heres something i have question about, i was analyzing a 3 handed game, I see this frequently.

chip stacks are pry irrelevant but 480, 550, 770

I analyzed when I was BB and it said Eq pushing range for sb is 50.5%, then i switched myself into the sb and the pushing range now says 52.6%. Why does this range change? all i did was switched hero from the BB to the Sb when analyzing.

The ranges for each position in the "Equilibrium View" (which are also used to set the default opponent ranges) are approximate Nash equilibrium restricted to a linear handranking whereas the Hero's range is the unrestricted best-response for the specified opponent ranges. For any given position, they should generally be pretty close, but there will often be some small differences.

so, would it be possible to have the program label every position in table view using unrestricted best response ranges?

to make sure i am doing this right. 6 handed, it is folded to me in the SB, Equilibrium strategy tree says to push 65.6% in this spot. Table view however says to push with 56.9%, I have not changed any details, everything is Equilibrium. My question is, is nash suggesting 65.6 and your PSM is contradicting that saying to shove 56.9%??? In this scenario the BB calling percentage was identical in ES tree and table view. So, this is a spot where your PSM comes into play?

i was never using "3(complete)" - these simulation times are taking quite awhile but are they going to be the most accurate??

Launched solver and I'm being told that my trial has expired even though I purchased the program.
Any ideas what happened? I emailed a few days ago but no response
Kind regards,
Colin

Hi Colin... I've replied via e-mail.

To solve the best response for every player simultaneously is the same as solving the exact Nash equilibrium for the entire table... And is therefore impractical for the same reasons. Once unrestricted opponent ranges are fully implemented, the difference between the best response range and the calculated equilibrium ranges should be a bit closer, however.

Because of the nature of the problem we're trying to solve, "accurate" is a never really a word we can use. Being able to say something is more or less accurate means that you know the actual answer... and when it comes to chip equity, we simply cannot. (In fact, the misuse of the term "accurate" runs a close second to "Nash equilibrium" on these forums... but thats a whole other topic )

What we can say is that if we were to calculate the PSM algorithm to an infinite depth, that would give us a sort of conclusion as far as PSM is concerned. Further, what we can say about the difference between PSM depths is that the deeper you calculate, the more likely the result is closer to that result you would get if you really could calculated out to PSM-infinity.

Does that make sense?

i believe so smarty pants, so in a nutshell complete 3 is best IMO but it's not accurate lol. thanks for the response

heh... well, its not that its not accurate... its just that we cant really know, in a mathematical way, how the accuracy with respect to real-world poker.

but this is all neither here nor there... my pedantic side often gets the best of me.

Hi I was wondering if there is any significant different between your ICM calculator and icmizer. I know you have your future consideration mode but if there is no sinificant difference between the two. I'd definitely be more inclined to get this software.

Thanks for your timr

Need custom ranges!

I'm dont really keep close track of what features are in other software, so I cant really give a detailed comparison. Of course the UIs are very different and certainly you should take advantage of the free trail to check things out (thought it sounds like you probably already have).

As you mentioned, the PSM model offers a tremendous advantage over any advice based on traditional ICM models. Based on the data I've collected from experiments, its hard to understate just how big this one is.

Another big feature that is unique to SnG Solver is the "EV-RMSD" display. This graph not only plots the value of a hand against your opponents ranges but it also accounts for potential uncertainty about those ranges. Because, can you really say for certain that your opponent is calling with 33.4% and not 35.5%? Probably not. So this graph is great for being able to sort out hands that are thin decisions from those where the answer is more clear. Its a feature that has gotten a ton of positive feedback from a number of high-stakes pros.


Indeed it does! I can tell you that this feature should be in the next update after all. I know its been a very long time since the last update, and for this I apologize. But it just means that there's going to be a lot of new stuff when it finally drops.

Just downloaded the software. Is there a way to copy and paste marked hands from HEM2 into sngsolver? Whenever I try I get the error message Unable to read hand history or I have to go via the Import Hand History route, which takes a lot longer than if I could just copy and paste hands of interest. It's an Ipoker HH btw. Thanks

Just looked at my first hand, am I doing something wrong?

Hand analysis by SnG Solver v1.0.10
Blinds: 150/300 Ante: 25
Payouts: 50.0, 50.0

BTN: 2855
SB: 1496
BB: 1679

Equilibrium strategies:

Action (position) Range% Range
==================== ====== ======================================
push (BTN) 0.5% AA
call (SB) 0.5% AA
overcall (BB) 0.5% AA
call (BB) 0.5% AA
push (SB) 81.9% 22+, Ax, Kx, Qx, J2s+, J4o+, T2s+, T6o+, 92s+, 95o+, 82s+, 85o+, 72s+, 74o+, 62s+, 64o+, 52s+, 53o+, 42s+, 32s
call (BB) 56.0% 22+, Ax, Kx, Q2s+, Q4o+, J3s+, J7o+, T6s+, T8o+, 96s+, 98o, 86s+, 76s, 65s

On the table view and Card graph it says AA is a fold. Per Nash ICM calculator I should be shoving 16% from the button. I have Predictive Simulation set at 3(complete)

Hmmm... if you can load the HH via the menu, it should be able to work via copy-paste as well... could you paste the HH into this thread or in an email to support@sngsolver.com and I should be able to see whats going on.

