So that is 24 value combos to 16 bluff combos that cover low flop boards...60% to 40%...What is your range on the flop (assuming you r playing perfectly)?Your checking range

I'm sorry - I'm confused. What is the relevance of how I play my hands in that spot to this snowie thread? I was adding to the conversation by pointing out some limitations that snowie has, but now I feel like I'm at the start of a coaching session. Or do you mean snowie's range? For that, I can't continue to share pokersnowie's strategies in a public forum.

The point is that Snowie knows your checking range (that's why I said "if you play perfectly balanced gto style strategy as like he is playing against itself)and is playing his range accordingly. He bets his no showdown value hands as you said 89%(for fold equty) or so and surrenders 11%(for balance).And his checking range(on the flop) probably is covered with some showdown value hands.Since you can't shere it,don't blame it.Just solve for why.

...You can check that 0.53 pot bet on a turn after you show weakness twice does Snowie is betting with his entire range?Probably it is because in theory you can bet any combo and show a profit .But GTO wise probably some showdown value hands will be checked for protection and bet on river(or check called).I like Snowie also because it teaches you how to play.With PIO, you have to figuire it on your own.

Please expound on what you mean by "surrenders 11% for balance." If bluffing all of these combos 100% of the time on the river is wrong, and bluffing them 89% is correct, then that would be because betting 100% would have us at too high a bluff-to-value bet ratio on this river, lest we get exploited by getting over-called by our opponent.

Regarding the check oop range, etc - I've shared that already. After hero snowie checks oop on flop/check-calls turn, he has AsKs 10% of the time and AA/KK 90%. Villain snowie does not have any value bets in this spot where he is betting, so that is by definition unbalanced. Showing a river overbet jam with 67s as being profitable against himself is untrue - because against himself, he gets called 90% of the time.

I'm not blaming snowie for anything, I use the product a lot, and I'm glad you get value from it as well. You do see the limitation though, right?

You should enter the scenario I listed into Pokersnowie if you are interested - I've closed and re-opened the program again.

I am basic subscriber,I don't have range analysis...but I have CREV/GTO+.It is easy to check if Snowie is right or wrong if I have both ranges at turn spot...but never mind.

I put it this scenario in GTO+...I doubt if Sbowbot is betting the turn only with low suited connectors(you said that flats a 4 bet with AA-QQ,AKs etc.)AA r 100% bet and also entire range with some mix % , only 87s is not betting the turn.
Ranges that I put were:
Hero AA-QQ,AKs,AKo,A5s-A2s,87s-54s
Villain AA-QQ,TT,87s-54s

I used to think that if Snowie was given more training it would converge on a GTO solution, but now I'm not so sure. I think that the best that an artificial neural network of Snowie's type can do is come up with a strategy that does "reasonably well" against the strategies (and near-relatives of those strategies) that it trains against. One of the problems is that introducing more bet-sizes increases the size of the decision tree - and the EV results database - which tends towards infinity. A neural net like Snowie could learn how to play a game that only used pot-sized bets, but if the half-pot sizing is introduced as an option, all the previous learning is rendered null and void, and a new "solution" needs to be computed that allows both sizes to be used at the optimal frequencies. From there, if you add another bet-size option, the solution needs to be re-computed once more, making it even more complicated (it's essentially a whole new game with another layer of options). And then you need to solve the river play for spots where the turn was played incorrectly at some (arbitrary?) frequency.
For the bot to be able to play unexploitably, it would have to train against bots/agents/humans that can do anything that doesn't break the rules of poker. (e.g. Use any bet-size on any street). Until it does that (solve the entire game), there will always be some holes in its strategy that allow major or minor exploits.
For now, I think the makers admit that it still has "problems" with both very small and very large bets. (Some posters have mentioned ITT that it is apparently also vulnerable to 5-bet bluffs pre-flop).
It's also been mentioned several times that spots come up where one player makes an "error" on an early street, and that causes the river ranges/strategy to be nonsensical. e.g. Hero's river betting range "should be" 98% air, but Villain still folds 100%.

My feeling is that Snowie does pretty well against "normal" play, but it is less 'accurate' when faced with weird/uncommon lines, ranges and bet-sizes. I guess it's possible that any computation time spent fixing those leaks/exploits (e.g. learning how to correctly deal with 5x opens) might actually diminish its ability to play accurately in other more common/typical spots, but I really don't know. For my purposes, it's a useful learning/practice tool, but it's demonstrably not GTO, because its own river betting/calling ranges seldom form 'equilibrium strategy pairs' and there is often a very clear imbalance, indicating an exploit exists.

