[JohnMakin]

Not really. Maybe in very specific applications of limit (fixed bet size) hold'em, but no limit texas hold'em, the most popular variant online, is very much unsolved, especially in multi-way pots. There are simply too many variables and strategies involved to calculate quickly enough on the fly. For games like omaha, which uses 4 hole cards, this is even harder.

Due to advancement of theory and study and popularity over the last ~20 years though, it's definitely much harder to be successful than it used to be.

[mattmaroon]

I don’t believe it. There are just as many variables involved in writing a short story.

Anyone who thinks machine learning can’t conquer poker is fooling themselves. I used to have bots collect every hand played on major poker sites in the early aughts so I’m sure there’s infinite training data.

And if it can be done we know there’s sufficient financial incentive. So I (former long time professional poker player) feel reasonably confident online poker must be unwinnable by now.

[JohnMakin]

You can feel all you want - there is no evidence this is happening.It would be massive news in the professional community. You’d need a comprehensive GTO strategy which doesn’t really exist, especially in multiway pots. Best we have right now are GTO (game theory optimal) solvers which need tons of assumptions plugged into them and require enormous amounts of memory and time to spit out results, and we still don’t really understand a lot about the underlying theory.

It doesn’t matter how many hands you “train” something on. Poker is a game of incomplete information and many assumptions must be made about an opponent’s range, bluff frequency, etc. One small tweak in assumptions and the entire GTO output changes with solvers. It’s very difficult to get these assumptions right. As I said it’s an unsolved game. Even the GTO solvers only work in 1 on 1 pots (assuming the assumptions you’re working with are close).

Respectfully, although you claim to be a former online professional (I have played for 20 years, at times professionally) - you don’t seem to understand what you are talking about.

[luxcem]

You don't even need training data, a bot that play itself à la AlphaZero will eventually collect more data than there are of actual games.

[teapot7]

> I don’t believe it. There are just as many variables involved in writing a short story.

Surely you're not implying that writing a good short story is a solved problem for computers?

[bluecheese452]

Heads up no limit holdem bots crush the best players in the world even 200bbs deep. So kind of like the same situation as chess. Not solved but not beatable.

[JohnMakin]

There haven’t been any real online NLHE heads up games online in at least a decade. so it’s kind of a moot point. Not sure which bots or players you are referring to - these games don’t exist online in any meaningful volume.

[hiatus]

Can you link to more info?