It looks like you've found a spot where the analysis is suffering from the "horizon effect". Long story short, because you're looking ahead 3 rounds at a 3 player table, there is one position that is getting sort of "double-counted"... and because this is a satellite with a massive bubble-factor, this effect is getting amplified. For 3 handed bubble spots like this, I recommenced using a depth of 2 or 4. (4 will be available in the next update).

What is the difference in the quick simulation compared to the regular?

The 'quick' PSM modes prune out portions of the simulation that are calculated to be unlikely, and therefore, do not contribute very much to the overall results.

For example, future games following a 3-way all-in that results in a perfect 3-way tie... these would represent a very small minority of possible future states, but take just as much time to calculate as other more likely scenarios.

In practice, this pruning is usually good for about 30-60% saving in calculation time and the results are usually nearly identical to the 'complete' calculations.

That being said, there is a certain risk to making a priori assumptions about the frequencies of future events. Its sort of a it-works-really-good-until-it-doesnt kind of thing.

So, my general recommendation is that the 'quick' modes are great if your sifting through a lot of hands looking for tricky spots to study... but once you've got a hand that is really close and you're going to spend some time studying in detail, its probably worth it to do the 'complete' calcs.

I dl'ed to test this out but free trial dont work.

Can you describe the problem a little more? Was it a problem with the installer, or running the program? Any error messages?

Getting trial period expired error, im 97% sure i never tried it out before.

I've sent you a PM with instructions to reset your trial period

Hey, I was wondering what the score is with the roi% result given in the chart in the bottom right in solver. Maybe I have this wrong but it seems entirely pointless to me.
I think it is:Your stack worth if you fold (always 100%) and a representation of how much you increase or decrease your stack worth relative to the 100% if you get all in or w/e/
So if it says a play is +1% it really means your stack worth will increase 1% (to 101% of it's original worth) if you make that play. This seesm pointless as it all depends on how much your stack was worth before the hand which will be entirely different for every hand you play.
Seems like it should be more like, what % in roi terms you improve by making the play relative to 1 buy in in Prize pool equity. Ie, gaining gaining 0.17% (or something close for 6 max 65/35 payouts) of the Prize Pool Equity would be a +1% roi play. As in making this play will in the long run give you a genuine 1% roi advantage over folding.
Does this make sense, am I missing something?

Is this program still in active development? Its been almost a year since you have said unrestricted ranges are coming along soon, and there hasnt been an update in almost as long. Should we ever expect this or any other update?

Colin252,

The numbers I believe you're referring to take the form of: EV_push - EV_fold = EV_diff ( RoI% )

The point of the RoI% over the %diff of the total prize pool is that RoI% remains consistent even when the prize pool values are changed to different, but equivalent values. What this means is that if you use %prizepool, it is possible to have situations that are mathematically identical but give different results.

For example... consider two games that are down to 3 players each. Both games have exactly the same arrangement of remaining chipstacks and positions. The only difference between games is that one game has payout structure of 50-30-20 while the other is 40-24-16-10-6-2.

Because we're down to 3 handed play on both tables, it turns out that these games are now mathematically identical, despite having started with different prize schedules.

If we analyze a specific hand on each table and take the results as EVdiff%, we will actually get different results for each. Using RoI%, however, will give identical results.

And this is why SnG Solver uses RoI%.

If you still want to see the EVdiff%... just make sure that the prize structure you've entered sums to 100... and the EV_diff in on the graph will now be the same as EV_diff%

Make sense?

Picasso,

Yup... still very much in development. I am sorry its been a long time since the last public update... I hope to rectify this soon.

Still trying to buy into this, but sometimes the results leave me scratching my head.

Payouts:
1st 50%
2nd 50%

Level: 100/200/40

Stacks:
BTN (Hero) : 1000
SB: 600
BB: 1200

Predictive Simulation: 3 (complete)

This gives me a shoving range of 100%. Now look at the results if I change the amount of my stack:

1000: 100%
1001: 66%
1002: 71%
1003: 75%
1004: 67%
1040: 41%
1060: 35%
1100: 41%
1110: 42%
1111: 34%
1112: 37%
1113: 34%
1120: 31%
1140: 25%
1200: 16%

Seems like some incorrect results. In a game of tiny edges, it's hard to trust these numbers. Any explanation?

still 180men structure is missing + import from a whole game hh is missing.