It doesn't have experience in certain bet sizes preflop because they're not optimal sizings. The point of snowie isn't to be able to plug in live spots and get a solution. You can't tweak snowie's preflop ranges (because it has already "decided" that the ranges it uses are optimal). Plugging in a 5bb open from a live spot is going to give you a drastically different range than what the live player actually uses, so you'll never get a real solution. It's not meant to solve exploitative spots, and opening 5bb in a live game is an exploitative play right off the bat.

I used to agree with every word here. However, now I have doubts about the complexity of alternate bet sizes.

I am really just getting started with multiple street no limit poker math, but here is my take on it thus far.

Please correct me and help me learn if any of this is way off track!

If stacks are infinite, there should be an optimal bet/raise size on every street. It is somewhere around the size of the pot. In toy poker games it is exactly the size of the pot. This is because bets of other sizes unnecessarily move the indifference point boundaries within range and lead to less range economy. Smaller or larger bet sizes also lead to smaller game trees, not larger game trees.

Lets look at the overly simplified toy game where only one player can bet and this bet can only be called or folded. Real poker has tons of options, and more options creates more indifference points in the ranges, but the overall range portion that survives each round of betting should be similar to the toy game.

A bet that is full pot brings along the most combos per dollar from both ranges. This is because all bets, regardless of size, must reside inside the bettors balanced range, and non-optimal smaller or larger decrease the overall payoff to the bettor. Smaller bets sacrifice too many bluff combos and larger bets sacrifice too many value combos.

Furthermore, when a bet can be raised, the non-optimal size of the initial bet can be exploited, since the initial bet must be balanced. Balanced bets must come from portions of range that are bounded by indifference points. When a bet size is not optimal it comes from a skewed portion of range and can be raised (or folded) for pure profit by an opponent's hand that is outside this skewed range.

Optimal bet sizing reduces the chance of a bet being raised, such that a raise must in turn also be optimal and can not be pure profit.

Lets look at a bet of twice the pot on the river.

To be balanced, such a bet will come from two different locations in a bettors surviving range. There will be a value to bluff ratio of 40% bluffs and 60% value. The caller knows this and knows that such a bet comes from the top 60 percent of a bettors value range, and the bottom 40 percent of the bettors bluff range.

Now play the hand backwards with 2x pot bets on each street.

The bettor has 56 percent bluff and 44 percent value on the turn.
The bettor has 78.4 percent bluff and only 21.6 percent value on the flop.

The bettor has run out of value!

Now lets look at a half pot bet, again only considering calling each street. Now the bettor has a ratio of 75 percent value and 25 percent bluffs on the river.

Now play the hand backwards with 1/2 pot bets each street.

The bettor has 31.25 percent bluff and 68.75 percent value on the turn.
The bettor has 39 percent bluff and 61 percent value on the flop.

The bettor has too few bluffs!

Now lets look at a pot sized bet. The bettor has 66 percent value and 33 percent bluff on the river. Now play the hand backwards with pot sized bets each street.

The bettor has 44.44 percent bluff and 55.56 percent value on the turn.
The bettor has 59.25 percent bluff and 40.75 percent value on the flop.

Now the bettor will get the most use out of both ranges. This ensures that the bettor will profit from over half the bluff range but still not run out of value by the river.

Now, if all these scenarios are the result of preflop play that has the options of bets and raises, then we can see how a player who arrives at a flop could really run out of bluffs or value by the river, if a healthy portion of bluffs or value was already reduced by preflop betting.

It should be pointed out that this toy game definitely is rigged. I believe, not for sure, but I am guessing that if a bet is required on the river, the caller has the advantage. If the bettor can instead checkback the river as well as bet, then the game favors the bettor.

If I recall what the humans said about Libratus, it was that Libratus could overbet the pot freely, but would then find enough bluffs by bluffing bottom value combos (hands with showdown value). This would be required by the above toy game.

What I have understanded from Janda's perspective is that those river betting combos (nevermind 60/40 or 75/25 or whatever) let say 100 combos are finite (they bet all streets).Those combos that bet turn and before that flop are mostly bluffs and some of them r shifts to checking range because they have gained some showdown equty or lost all of it.So basically u can't say we run out of bluffs or value hands on the flop but on the river.That is what I think.

