2. BB vs SB, why does it prefer to 3 bet hands like K9s and K3s, K2s (74% of the time) instead of lower suited aces like A5 etc?
If it's just because it's at the bottom of the range then why not Q3s or J3s? Also what makes K9s a good candidate?
1. I can't explain it in the usual human terms, because ATo clearly has more equity vs the SB's 3-bet range than the BB's.
By inputting a close approximation of the ranges into Equilab, I see that ATo has about 48% equity vs SB's 3-bet range, but only 45% vs the BB.
All I know is that Snowie learned by effectively playing against itself, and it must have found that ATo broke even as a call vs the BB, but lost money vs the SB. It must be something to do with equity realization, along with blockers/card removal factors.
2. In BBvSB, the BB has position and can utilize a somewhat polarized strategy when 3-betting. Hands like A5s can be be flatted, because it flops well and flatting means you keep worse hands (e.g. Kx, all the 5x hands) in villain's range. Hands like K3s are less good as flats (they are more often dominated and don't flop so well), but are OK as bluffs, since they have a blocker to KK/AK and a legitimate shot at folding out better hands like KT or A2.
It's the same reason why the BB often 3-bets A3o, but always flats ATo. If you have A3o, you raise to make villain fold hands like A7. When you have AT, you don't want villain to fold hands like A7o that you dominate.
Can anyone explain the math behind calling a 4 bet on the button with 66 and 77 but folding 88 and 99 ?
Is it because CO can have TT and JJ (and 88 99 at a very low frequency) which will block some straights?
It's something to do with that, but note that Snowie doesn't actually like 3-betting 99-66 in the first place, so it probably didn't do enough "training" to work out if 77/66 truly do better when flatting vs a 4-bet than 99 does.
That said, against a simplified 4-bet range for the CO (just the hands it 4-bets greater than 50% of the time: TT+,AKs,A3s-A2s,AKo) 66 has a smidgeon more equity than 99, which no doubt is do with card removal effects or blockers. There's also the thing that having some lower pairs in your range helps with board coverage. i.e. You don't "need" to flat 99/88 if you're flatting TT and some AJ+ that hit the high/medium flops, but 77/66 give you some coverage on the rarer low boards.
To all intents and purposes though, 99-66 basically all have the same value in this spot, and Snowie probably awarded them with nearly identical - close to breakeven - EV.
In the pro version of the software, you can experiment with different stake levels, where the amount of rake alters the strategy a little bit, and that can change the EV of each combo one way or the other. (You can also see an actual number for the EV, which in spots like this always resolves to almost 0.00 for the "worst" hands that continue). In the real world, those "close" hands have results that are subject to massive amounts of variance. (i.e. your database might say flatting 99 is insanely profitable, which it would be if you're really good at flopping sets of nines in 4-bet pots).
Thanks again for your time ArtyMcFly, really appreciated.
Ohh yes, I missed that the 66-99 was not in the 3 bet range at any frequency, weird that it's in the vs 4 bet range at all.
1. range compositions are different in sb and bb
sb is linear, bb is polar
ATo does better against polar because of the lower concentration of hands that dominate it
