As it turns out Snowie has virtually no hands in its range when it calls pre (only KTs 41% of the time) so the argument is moot.
One of the drawbacks of Snowie is that when you put in a scenario with a villain playing drastically different than Snowie you'll get completely useless evaluations.
fwiw by the turn the only hand Snowie could ever have is KdTd, hardly comparable to any real world villain.

According to GTO it has to defend with right % of his range to a bet (not to be exploited) so to say "it has no hands in his range" is a nonsense.Probably it has only overcards and some % of them must call.

Sent from my SM-J510FN using Tapatalk

Bruh I just looked it up and that's how Snowie plays, idk why you're arguing.

Yeah,Minimum defence frequency, it has to defend 66% of his range that got to that spot against 0.5pot bet .It is why some moves look like nonsense but it is a decision of its entire range.

Sent from my SM-J510FN using Tapatalk

Could be an idea to check out Snowies flatting range preflop versus a cut off open.

Snowie isn't GTO

You can check that if u have PIOsolver.I think that the difference is minimal.

Sent from my SM-J510FN using Tapatalk

Your point was 'According to GTO it has to...'

If it's not GTO then it doesn't have to do what you think it should in terms of frequencies.

It is leaning to play optimally balanced game,neural network playing against ot self.It is not far from GTO...

Sent from my SM-J510FN using Tapatalk

It's not GTO, it doesn't even pretend to be. Look at the homepage on the Snowie site and you won't see the word GTO mentioned. It may be close to it in spots but that doesn't mean it has to do X because that is considered the GTO thing to do.

You r leaning towards eternal question "what's older egg or chicken?"...

Sent from my SM-J510FN using Tapatalk

not sure if the first part is true, even that the program isnt gto it is built to keep playing always closer to optimal

This seems like the most obvious answer, but there are too many examples in snowie where the computer completely disregard this logic. Here's one of them;

https://www.youtube.com/watch?v=iWTnbL_w9xU

Cate Hall is supposed to check KK on the river, but what hand can a competent villain possibly have that beats us after calling a 3bet and a cold call from early position? Snowie's range for calling preflop in Brunson's shoes is QQ+ and AKss. So we only lose to a slowplayed AA, yet snowie doesn't even value bet here.

If you don't want to watch the video, here's the hand;
25/50 $13,000 effective
Villain1 raise $150 from MP, hero 3bets $475 with KK on the button, villain2 cold calls in the bb, villain 1 calls.

Flop: 956
2 checks to us, hero bets $900, villain2 folds, villain1 (MP) calls.

Turn 2
Villain checks, hero bets $2,100, villain calls.

River 9
Villain checks. ($7,500 in the pot, with $9,350 left to play for, and snowie says check is 6bb better than the optimal bet size of allin, and still even better than sub optimal size of halfpot.)


It seems like villain having 9x here is wayyyy more of a stretch than villain having Ax or 6x in the first hand.

According to Snowie, the highest EV play with KK in this river spot is to either check or bet 0.25 pot (same EV).

People are currently playing multiway pots incorrectly. You have to be way tighter everywhere.

Put it another way- imagine you are villain and face a 3 barrel on 965r29 no draw completed in a 3 way pot (added a bb cold call, and mp flat pre too which makes both ranges stronger). What do you put villain on? Certainly KK is at the bottom of that range. If villain checkraises at any point in the hand, what do we do with KK? Usually such a hand is never a slam dunk 3 barrel.

Also bear in mind, checking flop with KK as Cait and even check folding flop with JJ as villain is completely fine in GTO (folding JJ is out of the question only because of how current games are like). The reason being Cait is not even happy to get KK in 250bb deep in a 3 way pot vs either villain 1 or 2's checkraise (3b size pre was even small, so spr is still huge), and JJ is a pure bluffcatcher trying to get to showdown oop for 3 streets with huge stack left behind.

