This is the support thread for SnG Solver, a tournament strategy calculator.

Write on twitter, double in this thread

1) Can i buy sngsolver by pokerstars transfer?
2)download links are incorrectly ( http://www.sharktapus.com/downloads/...etup_1.0.0.exe )
3) are you read on 2+2 my ideas of name?xD

And more, are work with files with HH accepted in new version?

Unfortunately no... I live in the USA, so access to PokerStars is a little problematic at the moment.

I got a little ahead of myself there for a moment... but the download links should be all working now.


I did, thanks!... I'll be sending you a PM about that


I'm afraid not yet. But my top priorities are now 1) better documentation on the website and 2) hand history import... So it hopefully shouldnt be too long from now.


I'll be writing a more detailed post about version 1.0 a bit later... but it is now available for download. There is a free 30-day trial period that starts when you install (and actually, it wont start "day 1" until Jan 10th... so you get 37 free days if you install today )

Enjoy!

New wishes for a new year.
I remember you've said that manual ranges would make calculations too complicated.
Thing is, 2 types of range expansion should be enough for almost any spot. For example, heads up 10BB deep, I want to know the calling ranges if villian is pushing 55%. Problem is he should push Kx and not 54s.
I noticed that calling and pushing ranges aren't the same, but there are still some spots that would need to adjust.
One solution si to make solver notice before calculation which range expansion is more realistic, and other is to make 2 or 3 default range expansions and let player decide for himself.
Is this possible?

ginandbread,

There's good news, and there's bad news...

The good news is that I now do believe that I will be able to implement totally unrestricted ranges in to the SnG Solver system...

The bad news is that I probably wont get to this for some time.

... but the other good news is that the reason I wont get to it for a while is that it needs to come *after* another other major feature is added. And this other feature, I believe, is going to be a real game-changer...

So, right now, the feature roadmap looks something like this....

1. Hand history import
2. ***Top Secret Game-Changer***
3. Unrestricted hand ranges

What would a version 1.0 be without a few bugs?

If have installed 1.0.0 and had trouble getting the 30-day free trial period to start (that is, it wouldnt let you run without entering a license key), you should download and install the upgrade to version 1.0.1 now available on the downloads page.

Sorry about that.

So how long trial after download ?

30 days.

ICM is dead! The future of tournament poker analysis is the Predictive Simulation Model (PSM)...

Observant users may have noticed I've been replacing all the references to "Advanced ICM" with "Predictive Simulation". This is the new name I've settled on for the algorithm. Big thanks to everyone who submitted suggestions.


Also, there is a new version (1.0.2) available for download in the usual place.

Just a few small additions mostly related to automatically checking for new versions.

The next release should contain some kind of hand history import... knock on wood...

Hi,
This is a very interesting softwer, I am quite excited about it, I think calculating an sng hand as a part of the whole game and not only as a separate hand is the right way.

However, I am not sure why you do not reveal the exact calculation method. My concern is that deciding if ranges given by your program are better or worse approximations than those given by Wizard or holdemresources is quite dificult. One has to believe that you have good mathematical and theoretical background and you have better understanding of tournament game than anyone else who made calculators for sngs so far.

Although I understand the basic concepts that you use, (as a non-matematitian can understand them) and it seems to be good and makes sense, hiding the exact calculations worries me. Poker is business, investing money based on trust in others competence is usually not enough.

I understand that there can be reasons that keeps you from publishing your idea, but is it maybe possible to show at least some simulation results that show that decisions based on your model outperform significantly the decisions based on the already known ICM modell? I mean more than like "should be better because equities calculated in my model make more sence than in old models so theoretically it shloud be better".

I think this approach you started is really a good way, I would only need some proof that it really leads to better decisions.

Is the algorithm used for calcs not public? I think it should be if the software should be taken seriously. I understand you want to protect your software, Im a developer too, but if we dont know the method used to make the calcs there is really no reason to switch from what has proven itself to be quite a good way of doing things.

Mvh
Inga

Believe it or not guys, I'm really not trying to hide anything. I thought I had pretty much laid everything out in the old beta thread... but I know not everyone has read all that (nor would I expect anyone to).

