Does it help? Trying to figure out ways to study other than 2+2 and CREV. Has anyone tried Snowie? I'm on their 10 day trial and it's telling me to do pretty weird stuff. Like bet 1/4 pot a lot of the time, open A9o in the HJ and fold open fold 44 in MP. Also simulated the scenario I posted here.

http://forumserver.twoplustwo.com/17...llain-1670576/

And it told me to check back flop with AK wtf. Obviously not gonna try to apply Snowie directly to LLSNL but is there any merit in learning what the "optimal" play is?

Studying GTO is an excellent exercise but keep in mind, if you have to play GTO, you're essentially just paying the house to see cards because no one will be able to make a profit of anyone else.

I've tried snowie, and think it's not terribly useful for LLSNL. Having said that, assuming mid to deep stacks, I'm opening much wider than A9o from LP. On some tables, open folding low PP in EP and sometimes MP is ok, but for the most part, I do try to set-mine live, be it open raising or limping.

Stuff to do other than posting here:

*Read some books

*Watch some pokerbank

*Run equilab on all AI/river call decisions along other spots where it's good to know equity.

*Use flopzilla, and speaking of pokerbank, I just saw a splitsuit video on a program called "flop falcon." It's similar to flopzilla, but a bit different. I'd recommend looking into flopzilla, as I do use that, and flop falcon seemed useful as well.

I'd say its useless. The name of the game in LLSNL is exploitative play. The day I have to fine-tune my bluff 4betting frequencies in LLSNL is the day I am no longer playing.

I got a full year subscription of Snowie by mistake and they refused to refund me. I rarely use it nowadays.

It's good for 2/5nl zoom maybe but def not live. Nobody bets 1/4 pot live for balance and shizz.

Generally agree with the above. Play is not applicable to LLSNL. That said, I think it is useful to have an understanding of what GTO play generally looks like so that you understand better what departures you're making to exploit LLSNL V's and so that you can adjust when they adjust.

Also, when you move up in stakes, you're going to start running into some players that actually know what they're doing. It's useful to have some GTOish knowledge to fall back on while figuring out what mistakes they're making.

But I think that requires more reading and theory work than playing Snowie. I found the preflop play to be pretty easy to figure out, but it's much harder to understand why Snowie is making the plays it is postflop. And I've spent some time working to understand GTO poker.

If they had an engine with a library of "typical" LLSNL players, it'd be absolutely wonderful. But that would be an entirely different development effort.

As an aside, GTO poker will make money at LLSNL tables, just not as much money as exploitive play. If everyone at the table is playing GTO, no one is making any money, but if N-1 players are playing close to GTO and one is playing perfect GTO, that one will make some money from the rest. Just not likely very much.

Lastly, GTO heads-up is probably not the same as GTO FR. So the paragraph above assume correct GTO strategy for the actual game conditions (meaning number of players and blind structure/antes, not V's playing styles).

I don't know whether Snowie learned to play HU, 6-max, or FR (or all of them).

FWIW I've often wondered how an adaptable AI would handle live game dynamics. (High vpip, low fold to cbet, short and huge stacks, etc)

My understanding is that snowie plays with iterations of itself.

Yeah, the developers created neural net, fed in the rules of poker (but no poker playing knowledge at all) and then had Snowie play against itself until it converged on a strategy.

Then they created an opponent that specifically looked for ways to exploit Snowie's play and had Snowie play that. This led to the I exploit you/you exploit my exploit of you/etc... loop. Once that had converged, Snowie should have something resembling a GTO strategy -- for the game conditions under which it was developed and subject to standard disclaimers about neural nets. I would guess that this was for HU, but I could easily be wrong. The dev team may well have done this multiple times for different types of games or chosen 6-max or FR.

More complete description is on their website.

Janda's recent book uses snowie and pio solver both. Typically he uses it to test spots he's already drawn a conclusion on. Sort of a sanity check with a grain of salt.

One of the most noticeable things in the results he chose to use in the book were the use of both small sizing and overbets in post flop spots and the use of mixed strat.

I've been thinking about checking snowie out myself.


Sent from my iPhone using Tapatalk

I've watched some of the training videos for Snowie on youtube. It appears that it does not take into account a rake. So I don't see how it could possibly be any use for games that rake instead of paying time.

Went back & checked Little's video. .50/1.00. 1st hand $6.50 in pot & no rake.
2nd hand Little has 88. $14 in pot & no rake.

Can't see what use Snowie can be for a raked pot game. However, my understanding of GTO is limited to the 65 GTO 389 3 deuces, 4-speed that my father bought and then sold 2 months before I got my license.

Prob better to look itt http://forumserver.twoplustwo.com/15...2/index31.html

But snowie does account for rake in its EV calculations. I don't think the rake structure is published. But changing from very high stakes to very low will impact snowie's strat due to rake becoming relaxant.

I've only used the 10 day trial briefly but it seems the bigger disconnect from llsnl is that snowie has no experience playing in an environment where people will routinely cold call 7x bb raises with crap. So if you put in a hh from here for example the ranges that snowie villains show up with as flat calls are very narrow.

