Preflop solving with PokerSnowie. Weird results needs explanation.

Posted by

Posted by posted in High Stakes

Preflop solving with PokerSnowie. Weird results needs explanation.

Hi guys

Im currently reviewing my preflop game with extra focus on when hero gets 3bet OOP. I have for a period of time adopted Andres Artentinos default ranges from his course but I dont think they work very well due to too simple handselection(dont playing mixed strategies) and flatting with a very high frequency compared to 4betting.

My plan was to first study PokerSnowie and then Monker. And when possible use PioSolver.

When I dig into Pokersnowie I encounter some pretty weird results.

For example when we have narrow range spots, Snowie prefers to have a very narrow value range(AA) and then bluffs. A LOT of bluffs. For example UTG vs SB, UTG almost exclusively have AA in its 4bet value range. And then it 4bets and folds to a jam 82%(!!!). And explanation for this could be, that the EV of SB 5 bet jamming range is very low when called since it playing against AA all the time. But simple math shows us that 5bet jamming with 12% equity is very profitable, so this explanation doesnt really hold.

Thoughts?

Its obv 100bb deep and the we choose the sizings Snowie prefers - 2.25bb --> 8.25bb -- > 24bb

Now we get to the part where it gets weird....

Again we focus on strat versus a 3bet but this time in position. CO open -- > SB 3bet --> CO 4bet. Here CO value 4bets a broader range of AA-KK, AKs and a lot of AKo. SB then jams AA, all AK, KK with some freaquency and all(!!) Axs, some medium pocket pairs and AQs-AJs very rarely calling the 4bet. In itself an unusual strategy. Did I say that CO value 4bets AKo - it doesnt. It 4bet folds AKo against an extremely large 5bet bluffing range... This is just insane and doesnt make any sense. Btw CO 4bet folds 75% in this spot.

We see the same thing BTN versus SB where SB 4bet folds AKo, JJ and TT... Doesnt make any sense.

And now we get to the part where it gets really weird...

Against a 4bet Snowie prefers to make a 1/4 pot raise(24bb --> 41bb) and not to shove. This cannot change that much you might say. Wrong. It somehow changes everything. When Snowie 5bet 1/4 pot the strat versus the 4bet(in the BTN vs SB example) looks like this: 68% fold, 13% call and 18% 1/4pot raise. When we change snowies strat options from 1/4 raise to all in the strat vs 4bet looks like this: 33% fold, 5% call and 62% all in....

Am I missing something? Is there a good explantion for this? Or should I just delete Pokersnowie from my computer?

9 Comments

Loading 9 Comments...

Kalupso 6 years, 5 months ago

Snowie is weak for 4bet and 5bet stuff but quite good for opening and BB VPIP against opens.

I'd recommend getting Monker or Simple Preflop Holdem for preflop solving. Results are better than PIO edge from my experience and requires about 1/10th of the RAM to get similar quality. They can also solve multiway preflop spots but that requires a lot of RAM when you allow 3+ players to go postflop with high SPR.

Pape_Sux 6 years, 5 months ago

Whats your opion on Snowies Call vs 3b ranges? They seem pretty tight. I was also wondering from the board coverage perspective, since Snowie folds basically all SC´s OOP vs 3b (except BB 3b vs SB open).

Kalupso 6 years, 5 months ago

Snowie is pretty close to the better simulations I've seen for VPIP against a 3bet when raise sizes and rake are similar. Range construction and hand selection is slightly off compared to the stronger models. Tight seems to be right OOP vs 3bet in games with lots of rake and people mostly 3betting top hands.

Demondoink 6 years, 5 months ago

Snowie is very useful for some things, but not for others. this would be one of the examples where it is pretty useless, imo. definitely don't delete it though, as Kalupso mentioned a couple of aspects that it is very useful when working on.

you should also realise that poker software isn't gonna play for you, or tell you everything you need to know. if you have played a million hands at one limit you will KNOW that certain hands do well as 3 bet 5 bet jams, certain hands do well as 4 bet bluffs, what our stacking off ranges should be (roughly) from each position etc. if you take the Snowie, or indeed any other pre-flop ranges and stick to them like gospel you are gonna probably get crushed at your limit as you will either be stacking off far too wide or far too tight, probably on the far too tight side, having looked at Snowie briefly.

and you can always check your database, what hands do well and to what line. as certain hands may be 'winning' by Snowie's estimation as 3 bets, but that may not be the case for us humans. we can not play nearly as well as Snowie (or as weird) unfortunately haha.

just play a bunch, tweak different parts of your game on a daily basis using different software, rather than look for a one size fits all solution to pre or post-flop play.

