wow, the simple postflop 30day license is listed for 69.99$ and at the end of the payment process they want me to pay ~86$

What is the VAT in your country ? Ireland, by any chance ?

Hello!
The payment is proceed by the external payment system Fastspring. Sometimes they add taxes (e.g. VAT) if they are need to be paid in your country. If you have any troubles to proceed the payment do not hesitate to contact our support, I hope they could help you

Can PIO solver or Simple Postflop currently or will either one in the future be able to calculate minimally exploitative preflop ranges for simple game trees, one bet size not including shove per street for example?

No preflop yet, but both want to get there in the not-too-distant future.

Thanks, looking forward to it.

Do these programs calculate the same, and how are the calculations done? I tried in CREV by putting turn actions and ranges, and then selecting a river. I would max explo both sides back and forth, but what ends up happening is it goes in circles. Just having difficulty wrapping my head around it.

The results of the calculations are the same. The cyclical nature of applying maximal exploitative calculations is a well known (and solved) theoretical problem. The two major solution techniques are CFRM (counter factual regret minimization) and Fictious Play and advanced variations of those are used by all the major solvers.

Fictious play is the easiest to understand conceptually and if you walk through using fictious play to solve rock paper scissors with pencil and paper it will give you a good idea of how to resolve the cyclical issues from just calculating max exploits (which in Rock Paper Scissors can always be either 100% rock, 100% paper, or 100% scissors).

Wikipedia has a tiny bit of info here: https://en.wikipedia.org/wiki/Fictitious_play and links to more in depth papers are in the wikipedia entry. If I recall correctly, Will Tiptons books also go through the basics of fictious play.

They don't although the results look very similar.
There are many ways to do those calculations but all of the currently popular one work by making small adjustments at every step. So for example the solver looks like EVs the current strategy produces for raise/call/fold and adjusts the frequencies in the direction of the best action.
CFR/fictitious play/gradient descent based algorithms are all variation of this and the only problem is to choose the one which converges fast, is fast to execute and uses as little memory as possible.
One way to describe them is that they are all different strategies to climb the hill which you have no map for and can only navigate using what you see at given time. The difference is about the way you choose your steps, how big steps you take and how you handle stepping into wrong direction.

Cepheus (the project of Alberta University which solved limit Holdem) uses CFR+ algorithm which has some nice properties (mainly being good at close to 0 accuracies and having good theoretical convergence properties) and some not so nice ones (converging slower at the beginning, using a lot more memory than needed, having additional overhead for regret->strategies and back calculations etc.).

We use something different which is in my view the best described as a mix between fictitious play and gradient descent. This needs more tuning to be good at close to 0 accuracy (although last version is already way better there) but it allows for optimizations which will not be possible when using CFR like algorithms and is already faster at getting to very good accuracy so in a sense it's more of a practical than academic way of doing things.

I have no idea what other solvers use but you can ask the authors.
As to the results: what we did was to compare our results to Cepheus and to other small solvers we have available. The thing is it's easy to make mistakes when comparing those things and you can't really achieve it until you see the whole tree which was the reason for my reaction at the beginning of this thread (you can't claim you verified anything if you don't have the whole tree).
It is true though that all signs suggest that all available commercial solvers produce very similar results and that it's not possible to produce significantly different strategy once you are very close to the equilibrium.

Going back to my hill climbing anology: sometimes you see a path which leads up and you decide to take a step, say 100 meters long one. What happens is you end up on the other side of the peak. You turn around, you see the path leading uphill again and you take another 100 meter leap only to go back where you were before. The trick is to take a smaller step. While this sound like an obvious thing, it's not as trivial to make work when you have more dimensions (because there are more hands and available actions) so some care is needed to avoid jumping around like that but still ensure reasonably fast climbing pace when you are far away from the peak (if you always take 1m steps it will take a long time to climb 8000m high mountain).

I see now, thanks for the explanations. My mind can stop being blown. Still it's blown a little.

Given the recent updates, Which is the best/fastest now?

Pio ... and it's not even close.

Doug Polk AKA WCGrider recently answered some questions on reddit and it seems he doesn't have much good things to say about PIOsolver, which I find suprising. I hope others can elaborate on what he's trying to say.

"I think PioSolver is basically terrible. I don't know what the high stakes six max players are using, and I'm not claiming to be one of the top players in those games. But what I can say is that the players in HU that use this type of software to play, end up playing horrific post flop.

It actually upsets me a bit how bad the conclusions are, in that they are very far off what I believe to be correct. I DO think however, it can be a useful tool to think about situations/how to play. but the actual conclusions are not valuable (and often you need to make far too many assumptions for the conclusion to even be correct)."