It's like the old Sklansky post on "invisible antes". In a lot of ways from a theoretical approach a 1/2 no limit game should be mostly a dead and unbeatable game due to rake. Its only the anticipation that other players are so terrible that they can be counted on to enter a pot as a cold caller that has been raised to 7x with low quality hands which incentivized you to bring it in with such a large raise relative to the existing 1.5bb pot.

Since villain's can be counted on to make such theoretically horrible decisions llsnl strat is built around the idea that players will play incorrectly from a theoretical perspective. Snowie otoh I think spends its life playing what it believes to be "correct" vs other players playing "correctly".

This is just from my very limited time with it but basically I don't think snowie can imagine such fundamentally bad play as we encounter. I suspect this is because AI hasn't learned about the illogical actions that can be motivated by ego and addiction.




Sent from my iPhone using Tapatalk

CAmmAndo, How does it account for different rake structures? Some rake 10% [max $5] but $1 for BBJ @ $10 & again @ $20. Others rake 10% up to $6 and $1 for BBJ @ $20
In Jacksonville, FL - it's highway robbery.

No idea.

I read an FAQ on the snowie site that mentioned rake. Janda also showed in his new book some comparisons of high limit vs low limit and how snowie concluded certain hands became -ev at low stakes due to rake.

I've not seen the actual rake structure published but I would presume it to be similar to stars or some "typical" online rake.

Also it's possible that if you are importing actual online hhs that snowie is deriving the rake from the hh file. But I'm just speculating.


Sent from my iPhone using Tapatalk

I personally find snowie the most useful tool I have ever used. I constantly am a contrarian in these forums, and yes I am new to them.

Yet this is probably because I use snowie and tend to analyze and think like snowie a bit now. I personally find using Snowie's gto style as more conservative and allows hero to get value from his middling hands. How many times do you all go and fire with a pair or second pair heck maybe even third pair and villain snap folds on the river. Many of the times hero feels good since his hand was "vulnerable"

Snowie has helped me realize the equity in middle strength hands more often and analyze deeper during live games into villains hand range in the moment.

Personally, I use the scenario feature to analyze many of the situations that are posted in this forum, since live players with actual jobs only get so many hands per week. So I come on here read the forum. Formulate my opinion and comment, then run the hand the poster wrote up in Snowie to see what Snowie thinks. Then i read all of the other posters comments as well.

I feel that it allows my own thinking to converge with Snowie (complete analytical) and other human opinions to formulate the best strat of study. Active study and analyzing.

For new live low stakes players, Snowie is a way to formulate a basis. If every player played 75-85% like snowie in live low stakes games, and 15-25% exploitative play due to live reads, you would produce a much better result.

Snowie and GTO are the foundation of poker nowadays, exploitation is EXTREMELY important live, but it must be used situationally and correctly when it matters. Playing like Snowie keeps hero in good spots more often as our basis, then when conditions are right we exploit and reap the synergistic effect of both!

But hey F me right?

Not sure why your so snippy but maybe you would be willing to run a hand through Snowie. My trial expired and I'm not sure I'm not making mistakes. I'll comment and then post a link to the actual hand thread.


I had similar thoughts. As I mentioned previously I don't think in snowie's world the "invisible antes" Sklansky posted about exist nearly as they do in llsnl.

I ran Mike Stars QQ pahwm in snowie. Hero had QQ UTG and opened 7.5x and the button flatted. Iirc snowie suggested raising pot or something as a UTG open. I entered the hand as played with the $15 open and btn flat. Snowie listed a range for Villain of KK and nothing else. So I assume if snowie was playing an iteration of itself and faced a 7.5x open it's 3! Or folding everything except KK. ??? (Admitted noob on 10 day frail so possible error).

Now as I understand it snowie's strat as hero doesn't consider ranges so it's recommendation for hero have nothing to do with the fact that snowie itself would only have KK there.

AP Mike bet 20 into 30 with top set on QT9. Villain raised to 60. This was the first significant point of discussion.

Some of us wanted to 3! Smallish with suggestions in the 120-150 range. Some wanted to check.

When I entered the scenario as played into Snowie. Snowie suggested a min raise to 100.

I'd be interested to see if V3ttz3ao gets the same results as I did or can expand on the results a bit particularly the cold call range of villain.

The hand history thread is here:


PAHWM : QQ 225 BBs deep.

https://r.tapatalk.com/shareLink?sha...1&share_type=t


Sent from my iPhone using Tapatalk

Most posts that I comment in I run through Snowie, generally I only comment on spots I actually find interesting. Many of the "fold pre" etc ones I dont spend much time on.

Sorry I'm very tired i was up at 5am and i usually don't have to wake that early and came home to post I shouldn't have been snippy lol. Also looking back holy crap my post is ****ty. I meant to say that Snowie helps me check and reach showdown more often with middling hands live, and eureka I win against villains air. Many low stakes players bet some of these middling hands and fold out villains air and only get calls when they are toast. That was a nice leak it plugged.

ok next post i will analyze.

