Ok yes I get you now, thank you for your patience. So in both hands the EV of folding n calling is 0 as we have a bluff catcher in this spot and snowie has constructed a range of nuts/air and bets appropriately, in the second case the rake is a higher proportion resulting in a 0 ev situation becoming -ev

Yes, rake reduces your EV SO SOME SPOTS THAT ARE NORMALLY +EV become -Ev thus we need to fold.

Check out Hu strategy in 50nl vs 1000nl, one game has effectively no rake the other has very high rake. In this spot pre flop strategy of snowie is extremely different resulting in a very different strategy overall (much tighter obv)

I dont understand this software.

Hero raise co AQ
btn call
sb call (shorty)

flop: 7c4d2s

sb checks

Hero's bet was considered as a wrong move.

Snowie's advice is to check 99% of the time with my range wtf.
He suggest to bet only with 88,78 if I would bet half pot.
With a potbet he suggest to bet 77 (and KTcc, K9cc,T8cc,98cc 3% of the time) and nothing else.

Why is this?


edit: I modified sb's stack in scenario to full buyin and snowie still suggest to check 98% of the time.

it is mistake to bet that flop w AQ since most of CO opening ranges hit that flop poorly, no more than 20% top pair or better. if we are betting AQ, AK KQ...every time,that means we are betting more than 90% of our range and we hit only 20%. BTN and SB could 3bet , C/R us w blanks and print money.
shortstacker makes things even worse.

I never said I bet 100% of the time my AQs. It all about snowie's range advice.
You check this flop 100% of the time even with op?

snowie is based on GTO model, and betting 90%+ of his range on that flop is not optimal.

well my table is mostly 5 to 9 degenerate gamblers and me, so usually i just bet when i have it and check when i don't

You dont answering the questions. reread them again.

i answered everything you need to know in first reply.
just think about it carefuly, you will figure it out eventually.

Snowie doesn't play perfectly in multiway pots, because no one knows how to play perfectly in multiway pots, but Snowie plays them a lot better than most humans.
The general idea is that when you're OOP multiway, you often won't maximise your EV by betting, because you're not closing the action, and two or more opponents can "exploit" the information you're giving them, so it becomes very hard to make money by betting at a high frequency. It's often better to check, partly to induce worse hands to bet, and to give you the opportunity to gain free info and/or close the action. It's more complex than that (and not well understood by anyone), but that's kind of the idea. The other point was made by Pamik. On that particular flop, you don't have many very strong hands in your range, so you shouldn't be betting at a high frequency.

FWIW, if you're used to playing 2NL against horribad/unbalanced villains, Snowie's lines will look very strange, and following them slavishly against bad players could cost you money, because the 2NL villains will have completely different continuance ranges to the "optimal" ranges Snowie has trained against.
One obvious difference is that Snowie c-bets less often than a typical microstakes player but check-raises much more often. Check-raising for value is clearly more profitable if an opponent will fold to a c-bet, but will bluff when checked to. At 2NL, passive villains will routinely call c-bets, and not bluff when checked to, so it would be better to just bet for value when you're trying to build a pot.

Hi,

There is a difference between your 2 scenarios: in the 2nd one, you have not acted yet.

Since you have not acted, snowie hasn't calculated anything. Try calling the shove and see what it says then.

I have played this software HU for a large sample and I have two questions that maybe you guys can answer, if it is indeed GTO or near GTO why does it min raise pf when u play it at nosebleeds but 3xs every other stake? Am I wrong in thinking that GTO is independent of stakes? 2nd question is it safe to assume it can't adjust to players? I only say this as I notice when it overbets on any street b4 the river snowie has air or weak draw that folds to a raise close to 100% of the time. I give them credit for NL AI it plays very well but I also agree its not GTO, also their EV #s has anyone tested them comparatively to CRev? As in some spots they seem hard to believe.

It takes rake into account.

this thread got some people feeling some type of way

when playing vs 5 snowies in a 6m game, is it possible to see their stats while playing?

