Poker-AI.org

I'm curious how the current state of the art heads up poker bots fare against human pros. Have there been any matches recently between pros and the best bots? My hunch is bots would be better in limit hold'em whereas humans better in no limit but I haven't been able to find any examples of recent matches.

http://www.pokerlistings.com/videos/v/m ... i-c8846788 is the only recent or NL competition I know of, but 1500 hands isn't statistically significant. Variance is so large in NL that I doubt that any pro will ever play sufficient hands to get a statistically significant result against a GTO bot. I guess it might be possible for an exploitative bot to beat a human by a statistically significant margin, but the impression I get is that exploitative bots are not as well developed as GTO bots yet.

I agree with both of you in general.

Shameless plug, but my NL equilibrium strategy (see my profile) is probably 2nd strongest in the world at the moment, and I've had a couple pros play it in private and it's held up well, but can be beat. There are some counter-intuitive exploits you can use against it, mainly because of some design decisions I made related to hardware (re: budget) limitations when constructing it.

PokerSnowie is good for generic testing, like multi-way pots with different stack sizes, but for strictly HU competition I don't think it's that strong. Impossible to know for sure since they don't submit to the ACPC.

I plan to enter mine next year. It is a very aggressive exploitative rules based bot. I have taken a break from it to work on PLO but plan to finish it up later this year although it would probably do ok just the way it is.

What are these counter-intuitive exploits & design decisions?