For that post in particular it's a bit complicated since Snowie is unlikely to play like that, but here goes once i made it a 2/4 game to accomodate that 15 dollar pfr from UTG... snowie folds almost everything to a 7.5x raise from utg except AA KK and flats AK, once in a while snowie advocates being a dick with 65s, A3o, or A5s and reraising for balance.

AP Mike Starr or whatever, should do this OTF and broken by each sizing

Before the raise by villain OTF for this had the range of villain was JJ, AQ, AK... Once villain puts in the raise to 60 OTF here are the re-raise and frequency breakdowns..........

Raise to 0.33pot 97% of the time and call 3% (*Snowies optimal bet size is this probably awful in llnl)
Raise to 0.5 pot 30% of the time and call 70% (Probably very good for these games so it would be like re raise to 145ish)
Never raise full pot
Raise to 2xpot 7% of time time and call 97% of the time

Those are the re-raise breakdown options for hero. At these games clicking back 1/2 pot would be best to 140-150 range some times and checking others for balance if you want it, which I advocate balance just not here personally.

Anyway as for what villains range is according to Snowie at this point with that villain raise to 60 against our flop c-bet of 20... Snowie literally pegs him as JJ 100%

I assume the reason here is there is no flush draw, KJ or any other set either raises or calls. So the only hand with reason to want fold equity would be JJ, it is a good pair on a scary board with an OESD for equity.

And the cold call range for MP villain preflop is pretty narrow to hero's UTG 15 open according to snowie... (again snowie is going to be nitty compared to live games and doesn't factor in villain is tilting and ****... Snowie always assumes villains play just like Snowie)

QQ........probability is 8%
JJ..........probability is 18.5%
AKo.......Probability is 62.3%
AQs.......Probability is 11.2%

Curious if you adjusted stacks to provide a similar spr?

But yeah interesting both the KK range I ended up with and the JJ you got with adjusted blinds have the blockers to the nuts.

Unfortunately I downloaded the trial at a time where I just didn't have enough time to get into it adequately.

Sent from my iPhone using Tapatalk

Based upon what I am reading, Poker Snowie is not for LLSNL. Last night, UTG & +1 limped & I raised to $17 after 4 folds with JsTs. Everyone folds & the 2 limpers call. $45 in the pot after the rake.
Flop: 9s6s3c ck,ck I bet $25 they both call. $119 in pot.
Turn: 9s6s3d 5c ck, ck, ck
River: 9s 6s 3d 5c 2c Bet $50, raise $50, I fold, call.
Bettor had 6d5d for 2pr OTT [6th street chat: He was expecting me to bet again OTT so he could raise] raiser OTR had 6c4c for str8 OTR [6th street chat: He would have folded to the 2pr OTR if he 3! big, but thought his str8 was worth a min-raise for value. But he 'knew' V was the kinda' guy to play 74s]

This is not uncommon. Of course, Snowie would probably just call pre with JTs & ck behind OTF. That's my guess. I don't know what an AI would do. However, I have a sneaky suspicion that it would go broke in a 1/2 game.

Snowie support answered my e-mailed asking about rake:
They used the rake structure found on the major internet sites - 4.5%

Yes I edit the stacks to give an SPR to be reasonably the same.

lol look you need to read ALL of my post, I GET IT, Snowie is NOT the answer to how to play llsnl. What it is, however, is a way to practice playing high level thinking opponents. All Snowie opponents play at the same high stakes level of play. Adjusting the blinds on Snowie does not mean you cranked the level up or down. It doesn't matter, Snowie can only play a perfectly balance optimal strategy. What Snowie does is force hero to think actively about Snowie's range, and Snowie's perception of our range, and think how we can go to third level thinking to make appropriate decisions. For the most part, llsnl players barely touch level 2 thinking or incorrectly apply level 2 thinking, and most **** players are stuck at level one thinking.

Snowie helps hero think actively and deeply about a hand. Just firing bluffs won't work against Snowie, because Snowie doesn't tilt or have feelings. Snowie will teach you how and when to barrel, how and when to go for thin value, using blockers correctly (i hate reading these forums when that concept is always misapplied), and providing concrete structure to our ranges for opening/3!/4! from each and every position. Post flop improvements to my game were understanding showdown value, increased desire to reach showdown with middle strength instead of constantly betting to fold out air and only get value calls from villain (hero is good more often than he realizes so just see showdown stop betting with middle strength hands).

Being aggressive is more controlled against Snowie. This helped me massively in llsnl. I use controlled aggression most of the time when playing live now.

Caveat is that this helps me against GOOD players. The concepts Snowie makes us aware of more deeply don't apply to fish, bad regs, stations, etc. We play exploitative against the bad players and play a balanced strat against the Good players.

If hero plays every opponent hyper aggressive always betting/raising then hero is the fish, not villain.

"Just because your opponents aren't thinking opponents that utilize ranges, doesn't mean that you shouldn't still be thinking about theirs and yours."

Doug Polk.

Basically, don't play down to their level. You won't progress.