Image:


Link: https://www.reddit.com/r/acesup/comm...hand_analysis/

Prop to promote his 7$ preflop engine.

I think he is just using "PioSolver" as a representative of GTO Solvers in general, so it's not on topic in this thread.

Just wanted to stop by and make a couple points.

The Postflop Engine is an extremely basic product to help people begin to think about poker strategically. I don't want someone to think that they are getting solutions to complicated problems. Our goal with the product is to have something introductory for people who want to understand the underlying idea of how decisions in poker are made.

I think that almost all of the solvers out there at the moment are pretty terrible. I will reiterate what I posted on reddit, but I do not know what the six max players are using or the conclusions that they have come to. But I absolutely can say that in heads up no limit, I think most of the players using solver type strategies are playing pretty laughably bad.

I will probably write something more in depth down the line, although I would probably need to give them a good look through before doing so and I am not sure I have the time at the moment.

^^

It will certainly be very interesting indeed to many people to learn what could possibly be so "pretty laughably bad" when using GTO strats - aside from obviously not exploiting the opponent to the max ...

Hypothetically would you be willing to play HU for real money against a bot that just plays a solution from one of these "terrible" solvers under the restriction that both you and the bot would only be allowed to use a specific set of agreed upon bet sizes (you could pick a set of a few bet sizes on each street and preflop that you like to use and we would solve given that)? That is do you not believe the math behind CFRM or the conclusions of game theory on the unbeatibility of Nash Equilibrium play in HU games?

As the maker of one of these programs whose product you are trying to publicly discredit (without any specific evidence and while trying to also sell your own / your friends product) I trust that the results of solver software are just pure math. The theory behind computational GTO and the CFRM algorithm has been peer reviewed by top mathematicians in the world so I'd be pretty open to a challenge if it was within my budget (which is of course much smaller than yours I'm sure and maybe not worth your time).

swc123-
I bet you can find backers.

Pretty sure WCG either means

1) He thinks people are not implementing the strategies well enough or are trying to play too complicated of strategies and playing them very poorly

2) Abstractions people are using are not good enough

3) Population is too exploitable to not try to maximize our EV with quite exploitative play and 3b) doesn't think that people working with solvers are doing enough exploitative work

4) He has just gone full Durrrr

lol

Nice challenge - I hope this gets up

The nice challenge would be for a real HU match - not some abstraction with limited bet sizes.
Playing with forced betsizes is not real NLHE and such a match wouldn't prove anything.

Me - These solvers are bad at poker

You - What about if we played a game that isnt poker, for money?

Last I checked when I play poker we dont all sit down and agree on bet sizes. I find it a bit shocking that you even ask me the question above here. You do realize that if pick out the bet sizes you can use, that bots have a very large edge on people(As of course you do realize, because this is why you are asking me this). When you are solving for nash equilibrium, you cant solve for a simplified version of it, and then play it in the actual game. You just have a solution for a far too simple version of the game, which then has no application into the games we play.

Also for what its worth, solvers/bots will 100% eventually beat humans. I am not taking a Durr-esqe "But the bot cant make any reads" approach. Essentially you are solving for a solution that isnt even what poker is, but eventually that will exist.

Its hard to get the right answer if you are not even asking the right questions.

The reason I asked you if you would play such a bot is that your product that you are trying to sell also does not tell you how to react to every possible bet size (as far as I can tell) and you never explicitly mentioned bet sizing in your vague criticism of existing software, so I assumed your issues were not related to bet sizing. Thus I thought you were taking a Durr-esque view in which case you would presumably happily play.

If you just don't think the bet sizing abstraction is sufficient that is certainly an opinion that cannot be mathematically proven false. We'd have to program a bot to interpolate our GTO solutions and react to any bet size to challenge you in that case which I expect is possible with today's technology but it would be expensive/time consuming and we have no real incentive to do so. FWIW there was a poll by Sklansky of MSNL+ regs asking if they would take a rake reduction if they were only allowed to use 1 of 3 or 4 predetermined bet sizes on every street. Popular consensus that this would be +EV. I'm not personally convinced that an extremely strong strategy requires more than say 1-3 sizes per street (especially in pot limit) but I also can't mathematically prove that to you yet.

Also, as I'm sure you are aware many (almost all?) of the top HU hypers players use solver software so it sounds like given your beliefs you have the opportunity to win a lot of money off them.

It sounds like we are in agreement that the solver software does exactly what it claims to do, which is to solve a game with a bet sizing abstraction (otherwise you would presumably have accepted a challenge to play the abstracted game). How big an abstraction needs to be to be "good" and how good the users of such software that you've played against are at creating good bet sizing abstractions for your games are definitely valid concerns.