I call bs to this. No bot can conquer any form of poker including fixed limit.

That works until computing power improves 1000 fold and anyone can generate their own GTO (or close) strategy from a random seed strategy.

he is a High stakes HULHE pro lol

go away before you embrass yourself any further

That's never going to happen cause unlike you, HS pros are smart enough to understand why this thing is the real deal. All you have to do is read the paper and understand the math and logic.

They have evidence. Too bad it's beyond your abilities to understand it.

Anyways, I had fund with this but I won't waste more time on you. If you're a troll then well played sir. If not, then I feel sorry for your gene pool.

Could a mod please change the title to specify Limit Hold Em?

Okay makes sense and ties in with what David S was saying.

If I was playing heads up I think I'd still cap with aces 😀

Sent from my LG-D855 using 2+2 Forums

A match between one or more top pros and this program may happen, perhaps as an exhibition, perhaps for other reasons. Onlookers are likely to question the sample size in such a match.

Certainly top pros will play the program in private.

However, I agree with what I think are your thoughts about Donkem: that he is incorrect in a number of his thoughts as expressed in this forum.

Im sure this topic will have further developing, will see who's right.

el oh el poker luddites. Welcome to the 21st century.

Anyways, congratulations on the achievement.
Few questions to the team:

i) Would you regard LHE HU is from here onwards a dead game online? I.e. would you recommend that sites get rid of it because there will be a too significant risk of GTO botting. I read some previous post that "even though there is a solution, it's not useful for real-time play" - I didn't really understand it. Is this really the case? Why is your website processing queries so fast?

ii) I always thought 2-7TD would be solved first, even though there is probably more nodes in that game tree. Maybe a sub solution of 2-7 where one would disregard analyzing further certain "clear-fold" starting hand combos. Do you think 2-7 would be from technical standpoint easy to solve as well, had the AI teams put their effort into it?

iii) When will the bot be back online? I'm very keen on playing it, as I've always regarded my LHE HU game much stronger than NL. I played PokerSnowie a bit - and although a limited sample only - felt like it wasn't that good. Certainly at that time I thought I would not be able to beat players better than me.


If FLHU is truly solved, then RIP woofer & hats off. I guess he was the last guy to be the king of the hill before computers took over.

PS. This probably belongs to that "future of online poker" topic but w/e. Do you know if e.g. Apple tablets work so that iOS can prevent you from running 3rd party software if you're playing PokerStars app? I mean, if that's the case then I think a likely scenario is that in future online poker might be only available on specific devices. I guess it would add longevity for the game as well if usage of real-time huds etc. would be prevented.

I'll happily go on record as saying heads-up NLHE 100b won't be solved in anyone's lifetime. Adding more players won't make that problem easier...

It's an interesting question of how small the stacks would need to be. We could certainly solve a 2 chip game. I need more free time :/

Also, I've been consulting for the CPRG and generating samples against their AIs since 2007ish or so. I definitely wouldn't be near the top of the heap anymore, but I imagine I can still play a world class game. The earlier, inferior versions of the AI were starting to edge me out years ago (I never logged any hands against Cepheus because it wasn't useful to do so due to the increasing ease of best-response calculations).

Member of the Flat Earth Society imo. ^^^

The best HULHE players might be able to give the bot a decent run for its money over a small sample size, because edges are really thin and variance is high, but no human on earth can beat this thing over a significant sample size. Depending on how big the bot's edge is, "significant" might be half a million hands or more. Anyone could get lucky and win in the short run. Mathematical proofs need large numbers. The Cepheus team have some very large numbers.

I'm sure that this is an extremely stupid question, but here goes:

Is there a reason why this solution couldn't be applied to multi-player fixed limit holdem when you are on BB facing a single raise from a position other than the SB (and it would be headsup pot with only you and the original raiser)?

I'm assuming that there is, as it would be way looser and way more aggressive than the way humans play. But I'm not smart enough to understand the reason without someone pointing it out.

The starting hand ranges for the open raise would be different and so everything that follows after that would be different, too.

We probably didn't help an already-unhealthy online game. I certainly wouldn't recommend sites or players kill the game online because I don't think there will be a real problem, but a) what possibly authority do I have to make that recommendation? b) I could be wrong, and c) at the end of the day, it's all up to what the players want.

I don't think the strategy tool will be a big problem because it's slow: that it's as fast as it is, is a bit of a miracle. It's surprisingly idle: I can watch the back end try serving requests, and it's sitting idle pretty much the entire time I check. If people ever start trying to use it in bulk, they're going to be very disappointed in the response time.

I'm not very familiar, so I could be getting something, but it looks a lot bigger. It's generally not very easy to carve of pieces of a poker game and still end up with something where you can provably say you did things right in the full game.

No real ETA, sorry. Hopefully Real Soon Now? It was up this morning, ever so briefly, before crashing back down. The guy working on that part has apparently been working on this more or less non-stop since Thursday.


Sorry! I know nothing about the details of iOS tablets.

Let's say CO raises, the button and SB fold and you are the BB.

To figure out the GTO response for you in this spot you would first need to calculate a GTO strategy for the CO's opening range (I think, not 100% on this).

To do that you would have to first solve a 4 player game.

because the first raise isn't heads up

can you estimate your/someone like you winrate (lossrate I guess) be vs cepheus? If you had a year to train for it how good of a winrate could you get

Because they're are two entirely different games.

The range someone should raise varies on based on position. So you can't defend with a static range. Also you're assuming SB folds always so there's an extra SB in the pot that you need to take account for. The solution they have posted now wouldn't apply to LHE HU game where each player posts blinds + an ante of 1/2 SB.

Make the river bet 3 BB instead of 2 and they have to deal another 20 trillion hands.

Put a preflop flop out where those five cards are dead and they have to deal more than 50 quintillion.

You're right, but I assume it would be trivial to adjust the solution to a different blind structure. At worst you could rerun the algorithm with adjusted parameters in the same time it took for the original solution.

So I'm basically querying a supercomputer, no? A one which regular people don't have access to? Could someone grab the source codes and be able to build a similar system which can query the database almost real-time with a moderate investment? (Moderate in this case something in the range of 10-50k?)

No worries that question was addressed to poker community at large.

There are a lot of evidence, you just have to read it. The fact that you bring up NL shows that you know nothing about the math behind a thing like this. NL is magnitudes harder to "solve" than FL.

I've absently wondered a few times over the last bunch of years if it would have served poker better for the sites to take more of a 'league play' format rather than the games being static forever. Say there's some site, HUpoker.com, and every year it offers three different HU formats that people can compete at and at the end of the year they go away. These could have been things like Pineapple or other relatively simple variants that would have kept the games fresh, interesting, and competitive. I'm not sure if the culture of poker players at large would have supported this, but if it had become accepted at some point I guess it would have been interesting.

No surprises here.

Congrats to the programmers.

NL is next..