Pape_Sux 6 years, 5 months ago

Really good post mate. But I still think low stakes players should have some rough guidelines (maybe not Snowie), especially for preflop play. You need to play huge volume so that you can analyze your DB for some specific spots like calling small PP´s or SC´s in 3bet pots in different positions.

Demondoink 6 years, 5 months ago

cheers man :))
yeah for sure, Snowie is very good at using for RFI ranges, for calling ranges and for knowing what to flat vs 3 bets. for example you will see some sort of cut-off between, let's say, pocket 5s and pocket 6s, where one hand is winning marginally as a flat vs a 3 bet and the other is slightly losing. so it's good to know where to draw the line and makes it easier to know in game what hands you should do what with.

as I said, for the exact 3 bet/4 bet/5 bet ranges I think that it is way off, at least in terms of what would be winning at the limits in and around what I play. for example 4 bet folding AKo, 5 bet clicking AA or whatever. don't think I am capable of doing those plays haha :P

ohmyrage 6 years, 3 months ago

I think snowie takes removal a TON into its accounting.
-While our AA is a value range, our bluffs all all Ax based, which completely halves villains AA combos. The ranges are so narrow that having exclusively AA against a super narrow 3bet range to begin with makes more sense
if you take a look at how narrow snowie 3bets UTG from SB, you'd see that 4betting KK isn't exactly that appealing, by flatting it it helps the rest of the calling range.
-We also see that by blocking villains AA, we get to bluff alot and really put hands like JJ-KK in a spot where they either jam into a huge equity problem running into AA, OR they flat and have to deal with Ax equity

Bingo 123 6 years, 3 months ago

Its just a result of how Snowie works. Not saying Its bad, its quite interesting how it works. And I think it has developed decent open ranges and defending ranges. 4Bet/5Bet pre I is just a ton less common than PFR that's probably why its still weak in that regard. It just plays against another dumb robot and try to learn what works and what does not.

To begin with, PokerSnowie played completely at random. After each
hand played, successful betting lines were reinforced and unsuccessful
moves learned from and reduced. For example, a call with a low high
card hand on the river is mostly a losing situation, so PokerSnowie
would call less and less with such hands, whereas trips win most of
the time, so calling in this situation was reinforced.

Many people are surprised that computers can learn something
'psychological' like a bluff. In fact, this is one of the first things
PokerSnowie learned. If a bluff is often successful in certain
situations, bluffing is reinforced and PokerSnowie bluffs more often.

No expert knowledge There is no expert knowledge built into
PokerSnowie's strategy. This proved to be a disadvantage at the start
of training: strong hands like a full house were randomly played and
even folded. Playing quads or straight flushes was especially
difficult to learn because it's very rare to hold those hands. Here, a
big difference can be seen between human learning and PokerSnowie's
learning: a human would know that quads is a very strong hand that
wins almost every time. This leads to the obvious conclusion that
quads should never be folded. PokerSnowie, however, sees a hand that
it doesn't know and with which it has very little experience. Only
slowly did it adapt its strategy with these hands in the right
direction. Of course it's easy for the neural network to learn how to
play these kinds of hands, but it takes time.

On the other hand, giving PokerSnowie the complete freedom to learn
whatever it thinks is best has extraordinary advantages. If an expert
were to stipulate parts of the strategy, those parts could not be
improved by PokerSnowie, even if the expert's strategy proved to be
wrong. The beauty of the expert-free approach is that PokerSnowie can
become a much better poker player than the programmers and also better
than humans in general!

Be the first to add a comment

Runitonce.com uses cookies to give you the best experience. Learn more about our Cookie Policy