Also snowie is assuming pseudo optimal play, but in multiway pots it's far more important to play exploitatively, so I wouldn't listen to snowie's advice in most cases.

hi

is there a PokerSnowie support here?

At the moment, when I launch a challenge, the play doesn't start : the 6 players are sitting at the table, cards have been dealt, but the first player to act does nothing.
And if I am the first player to act, when I press a button, I get an error message box from microsoft windows saying that the Pokersnowie challenge application stopped working.

As if it had frozen.

I have restarted Snowie, rebooted the PC but it's still the same.

One hour ago it was running fine.

Any idea what may cause this issue?
Can you guys launch a challenge?

Hey bro, same thing happens to me occasionally just need close out the challenge window and reopen it, sometimes change the parameter. It's almost like hitting the TV to get it to work. Gl.

Oh, also make sure you don't have pokersnowie open on another laptop or something.

Good stuff. Yeah definitely having the cold caller in the hand (even if was just on the flop) is a lot more significant for river value betting, and alternatively villains bluff-catching ranges, than I originally thought.

in the challenge mode, are the higher stakes harder to beat than the low ones?
or do they all play the same?

basicly the same, the only difference are in few spots where rake is relevant in small stks

This. The Neural Network programming used does seem to have trained snowie to have a reasonably balanced strategy when using its own bet-sizing limitations (or at least I'm not sharp enough to spot large imbalances).

An interesting question to me is whether a supervised deep learning program making EV decisions while playing against itself and improving would converge to the GTO solution if it were exposed to enough strategies. Curious to hear rationales for why this would or would not be the case.

In practice, as mrno1324 points out, PokerSnowie can come up with some pretty useless (I'll say verifiably inaccurate) range advice when playing against players with too different of a strategy than snowie, heads-up or not.

Using range advice after setting up scenario evaluations can often result in lines that snowie is taking that are not only bad, but show snowie's inexperience. I often get to spots where my opponent can supposedly profitably bet only 2-pair plus and no bluffs on the river, but somehow I am profitable to call him down with 1 pair hands. These spots are not Nash Equilibria, needless to say. Then again, I expect snowie to have nearly zero experience with the bet sizes used throughout the hand, so without the requisite experience or GTO calculation, of course he gets things wrong.

Can you give a couple of concrete examples?

So I don't save the scenario evaluations as indicated, since they do not help me much, but something funky did happen in one of these regarding a recent session of mine, and snowie is still running, so I will share. I do want to say that snowie has improved my play tremendously, and if a snowie mod is out there, you can take this as a bug report.

2-5 NL

Folds to Hero in HJ ($2275) who raises to $25
Villain in CO ($2275) reraises to $90
Everyone else folds (all other players started the hand w/ $2,000)
Hero 4-bets to $275
Villain calls

Flop ($557.50)
J82

Hero checks
Villain checks

Turn (557.50) J829

Hero checks
Villain bets $300
Hero calls

River ($1,157.50) J8292

Hero checks
Villain goes All-in for $1,700
Hero folds

Pokersnowie himself chooses this all-in sizing as his preferred bet-sizing. Snowie shows up to this river (as Villain) with 54s, 65s, and 76s in diamonds, clubs, and spades and jams all of these hands at an 89% frequency with each, checking back the other 11% of the time. That's it - snowie likes a jam and he doesn't have anything better than 7-high in his jamming range. 63% of the time after we check this river, snowie will punt it off with no value.

This scenario has Hero calling nearly 90% of his range (AA, KK) and folding the other 10% (AsKs).

This was an actual hand I played.

Maybe it is because you come up with 5 bb open that Snowie isn't trained with...What is Snowie's calling 4bet range preflop?

I thought that was pretty much the point I was making in my prior post - that snowie doesn't have experience in certain bet sizes.

At different frequencies, snowie flats the 4-bet with AA-QQ, AKs, and 54s-87s, as well as TT (he somehow folds JJ).