Quote:
Originally Posted by branch0095
You're absolutely right. I normally fold turn against known nits, but I had never seen this guy before in my life. I admit that I gave him a turn bet that I shouldn't have, but I was feeling a bit sickly given that this was a Saturday night against a complete unknown. I also concur on not automatically ruling out KK, as there are plenty in my room that will never 3! KK. I discounted KK some because I had seen him 3! once when I first sat down (flawed logic, I know), so with what very little history I had with him I assumed that even though he was tight he was 3! his big PF hands. I've seen people check the absolute nuts either because they mis-read their hand or played it off as "i knew he wouldn't call a bet", but had never seen someone so terrified of quads in a HU pot that they wouldn't bet river with tops full.
It's semi-logical to suppose he 3-bets KK if you see him 3-bet over a relatively small sample. The other alternative is he got AA. Or he's one of those players who limps KK+ and only 3-bets JJ - QQ and AK because they're too hard to play post-flop. It's always hard to make good predictions based on such limited information, but we can use Bayesian inference in spots like this.
I don't know how many hands you saw, but let's suppose you had been playing about 90 minutes and saw 50 hands, and suppose about every other hand there's a raise, so the guy had maybe 20 chances to 3-bet. You get AA once every 221 hands. We've seen him 3-bet once when he had 20 chances. From Bayes Theorem (note the | means "given"):
P(AA | 3-bet) = P(3-bet | AA) * P(AA) / P(3-bet)
We'll assume P(3-bet | AA) = .95. I do see people flat surprisingly often but I'd guess it's 3-bet at least 9/10 times, we'll say 95%. We know P(AA) = 1/221. For how likely he is to 3-bet, we just have to guess based on stereotypes and our observations. If we have literally no info we can guess based on the typical super strong raising range of 3%, or we can use our empirical sample size and guess 1/20. Let's use the empirical guess, though I think if you have a read it's probably not the best way to go. Anyway, this gives
P(AA | 3-bet) = .95*(1/221)/(1/20) = .086 or 8.6%
If we suspect he just 3-bets JJ+ AK that's P(3-bet) = .0302 and P(AA | 3-bet) = .142 or 14.2%
If we suspect he's a nit we might just give him KK+, half the QQ combos, and 1/8 the AK combos for P(3-bet) = .0128, and now P(AA | 3-bet) = .336 or 33.6%
And he might be even tighter than that. But we can see the problems with empirical estimates over small samples. If we had only seen 5 raised hands where the guy could 3-bet, now our initial estimate would be 2%!
We don't really get a decent estimate until we've had several hundred hands with the guy and even then he could be getting cards or be card dead.
Point of all this is it's not illogical to assume if he 3-bets once over a shortish sample size he probably 3-bets KK, but...we just have to remember it's far from certain, and it's also non-binary. He may 3-bet KK when he feels like it and flat call when he feels like it.
Pretty funny he thought you had quads so much he can't bet the river, given the way the hand went down. Like you don't 3-bet flop or raise turn with boat/quads and then check river? Nah I think your line looks like what it is. KQ AK or maybe AA.