The SnG Solver beta is now freely available for download.

Go here to get a beta key and download the installer: http://www.sngsolver.com/beta.html



Short FAQ:

Q: What is SnG Solver?

A: Its an equity calculator and push/fold strategy analyzer for NLHE STTs and MTT final tables.

Q: I already have one of those, why do I care?

A: I'm glad you asked. Here's a few points you might consider:

1. SnG Solver represents the first real improvement to ICM since Malmuth wrote about it over 20 years ago.
2. SnG Solver thrives where other programs fall flat on their faces. Super short stacks, out of position situations, bubble spots, etc... SnG Solver continues to give mathematically solid analysis without the need of awkward "fudge factors".
3. SnG Solver is aware of the value of position.
4. SnG Solver considers the movement of the blinds and the possible action on future rounds.
5. SnG Solver calculates the *best* Nash equilibrium approximate ranges available.

Q: Wow, that sounds pretty awesome.

A: Thats not a question, but yes, it is awesome.

nice work!

Just a quick FYI....

I'm aware of a bug that when you enter your activation key the first time you run the program, you will still continue to be prompted for a key when you push the "solve it" button.

The workaround is to close and re-open the application after you've successfully activated the app.

Sorry for the inconvenience, this will be fixed in an update *real soon*.

what is the major different from SNGWiz? can i use it to make push and fold table (or you have made one already)?

The short answer is 1) SNGWiz is wrong and 2) you could make the best push/fold table you've ever seen with it.

Okay, so here's the loooong answer...

Fundamentally, any program that uses the standard ICM equation (aka Malmuth-Harville, or even Malmuth-Weitzman) as the basis for its analysis (like SNGWiz does) is going to have some pretty big problems. This is because standard ICM equations are completely oblivious to position.

Standard ICM equations will tell you that 1000 chips UTG always has exactly the same equity as 1000 chips on the button. As I'm sure you know, and as anybody who knows anything about poker will know, this just cannot be right. In fact, its absurd. In fact most winning poker players will tell you that value of position is one of the *most important* concepts in all of poker strategy. Yet, the standard ICM equations have no clue about it.

You would never take strategy advice from another player that tried to tell you that position has no impact on chip equity, so why would you take advice from a computer program that thinks that?

This is the fundamental advantage of SnG Solver over every ICM based program out there... SnG Solver uses what I have called "Advanced ICM". This is a new model that is aware of position and will calculate vastly superior stack-to-equity estimations.


I mean no disrespect to SNGWiz. It has served the poker community well and helped to turn many a player into winning players. But at the end of the day, its analysis is not that deep and it is possible to do much, much better. Which is exactly what SnG Solver does.

any idea how I get Norton to NOT automatically remove your app?

Hmm... does this happen when running the installer? Or the program itself?

I'm afraid I dont know enough about Norton to give any specific advice.

I'll try and find out if I need to be put on a "safe list" of some kind.

*update*

Okay, I just found this thread that might help on one of the Norton forums: http://community.norton.com/t5/Norto.../192632#M96286

This is apparently a very well known issue that continues to be a serious problem. Franky, I find the whole situation pretty disgusting. I'm going to do what I can to bug Norton about this and to get SnG Solver to somehow pass their insane sniffer test.

Will your program give real time advise while I'm playing?

I have no intention to create a program that would likely violate the T&C of the major online sites. SnG Solver is meant to be a learning aide, not a playing aide.

In its current form, its about as "real-time" as its going to get. If you download and try it out you can see what its like.

Before I even run the installer.

So you mean as soon as you finish downloading, Norton deletes it? Wow, that's particularly disturbing since I have used the very common NSIS system to create my install EXE.

Correct

Did you look at that link I posted above? It has instructions for getting files out of Norton's "Quarantine". Have you tried that?

temp disabled insight protection then unquarantined installer and it worked

I think it has potential, but without knowing the mathematics behind the equilibrium result, it's hard to understand the result.

From your website, you say that the "Complete" ICM mode:
"Complete - It is in this mode that SnG Solver truly shines. It will take into account every possible outcome of the future round of play to arrive at the most accurate strategy analysis possible. Use this mode whenever possible"

It sounds like you simulate one round of blinds before calculating the equilibrium solution. Do you do this to determine the optimal "edge" that each player should require? How do you calculate equity based on the combination of stack size and position?

Can you provide more details on how are your calculations different from straight ICM?

Is there any available literature that you relied on to develop your "Advanced ICM"?

Thanks for making this available.

Is it possible to have a mac version of this?

Please really consider the opportunity to create one: there isn't any ICM calculator for mac and yours can be the first!

Please Please do it!

Well, its a bit complicated... But goes something like this...

For a given hand, I populate a game tree with every possible outcome for up to 3 of the players being "all-in". This, of course, can be repeated recursively to any depth. At each node of the tree, equilibrium strategies are calculated for every player. These strategies imply frequencies for each node. The whole ball of wax is collapsed back into the "present" and we get our answer. Got it?

The result is that when you have a short stack on the button, the math "knows" that the big stack is going to be trying to bully the table... and therefore how calculate what the *real* +EV play for you is.

There is no "edge" (to use the sngwiz term) or other artificial correction factor that must ultimately be based on "feel".

My "Advanced ICM" is an original model by me, so there's no white paper you'd be able to read about it. Its certainly based on/inspired by the original ICM models by Harville and Weitzman. I've applied concepts from Nash equilibria, "fictitious play", and Counterfactual Regret Minimization (CFRM). I've also drawn a lot from something called the "Multi-grid Algorithm" for how it handles error correction in very large systems of equations.

At this point, I'm not entirely decided on how much detail of my process I'm going to reveal. So I hope you'll forgive me if, at least for now, I play things a little close to my chest.


You're welcome! I hope you find it useful enough to eventually pay for it.

I have good news and I have less-than-good news...

I would *love* to do a Mac version. I am a semi-regular Mac user myself and am motivated to do it.

That said, I do not have plans to do it in the near future. I'm afraid its pretty far down on my "to-do" list. That said, your post alone has moved it up a few spots.

This software looks really good, but as others in the thread I have trouble downloading and installing because of Norton.

Have you tried having your software approved by Symantec or Norton or whatever the company is called?

Looks very interesting.

Are there any plans to enable importing hand histories?

How far into the future do you look and (assuming you don't expand the full game tree) are you just using "vanilla" ICM to evaluate the equities at the deepest level?

If this is the case then have you considered/evaluated the "horizon effect"?

Juk

Thanks!

Yes, hand history import is one of the next items on my list.

The future depth is always *at least* 1... that is, I always fully solve for at least the next hand. How far beyond that is somewhat variable... with extra depth focused on areas that are deemed to be more important. So I guess, yes, this does help minimize the "horizon effect" you mentioned. But I think also this is much less a problem for poker as opposed to, say, chess. In a game like chess, an entire branch of the game tree might have an actual "equity" of zero, but this is never true in poker. So, the effects of becoming "trapped" can never be as devastating.

And yes, I do use a standard (Harville) ICM to evaluate the leaf nodes... at least for now. But ironically, I might end up using a *worse* model (like proportional equity) in some spots instead because if faster leaf processing can lead to additional depth, the value of that depth can trump accuracy at the leafs.

I have contacted them about getting on their "white list", though I'm not sure how helpful this will ultimately be.

If you (or anybody having this kind of problem) could let me know exactly which of their products you're using (and exactly what version), it would be super helpful. That and at what point the trouble happens... e.g. the installer get quarantined right after download, or the app gets quarantined after running the installer, etc...

Can we see how these calcs are being made for advance icm?