Fooled around with Snowie for the first time in a few years. I play 200nl 6 max online fwiw.

My impression is that its 3+ handed game is quite weak, and a little too click-backy in certain spots. I played it for ~160 hands 3 handed before it crashed, and its lines were whack.

Snowie's heads up game is much better, but still has leaks. It plays certain spots - specifically IP SRPs quite well. However I was less impressed with its OOP play, especially in 3-bet pots. Being limited to a 3x 3bet sizing is also... kinda limiting. It made a move I thought was really terrible (donked 20% OOP in a SRP on TTT), and also donked 20% on 622.

I let it give me an analysis after HU, and disagreed with around half of the spots it said I made a mistake in. I also strongly disagreed with some of its preflop suggestions. (Labeled not 4 betting K6o as an error)

I'd say 5nl 6max is tougher than snowie's ring game, and maybe 50/100nl HU. Hard to make a comparison though.

Beat it over like 15k hands HU at least and post a screenshot if you want to be taken seriously. Random forum guy not being impressed doesn't mean much.

I played snowiest quite a lot and it took me several hours to get in about 2k hands. You did 15k?

No but I'm not making big claims about Snowie's leaks. There's a good chance the plays he considers bad are just over his head so it would add a lot of credence to his analysis if he played a relatively decent sample and didn't get owned.
Not trying to attack him, js. I'd actually be really curious to see how someone somewhat competent (not me lul) does over a sample.

Snowie is a very good tool. I think I was up maybe 2bb/100 or something marginal over the sample I had, but it should be noted that my goal wasn't to beat it.....

I misinterpreted your response.

Input any sort of reasonable HU srp ranges into pio on a TTT flop. There will never ever be a donk freq, even at 20% size, I'm certain of this. And it's not close to being a good play either.

Calling K6o as an error to not 4 bet is also a joke. Similar with calling a QJo flat HU a mistake.

It made some other very questionable suggestions postflop, but i'm not 100% confident in them so I refrained from posting. But given it's silly preflop suggestions, I don't think I trust it too much.

Now i'm not saying it's a completely useless program. A few of its suggestions were actually corrections on my play, and it can be a good tool for beginners to improve their play. But i'm just saying that people shouldn't take it all too seriously. I'm certain that good humans can beat it, since it deviates from Pio suggestions significantly in many spots.

Also, can you link me to where jungle says he uses Snowie to study recently? Not doubting you, but it does seem surprising to me.

https://twitter.com/junglemandan/sta...69703984775168

Non zero chance he was levelling I guess but it'd be a weird level, even for jungle.
Which sizes and stakes did you use to get K6o 4b 100%? Can't seem to recreate that suggestion from Snowie.

103bb deep, I open 2.5bb Snowie 3bets to 7.5. I get that it's not perfect preflop, but I also have my suspicions about postflop play, including play in larger flop and turn spots.

Yes fair it will do this at lowest stakes I presume as a consequence of 4betting more lightly for value (because it dislikes seeing the flop with rake) and hence needing more bluffs.
fwiw at highest stakes it only 4bets 56% and folds the rest. Its preflop ranges (and opening sizes) change rather drastically based on which stakes are selected due to rake.
Maybe if you play it at 100/200 you find it's closer to pio.

Yes, play the highest stake one.

edit: I used it for 6max primarily fwiw.

I'm barely competent (and used to be absolutely terrible when I first tried Snowie a few years ago), but I've beaten Snowie every year since it was released, while dabbling in all formats (freezouts, full ring/HU/6-max cash at different limits and stacksizes).
I mostly specialize at 100bb 6-max 100NL simulations when playing challenges. In those, I've won at 2bb/100 in 26,659 hands (7bb/100 in the most recent 5k), but I wouldn't call that a representative sample, as a 20 buy-in downswing could start tomorrow. #Variance



If the challenge games were raked, I'd be a losing player, but then so would the bots. Snowie is the toughest opponent I face on a daily basis, but then I mostly play against microstakes amoebas that can't spell their own names. Snowie does loads of stuff that infuriates and annoys me, but I've learned a ton from studying with it.

33k snowie hands?

Wtf.

@Arty Are you trying to emulate Snowie as closely as possible to get lower error rates or do you make a concerted effort to exploit its possible leaks? Do you ever go "I know Snowie is underbluffing here so I should fold but it's going to consider folding a mistake in the review so I'm going to call."

legit lol