starting gun,

Yes, it can be tricky to interpret situations like this when comparing the results from different simulation depths. The problem here is something called the "horizon effect" where, when trying to decide between two paths, the best choice might keep changing if you are able to keep looking just a *little bit* farther ahead.

I can tell you that the results from a deeper simulation is always *most likely* to be the best result. I'm sorry if this sounds a little vague, but this is why the extended depth modes for PSM are not 'on' by default.

The good news is that for the last few weeks I have been working on something to bring some clarity to spots like this. I hope to be able to unveil it soon...

Indeed, you have found a bug! I will have a fix for this in the next update. Thanks for pointing it out!

The EV-RMSD bands represent the effects of uncertainty about our opponents ranges. Even though we might put an opponents range at 10%, how often are we able to really say that its 100% accurate? Almost never. The reality is that that opponent could often just as easily be on 8%.. or 12%.. or ?... and how might the EV of our hand be affected by this uncertainty? This is what is what the EV-RMSD band visualizes.

The green band shows the bounds of our all-in EV if our opponents ranges might vary by as much as +/- 20%. Now, this is 20% is *relative* to the original range... so if the opponents nominal range was 10%, then the uncertainty covered by the green band would be 8-12%. If the opponents range was 30%, then the green band would cover 27-33%, etc...

The yellow band behaves the same way but represents a +/- 40% variation in opponents ranges.

The calculations for these bands includes all the players ranges in a hand that have acted or have yet to act. The potential variation of all these ranges are aggregated into the results.

The portion of the band that hangs below the Hero's green all-in line typically represents opponent ranges that are wider than nominal, and vice versa. (I say "typically" because there are some extreme situations where this is not always true)


Well... I dont really like to specifically speak about other peoples software... so I'll just say this...

I think with things like PSM and EV-RMSD, that SnG Solver is clearly the leader when it comes to innovation and "outside-the-box" thinking. And I can tell you that I've got a lot more on the drawing board...

Also, note that *any* program that creates strategies based on any of the standard ICM algorithms (Harville, Weitzman, Roberts, etc.. ) is at proven disadvantage to the PSM in SnG Solver: http://blog.sngsolver.com/2012/04/ad...-verified.html

True, I dont have the ability to customize opponent ranges... yet. This is one of the most requested features and I *am* working on this. (Its been taking me a bit longer than I had hoped... there is another new feature that I am adding and combining it with custom opponent ranges is very complicated).

I would like to add the ability to handle limpers at some point... but at this point I cannot say when this might happen. There are a number of other things with a higher priority that I need to finish first.

Thank you for your thoughtful response, looking forward to the update

Where's raise call EV on priority list?
I would really like to see it in Sng Solver cause of PSM, it's obvious that in other software raise call EV, especially when we have low stack ,will be flawed..
I know there could be a problem with finding equilibrium for raise call, and for that you also need equilibrium for 3bet and so on, but for start we could manually set villians 3bet range and other future rounds would be calculated as now.

wow thanks a lot for this awesome answers,
I got 10%+ of my marked hands vs people limping! would be awesome if you can do that!

In witch spots the use of PSM are the most diferents from ICM classic,
For psm you use nash push/call for the future hands? can i change these ranges ?
thx!

What is the faster way to past marked hands in HEM1 to sng solver?
(i use right click on the hand/view hand/ highlight + ctrlC , / control v in sngsolver)

I would like to set a "default payout structure" !

thx!

Yes, the problem with raise-call type spots is that it really complicates the possibility of coming up with a equilibrium solution. This is actually not an issue for the "future rounds" math of the PSM algorithm, but it does make finding credible default opponent ranges really difficult.

Probably, the first version of something like this would have to put the onus of setting the "caller" ranges on the user.

As far as where this is on the priority list... I dont want to make any promises about this right now, but I do consider spots like this (and other combinations of limp/call/min-raise/etc situations) to be "low hanging fruit" when it comes to making SnG Solver more useful. I have a few important things to finish first... but then I would like to be looking these kind of things next.

PSM tends to diverge from ICM more anytime the stacks get shorter and as the "bubble effect" increases.

Yes, the future rounds are simulated as if the players are playing according to an approximate Nash equilibrium push/fold strategy. You cannot manually edit these future ranges... to do so is just not practical. For all but the simplest scenarios, there can be thousands of relevant ranges to adjust... not a very appealing prospect.


SnG Solver tries to figure out the correct payout structure when copy/pasting a hand history... if its not working correctly, you can post/e-mail the offending HH and I should be able to get it sorted.

hello, i just emailed you also,
am trying the demo, carbon/merge player with HEM2 database.
can't seem to import a tourney and or hand.
Can you explain how i can do this with solver.
I can send you a tourney file if you need.
thanks

toprange exclusion slider?

no limp no need imo!

is jason on vacation? i have tried to email him and no response on here as well?

Hey guys, I'll have some replies soon... been out of action for a bit and I've got an inbox thats bursting at the seams.

