Hello David, thank you for bringing up a question.

Just checked frequencies and it is 3-betting 5.05% (calling 2.57%) against UTG 3X Open in 6-max format, while in 9-max the 3-bet frequency is 1.62% (calling 2.17%).

Stake was 25/50 with 100BB.

Thanks for that but I meant to ask regarding ak specifically

At 100bb deep 6-max it 3-bets 100% of the time with AK in MPvUTG. I'm not sure about 200bb+ though, or full ring.

When creating scenarios for 6max Snowie recommend for AK is 100% 3bet for NL20(high rake) and NL20000(low rake environment).However the difference is that it recommended UTG to open pot bet for UTG 100bb deep at NL20 and min.bet(2bb) for 200bb deep at the same level.For NL20000 it recommended UTG to open 1/2 pot either 100bb or 200bb deep.
No ante ofc.

For 9max it goes:
-NL20(100bb) UTG opens 1/2pot...UTG1 AKo is 87% 3bet pot,13%call
-NL20(200bb)UTG opens min.bet...UTG1 AKo 3bet pot 100%
-NL20000 (100bb) UTG opens min.bet...UTG1 AKo 3bet 1/2 pot(Snowie recommends)12%,Call 88%...If u go for pot size 3bet it is 63% raising and 37% calling.
-NL20000(200bb)-UTG opens 1/2pot...Snowie recommend 1/4pot 3bet where AKo is 100%(but AA is 100%call and KK is 50/50)...If we choose 3bet pot size then AKo is 100%call.
In training mode for 9max this r the frequencies recommended.

Serious answer:

Subscribe to Snowie for a year.

Janda can help you learn how to use the program, if you like.

* 7 day free trial but time flies, as they say.

Gross. Rake is way too high at the lowest stakes.

Is there a way to tweak snowies cash game settings to better emulate live ranges ( eg more flatting, wider ranges)? I saw the tip about adding a .25 ante which has helped. Is there anything else that can be done such as increasing or decreasing stack sizes?


Sent from my iPhone using Tapatalk

Probably adding ante will get to wider ranges but what u mean when u r saying "live ranges"?Playing live cash games at Casino?

yes, or even microstakes online...people are calling much wider and three betting much narrower I find than snowie which throws off the post-flop game tree....

Calling wider is a mistake because of the high rake.A lot of hands r -EV according to bot(PIO will come with almost same results if u solve with 5% rake).Snowie learn that and adjusted ranges.If u compare NL20 and NL20000 in training mode Snowie plays different preflop (wider at NL20000 ofc).Also if u make a mistake preflop according to Snowie's recommendation u can be misled on later streets recommendations.So all u have to do when using Snowie is to learn patterns.What the bot is doing and think about why is doing it.
So,calling wider is mistake in high rake environment.So when thinking about adjusting u should think about going tighter (and believe me Snowie is playing tight pre).

I've forgotten to say,for live play u can check for Hunter Cichy's book,Advanced concepts...It is based on Snowie's play and how ti adjust it to brick and mortar games.

I know its a mistake..im not saying I plan to emulate that strategy...I run sims in snowie and then run my errors through pio...the problem is that snowie calls so narrow that pio wont give me very useful results because it is not solving for what human being villains actually do...

That's because u will never know their ranges 100%.If u run a sim u have to put a range in PIO.If u do that u will have to also put Snowie's bet size recommendations.And than u will get some results that u will have to interpret. Snowie is doing moves according to it's own data.If bot is making money by calling in some spots based on his similar spots played before(they've said it played trillions of hands against itself)it will call with some frequency (0 to 100).The problem is that the opposition is not a perfect and u r getting some "pseudo strategy"to play against it(all those nonsense about exploitative vs GTO play bla bla...).
I would say just pick Snowie level and play against it.If it is 6 max micro NL20 imagine u r getting 5 LLinusLLovers as opponents.U will learn by practice.

How much faith should one place in Snowie's recommendations when they run counter to received poker wisdom? For example, Snowie seems to have a donk betting range on some flops (example hand below) but received wisdom is that you should generally check to the preflop aggressor.

500NL, hero has KdQd in the big blind. CO raises to $11, BTN raises to $40.5, hero calls, CO folds.

Flop is 8d6dTc with a pot size of $94.5. Hero checks with the intention of check-raising. Snowie says this is an error and hero should donk bet 1/4 pot; it thinks the EV of checking is $9.82 and that of the small donk bet is $11.83.

Probably it favours his range ,1/4 pot is Snowie's recommendation for betting in that spot and for that size it is choosing a blocking bet.Villain will check behind a lot of his over pairs and over cards on that board.(ofc. GTO villain playing like Snowie).Snowie needs some Ace high hands to fold to a 0.25pot bet.

