his creation by analyzing which parts of my success came from skill and which parts had emerged from the thick fog of randomness inherent in no-limit poker. He wrote, “You should expect to win 42% (margin of error of 5%) of your matches against DeepStack. While you won 15 and lost 14, your play (after removing luck) suggests you should have won 12 matches and lost 17.”
It is a model.
Thanks for the Archive link so I could read all of the article and the comments. Good post - should keep me out of the poker rooms.
How long until DeepStack cuts a deal with the random-card generator?
I used to play semi-pro a couple decades ago when Poker was having its hey day. Won a few modest tournaments. Played a lot of high stakes limit poker and no-limit. Limit poker is a little more mechanical, and the most you can lose in a pot is limited to the bet size though with several players your position matters a lot since you can get raised in front of you and behind you if you don’t get the other players to fold. No limit, well, you can bet any amount at any point which makes it scarier when faced with a big bet decision.
Today the game is tougher. Cash games are truth or dare. Tournament games are as much about not getting knocked out out of the tournament as they are accumulating chips. So tournaments are about saving chips as much as winning chips. Reading your opponents’ playing style and figuring out which kinds of hands they are likely to raise or call with, and then reading the board to see if those hands are likely to have hit or miss or give him/her a drawing chance. Which of course means opponents have to vary their starting hands and style a bit just to make them harder to read over the long haul. A lot of bluffing and semi-bluffing due to that psychology.
That's a pretty lame interpretation.
The computer is getting luck, too, unless this model cheats all the time.
I installed the "nathan-149/hover-paywalls-browser-extension" extension from GitHub into Google Chrome to get around WSJ (and other) paywalls.
Poker Ping
When the machine can learn to get drunk, get distracted by the waitress, get gun shy or overconfident, and have an overwhelming urge to take a pee, I’ll believe in AI. Until then, it’s just math, not poker.