It takes rake into account. I'm not sure if they've published which rake structure they use, or how many levels it has, but Snowie's ranges are tighter at 100NL in comparison to 1000NL, just like in the real world.
It doesn't adjust. You may have found a way to exploit it, but check the scenarios/ranges and see if your feeling about its ranges is correct. In my experience, it also overbets with hands it wants to play for stacks with.
There's no HUD as such, but post-challenge you can look at a wide range of stats for how the opponents played on average. Since all 5 Snowies play the same strategy, you can get a decent idea of "Snowie stats" from a fairly small sample size. (You can also do this with imported hands, if you want to know, for example, how often the player pool defends the BB on average, or folds to river bets).

ok thank you

stats are not needed. snowies ranges are known, as such its stats are known as well.

you can look up any spot and see how it got there and what it will do from there.

Are there any spreadsheet of error-rates vs snowie compared to the limit the player plays?

By this, I presume you're asking "What kind of error rate would I need to beat $XNL?" or something like that.

As far as I know, no such thing exists, let alone been published, but (unless I heard it elsewhere) I think that in the Thinking Poker podcast featuring Johannes Levermann, it was said that the guys in Montenegro that learned how to play poker solely from Snowie had error rates of about 8 and were beating 200NLz. Forgive me if my memory is at fault.

If you just want to see winrates vs error rates in the Snowie challenge, it's on the blog. ("Expert players", who'd I'd guess can beat 100NL fairly easily, lost about 5bb/100 vs Snowie's old AI, but the bulk of challengers were merely "advanced".)

I regularly get an error rate of 5 or less , sometimes ET level at 2.X and I struggle to beat 50nl

1. sb raise, hero 3bets to 9bb, sb 4bets to 20bb. snowie defend with 90% of his range. He calls with almost all of his 3betbluff crap hands like 74s 85s Q6s T8o...etc. why?

2.hero raise co KQ, btn sb calls. flop Q96r sb checks. snowie recommended a check.
hero bets, btn minraise sb folds. snowie recommends fold 100% of the time. why?

3. QsQc hero raise btn to 2bb. sb bb call.s 6bb pot. flop Ks4d4h check. turn Kh . sb leads 2bb bb calls, snowie would fold. weird.

that is because 50nl is unbeatable. check how much you win pre rake and post rake.

the effective rake at these levels must be around 80%. at that point it will be very hard to beat the games.

for me personally the rake is pretty much 100%. when fll tilt came back online i played 60k hands at 50nl and won 100. But pre rake i was up 1500. In order for me to win 100 the opposite player to me must have lost 3k (since he paid for my 1500 and in doing so would again lose 1500).

If it takes 3k in losses to win 100 you will have a hard time winning anything unless you are really in the top 1% of players.

You need to play 2/5 or so in order for the rake to become reasonable.

On the flip side i have had good succes in tournaments. Rake there for me is around 20-30% which is the same when i play live cash.

So when playing online for low stakes one should stick to tournaments and play cash in casinos.

If you buy one, lesser, Snowie subscription now, if you decide could you later 'upgrade' to a higher one by paying just the difference or do you have to pay the whole higher subscription/ fee ?

Similar story here. As my error rate wet down, so did my profits!

I think a lot of it is variance, but I think it's also an indication that against weaker players, you should deviate somewhat from Snowie ranges/frequencies. (One of my biggest "mistakes" is making Snowie-style hero-calls when my opponents are aren't bluffing anywhere near as often as Snowie would.)
It's hard to beat the rake in the micros unless you really exploit the weak when you get the chance.

We don't know why snowie does anything and snowie cannot tell you. snowie can only tell us what it does.

i assume however that this is a call because the 4bet is a min raise. so we are getting close to 3:1. as such snowie calls with 1/2 its range and splits the rest into folds and raises.

i assume we need to call so much because we are getting quite the right price and should have 30% EV with that range.

However snowie does fold in this spot 1/4 of its range too. And snowie would always 4bet pot.