Is it "received poker wisdom" to cold-call a 3-bet with KQs in the BB at 500NL? I have no idea whether it's correct to have a donking strat in that spot, because I mostly cold 4-bet or fold pre. (Very occasionally I flat JJ or TT, and I can see good reasons for donking with an overpair or top set on that T86 board, as villain may well check back his best overpairs and AK/Ax. If you wanted to donk your top set, it's easy enough to balance with a FD that you misclick called with pre).

In that spot BB calling range is KQs,AQo(4%),JJ(26%),TT(67%).So it makes sense to donk with a set of tens and protect that with 1 combo of flush draw.I am not sure if Snowie is betting his whole range or JJ and AQ r checking behind.

I have checked JJ is also betting,AQo and other KQs combos are checking behind.Snowie will also bet 0.5 pot but check his entire range for bigger bets.0.25 pot is optimal.

KdQd is betting 95% and check 5%of the time so it is mix.It will not show it as error if checked according to my scenario of 100 bb deep.

Yeah, I know the call is a bit iffy. FWIW, snowie reckoned the EV of both calling and raising was -0.06 BB so I guess it's kind of a Doug Polk spot in practice - you can call sometimes, raise sometimes, and fold sometimes. That said, the preflop action isn't really the point - snowie's donked into me a few times as well, on various streets, where my action is much more clear-cut. For example:

Hero has AcKd in the CO. Snowie raises to 2BB UTG with AsQs, hero raises to 6.5 BB, Snowie calls.
Flop is 2sTcAd. Snowie checks, hero bets 8.5 BB, Snowie calls.
Turn is Ks, Snowie leads for around 1/4 pot.

Obviously, in that case it's picked up a ton of extra equity from the turn card, but still, it's donking into the previous street aggressor.

Everything sbowbot is doing is for to realize equity.And for that u need to make it at least to the river.

I'm still finding it hard to make any kind of coherent strategy based on Snowie's recommendations when it comes to 1/4 pot bets. Take this hand, for example:

6max 500NL, hero has KdQd in the SB. Snowie opens for $11 in the CO, hero pots it, snowie calls.

Flop is 2hJhAc, action goes check, check.

Turn is Jd, hero bets half pot (snowie says betting here is correct but 1/4 pot is better than 1/2), snowie calls.

River is Kh, giving a final board of 2hJhAcJdKh. Obviously, at this point hero's pair of kings is just a bluffcatcher. Hero checks, Snowie bets 1/4 pot, hero calls. Snowie can of course have tons of boats, two pairs, trips, straights, flushes, and even just paired aces that beat me, but at the same time, I'm getting 6:1 to call with a bluffcatcher, which seems like it's a mandatory call if your opponent has even *one* bluff combo in his range. As it happens, snowie was making a play with 77, so it clearly does have some bluffs in this spot. Nevertheless, it classified the call as a blunder and says that the correct move is to fold.

The short story is that although donking is mostly considered a fishy play, it is part of the overall GTO strategy at some frequency. Good players will donk exploitatively in some sports (e.g. for value when flopping a set multiway). It's just that donking is rarely much more profitable than check(-rais)ing in the same spot, so you can have a very similar EV or winrate even if you never donk-bet. So in a sense, the "perceived wisdom" that donking isn't a great strategy is kind of correct. (It's also the case that players that donkbet a lot tend to be losing players.)
Finding spots where donking is good might add a little to your winrate, but it's not massively important to develop donking ranges imo. In addition, building donking strats necessarily increases complexity, since that's another branch of the game tree you have to learn to balance. When using solvers, many players don't give the OOP player the possibility of donking, as they rarely do it in real life, and it just makes the "solution" harder to understand.

I've played a few thousand hands against Snowie now, and while it's an interesting opponent, some of its advice just seems flatly wrong. Take this hand as an example:

500NL 6max, 100 bb deep. Hero has K♥Q♣ in MP, Snowie has A♣Q♦ in the CO.

Action folds to hero, hero raises to 2.25 bb, Snowie calls.

Flop is T♠5♣5♥ with 6 bb in the pot. Hero bets half pot, Snowie calls.

The turn is the K♦ with 12 bb in the pot. Hero bets pot, Snowie calls.

River is the 9♥ with 36 bb in the pot. Hero checks, Snowie jams for 2.3x pot (82.75 bb), Hero calls. Snowie says that hero's play up to this point was perfect, but that the final call was a blunder with a massively negative EV of -45.4 BB.

The thing is, according to Snowie, its value range when it does this should be TT, 99, and QJs (fair enough, I guess), and its bluff range is AQ and AJs. Taking its frequencies into account, its 2.3x-pot overbet shove range consists of 5.7 value combos and 13 bluff combos. So if hero ranges Snowie correctly and calls with all his made hands, he wins the pot 70% of the time. But given the pot odds, hero only needs to win 41% of the time to break even on the river call. So how can the call be -EV, how can it make sense to have so many bluffs here, and just generally, what the ****, Snowie?