Poker: 4-bet pots
Possible 4-bet range (46 hands, 3½ % of starting combinations)
This article recounts a big loss I suffered in a 4-bet pot recently (relative to the stakes we play, anyway). I compare the play to the actions suggested by GTOWizard, which offers one free analysis per day. It turns out that mistakes were definitely made, even before my BIG mistake of calling on the river!
In general, my poker group makes single raises with too many hands - we play way to loosely. Our 3-betting range is a lot tighter though, a lot closer to game theory optimal (GTO). Our 4-betting range generally consists only of the very best hands: JJ, QQ, KK, AA and AK, say. This is even tighter than GTO, which will throw in some weaker bluff combinations to balance the 4-bets that are for value.
To the hand:
Preflop 5 handed, 150 big blind (BB) effective
HJ position calls 1 BB (20c). Verdict: mistake and not GTO. The solver never ‘limps’ when first to act. It raises or folds. Why? Whilst it is tempting to limp in with hands that are too good to fold but not good enough to raise, it caps your range. Having middling hands is generally bad in poker.
CO position raises to 5BB ($1). Verdict: a 5 BB raise is a little large. The computer is only raising to 2.5 BB. Raises that are too large don’t give the players in the blinds a difficult decision, as they can just fold their weak holdings. That said, the raise size is a bit more defensible when there has been an intervening bid and stacks are deep. This player would likely raise with many more than the recommended 30% of hands, though happens to hold J♠️J❤️ this time, which is certainly in the computer’s CO raising range.
Button position (hero) 3-bets to 14BB ($2.75). Verdict: button should only raise with 12% of hands. Hero’s actual hand (A♣️Q♠️) is a 60:40 mix between raising and calling. Expected value (EV) either way is a pretty modest 0.3BB.
It folds back to the CO, who 4-bets to 26BB ($5.25). Verdict: the solver does this with a tight range of suited broadways and Ax suited, totalling 77 combinations of hands (viz. 6% of the 1326 possible starting combos). J♠️J❤️ raises 37% of the time, or else it calls. The GTO raise size is to 28 BB ($5.60) so sizing is on track.
Button calls. Verdict: the solver continues with just 44 combos of hands to a 4-bet, and A♣️Q♠️ is a mix between raising, calling or folding. All these possibilities have the same EV, by definition, and since folding is in the mix they all have an EV of zero. In other words, once you get 4-bet AQo is a marginal hand!
Flop A♦️K♣️Q❤️, Pot is 55 BB ($11)
The CO checks. Verdict: the CO has 49 combos left in his range (the flop reduced some of the 77 combos he had preflop). Despite that fact that AKQ is pretty much the worst possible flop for JJ, as all the overcards came out, CO is meant to bet very small (25% of pot) with 100% of his range. This is basically true regardless of the flop texture in a 4-bet flop. This is because the preflop aggressor has a range advantage which is is preserved in practically all flops.
Button (hero) has made two pair with his A♣️Q♠️ and bets 36BB ($7.26, 66% of pot): we are off-piste now as far as GTO is concerned, as CO was meant to bet. This is clearly a bad bet for the in position player though. Two-pair is a good hand against CO’s range, having 57% equity, but is far from the nuts. If we are guessing that CO would have bet with a stronger hand than 2-pair, then a much smaller bet would have done the job.
CO calls: J♠️J❤️ does have 4 outs — a gut-shot to a straight, but otherwise is behind almost every hand that button might have (not to mention the hand the button actually has). This is just a fold.
Turn J♣️ Pot size is 127BB ($25.52)
CO checks: he has made trips now, and may have thought this was a slow play. Unfortunately we don’t really know how the solver would have played this because we never arrive at this node, but assuming that the CO had bet small on the flop and the button had called, JJ only has 51% equity on this turn card. Hero might have AA, KK or QQ, and while there can only be 3 combos of each of these hands, the button should only have 14 combos in total in his range by this stage. He might also have the occasional 10 in his hand for the straight, and even his 2-pair hands still have equity. Given hero bet strongly on the flop, JJ is probably behind at this point.
Button: also checks. We don’t need computer confirmation that this is wise.
River: 8♣️. Pot size is 127BB ($25.52)
The CO bets 90BB ($17.97) effective. Final board is A♦️K♣️Q❤️J♣️8♣️. The river card is mostly a brick, though a few flushes come home. CO may have read weakness in button’s check back on the turn, but this is still a brave bet. The solver is checking JJ and only betting higher trips or better (balanced the value bets with bluffs, as it always does).
Button calls: looking at the pot odds, A♣️Q♠️ needs to be best 17.97/(25.52 + 17.97) = 41% of the time here, for this call to be good. AQo is behind all value bets though. While this CO is known to bluff frequently, it is quite hard to find bluffs in this spot. This is just a fold.
CO wins a pot of 307BB ($61.46)
Summary
My poker group’s bet ranges and sizing are so different to GTO that the computer recommendations are of limited value in most circumstances. However, as we get to 3-bet and 4-bet pots, ranges start to converge on GTO and the solver output becomes a bit more relevant.
If you’re the preflop aggressor in a 4-bet pot, the overwhelming GTO recommendation is to c-bet with your entire range for a small size — 25% of pot.