Yeah, if you're having a problem with importing a HH, either send it to support@sngsolver.com or even just post it in this thread and I should be able to get it working.

Once I get unrestricted opponent ranges in, I can add such a slider to the range pickers.

ok, just emailed you a hh from Hem2.
rambo911

Can you include some variables in your PSM calculation

Would be awesome if we can change the "nash ranges" in the futurs hands,
Just choosing the style of vilains,

Pusher/folder Pusher/caller , Walker/Folder , Walker/Caller

And changing litle bit the ranges

As a matter of fact... yes, I have a plan to add a kind of opponent modelling... no ETA though.

Just bought the programm to support you

That addition would be huge, if you need a good Turbosng/MTT-Player for input for the ranges write me a PM.

Hey Jason, I've been reading up on how sng wiz makes errors by not calculating ties properly. I understand Solver does calculate this in the correct way.
The cost of the these errors seems remarkably close to the difference between solver & wiz in a lot of calling spots. This being the case it seems that PSM has little effect at all on the outcomes I am seeing. This doesn't really make sense to me, as I thought PSM was what was causing the slightly different results between the two programs

Depending on the exact situation (stack depth, position, etc...) the differences between ICM and PSM can range anywhere from negligible to enormous.... it really just depends.

If you can post the details of a particular spot, maybe I can help make sense of things.

Hey man.

Can you try and get support for full tilt in teh next update. Heres a hh. lmk if you need anything else. thx

Full Tilt Poker Game #31321760060: $60 Sit & Go (Super Turbo) (243507577), Table 1 - 20/40 - No Limit Hold'em - 11:50:46 PT - 2012/11/06 [14:50:46 ET - 2012/11/06]
Seat 2: HardcoreParkour (535)
Seat 3: marazzzm (445)
Seat 5: pearnip (535)
Seat 6: JSH1991 (285)
pearnip posts the small blind of 20
JSH1991 posts the big blind of 40
The button is in seat #3
*** HOLE CARDS ***
Dealt to JSH1991 [Ac 6d]
HardcoreParkour folds
marazzzm raises to 445, and is all in
pearnip folds
JSH1991 calls 245, and is all in
marazzzm shows [As 4d]
JSH1991 shows [Ac 6d]
Uncalled bet of 160 returned to marazzzm
*** FLOP *** [Jc 4c Kd] (Total Pot: 590, 2 Players, 1 All-In)
*** TURN *** [Jc 4c Kd] [Qh] (Total Pot: 590, 2 Players, 1 All-In)
*** RIVER *** [Jc 4c Kd Qh] [4h] (Total Pot: 590, 2 Players, 1 All-In)
marazzzm shows three of a kind, Fours
JSH1991 shows a pair of Fours
marazzzm wins the pot (590) with three of a kind, Fours
*** SUMMARY ***
Total pot 590 | Rake 0
Board: [Jc 4c Kd Qh 4h]
Seat 2: HardcoreParkour didn't bet (folded)
Seat 3: marazzzm (button) showed [As 4d] and won (590) with three of a kind, Fours
Seat 5: pearnip (small blind) folded before the Flop
Seat 6: JSH1991 (big blind) showed [Ac 6d] and lost with a pair of Fours

Im not sure if it has been asked for or is already in development, but adding in a minimum edge would be cool.
Also, is there an eta on the custom ranges? Seems like it's been in development for a long time.

Thanks for the HH. I've had a quite a few people send me FTP HHs now... so, for sure, I'll be adding support ASAP.


I'm not likely to be adding any kind of "edge" control. The way this gets implemented by other programs, its not based on any kind of solid mathematical theory or rigor. As such, I feel this would fly in the face of what SnG Solver is all about.

I think the two most common reasons to need an "edge" correction is 1) to adjust for position, and 2) to adjust for opponent skill.

The effects of position are already accounted for by the PSM algorithm, so there already shouldnt be a need to add more adjustment for this.

The opponent skill factor can be a legitimate concern. So if you think you should need to be more or less risk averse in a particular spot because of some opponent tendencies, then, well... just make a mental adjustment to the recommended range yourself.

I dont meant to sound flip about this but I really believe that adding some arbitrary magic number to the end of the equity equations would just be a cynical ploy... an attempt to prop up "feel" with the air of mathematical authority. It would just be an insult to your intelligence... and nobody wants that.

Yeah, its been a while... longer than I had hoped. The holdup is that there are some other big additions that I've been working on that have 1) been time consuming in their own right and 2) are additionally complicated by the change to unrestricted opponent ranges. Its been a challenge to get everything to fit together and be able to release things in a way that wont require everyone to have to re-download all the hundreds of megabytes of data every time there's an update.

I appreciate your guys patience with this... hopefully I will have made it worth the wait.

On a slightly related note... while my plan had been for the next release to be a "major" update that would be version 1.1... since its taken so long, and I've now accumulated a number of small fixes for things, I'll likely be release another minor update first (1.0.11).... probably some time next week.