In fact, I consider that the success of SnG Solver is going to be directly tied to my ability to educate the poker community about how it works and its benefits... its *exact* benefits. To this end, I'm working on a couple blog/ariticles that detail the algorithm, gives side-by-side comparisons, etc...

This will take a little bit of time though (as SnG Solver is pretty much a one-man operation for now)... but for anyone who may have missed the old thread, I'll give a quick rundown and an example here:

In most ways, SnG Solver works just like any other strategy calculator you may have used. And, if you set the "Predictive Simulation" mode to "off", you will see that the results will be essentially identical to those "other" calculators.

All equity/strategy calculators do, fundamentally, the same thing: they calculate the risk vs reward of situation. And all tournament equity/strategy calculators do just one extra step that makes them different from a regular equity calculator (e.g. PokerStove): they convert chipEV to $EV. Thats it. Everything else is just window dressing.

In most programs (well, all of them except SnG Solver), this conversion from cEV to $EV is done with the ICM equations.

A program like sngwiz then calculates the $EV for the different possible outcomes of a situation ( what if I fold?...what if I call and win?... what if I call and lose?.. etc..) and then presents them for comparison.

If one sums up all of the "what-ifs" for a hand, you arrive at a kind of "composite $EV" for each stack.

So now rewind a bit... back to when I mentioned using ICM to convert cEV to $EV. What if, instead of using ICM to convert cEV to $EV, we use this "composite $EV"? What you get is a new $EV that has taken into consideration the future round of play. In other words, we calculate the "future $EV" to find the present $EV. This process could then be repeated
ad infinitum (or until every possible future game state is evaluated and the game is "solved").

This is what the Predictive Simulation algorithm in SnG Solver does. It recursively evaluates future $EV to arrive at $EV for the present situation, thereby properly considering the effects of position, impending blinds, etc... Of course, in order to evaluate a future situation, you must know how each player is going to act. This is where the "Predictive" part of the algorithm comes in. For each potential future, SnG Solver computes an approximate Nash equilibrium and assumes that that is how each player will act. Naturally, not all (or even any) players will be playing exactly according to a Nash equilibrium strategy, but such a strategy should be *reasonably close* to what a solid player would do and it ultimately make a good theoretical foundation to build on.

To see this in action, consider the following hand:

BB/SB: 1000/500
Payouts: 65-35
Stacks-
BTN: 700
SB (Hero): 600
BB: 7700

Hero is dealt 52o.

SB folds, Hero?

Any ICM based program will tell you that the SB must call with ATC, and therefore we call with our 52o.

There is no good player that would agree with this analysis (if you do agree with this analysis, then I have a prop bet I would like to make with you ).

It is also true that the Predictive Simulation Model of SnG Solver does not agree with this either. SnG Solver recommends calling with a range of about 17%. I think you'll find that most good players will agree that this is a reasonable range.


In case anyone missed it, here was a post where I made some more comparisons with some other ICM based programs: http://forumserver.twoplustwo.com/sh...2&postcount=38


So, I hope that helps to remove some of the "mystery"... by all means, if you've got questions, ask away. Like I said, I'm working on a number of articles to further illuminate what going on under the hood of SnG Solver.


Also, I should mention that the "Details View" as it stands now, is just the start of whats to come. Eventually that panel will contain so much data it will probably be considered overwhelming and you'll begging me to hide things from you.

Hi Jason,

first of all, you are doing a fantastic job here!
I am thinking about possible ways to actually compare the quality of your equity estimates with the ones we get using ICM.
Let us suppose that your estimates are more accurate than the ICM estimates.

Clearly the ICM estimate is static in the sense that it doesnot depend on the position and size of the blinds.
Your model is in this sense a forward looking one.

So I am thinking that if your estimates are the better ones then of course playing the resulting Nashranges should earn more money or loss less money than the Nashranges resulting from the static ICM in the long run. (At least if we suppose that the other players are playing "ICM Nash" for simplicity.

So what do you think about a simulation in which you let one player play your ranges vs two players using the common ICM Nashranges on a shorthanded SnG bubble with eg. every player starting with 10BB.

If your ranges and therefore your strategies are superior to the ICM Nashranges than your player should clearly show a profit compared to some other player who is playing the ICM Nashranges in the exact same situation.

If you want to consider a simpler example you could simulate a HU match in the same fashion.


This simulation could be able to prove the superiority of your ansatz.
What do you think about it?



Cheers,

Bobbs

Thanks, for your answer.

I have read that thread, and maybe I misunderstood the things (since I am not a native English that could easily happen), but I thought you said you didn't want to to go into many more specifics beyond those basic things you mentioned at the start.

That sounds great. I am not saying that I would understand anything from complicated calculations, but if the information is not hidden, that itself gives more confidence.

And I agree with Bobbs that you are doing a great job.

Thanks for the encouraging words, guys!

floxedG,

I think there were some parts of the conversation early in that beta thread that I felt were veering away from "big picture" algorithm concepts and into low level implementation details. I wanted to keep the topics at a higher level, but I ended up coming across as secretive or evasive. And, the internet being the internet, sometimes that kind of thing is hard to shake off.


Bobbs,

I think that is a fantastic idea! I cant believe i didnt think of myself, lol. Having some experimental data to draw from and being able to come up with some hard numbers for winrates and that kind of thing would be huge!

There's be so much you could do with something like that... benchmarking vs other algorithms, baselines for algorithm improvement, comparative advantages in different game formats... head asplode.

I've already got a huge to-do list staring back at me, but for sure i'm going to try to jam this near the top somehow...

I just downloaded and it says that I have 23 days left on my trial.

Sorry about that paul... There is a bug that affects people who had installed an earlier beta version. I'll have an update up in a few days that will reset the trial clock for beta users.

Hey guys,

I need hand history samples!

As I am elbow deep in writing HH import support, I need some sample histories for whatever site you want supported. And being that I am in the USA, I am not easily able to generate my own.

I already have PokerStars, Full Tilt, and Everleaf but other than that, I need it!

So if you've got them, I would greatly appreciate a PM or an e-mail (support@sngsovler.com) with a couple HHs from your favorite site.

Thanks!

The Hero's range as shown in the Table view is the sum of all the hands that are +EV vs the ranges of the opponents. This is often slightly different than the calculated equilibrium range for the same position. This is because the equilibrium ranges (which are then used as the default opponent ranges) are restricted to a linear handranking to help speed up the calculations.

Bump!

How is it going?

One big idea, maybe mentioned before...

I don't know how much information Sng solver needs to remember when hand is solved (cause i see that everything we do afterwards is fast so probably already computed) but if it isn't too much, idea is to make some kind of a buffer that remembers everything about hand. For example we solve and analyze hand "A", then we solve hand "B" and then we want to go back to hand "A" because we missed something. Now solver doesn't solve the hand cause solution is already in buffer and it's instantly shown.

If that's possible next step would be to make a button that allows us to solve multiple hands, and with that you get rid of most annoying thing in sng solver which is waiting, it's always between 5 sec to 1 minute and this time is too short to do anything else, so you can only watch progress bar slowly turning green.

The key for hands would be stack sizes, blind level and payout structure.
If all that matches a hand in buffer, copy information from buffer, else solve hand...

When entering "stacks", is the amount that I enter for the big and little blinds before or after the blinds have been posted?

ginandbread,

I think thats a pretty reasonable idea. As I've been working HH import stuff, I've been thinking along similar lines. It might look something like sngwiz's "tournament view"... but in addition to displaying the hands in a HH file, you'd be able to select a number of hands and do a "batch solve". The imported HHs, the data for each "solve" could then be saved off in to another file and reviewed later without having to re-solve.

I agree... re-solving a hand that maybe you just did a little while before is annoying... especially if it was a lengthy solve. I feel your pain... I'm not just writing SnG Solver, I'm a user too!


yukoncpa,

The stacks values should be *before* any posting of blinds or antes.

I cant believe that no one has noticed/mentioned that the "big blind only" button had disappeared in the last version...

At any rate, don't panic! It will be back in the next update.

Just downloaded the free trial and for whatever reason I can't fill in stack sizes. Also do I fill in stack sizes before or after posting the blinds?

Are you able to take a screenshot of what you're seeing? If so, could you attach it to a PM or e-mail to support@sngsolver.com

The stack values should be before any blinds or antes are posted.