I am flying to Korea this week to play Triton Jeju. If you’d like to become a paid subscriber, you can do so for one month at the cheapest rate I’ve ever offered, $1 here. If you’d like to read about how the POTD schedule will be changing when I’m in Korea, you can do so here. If you’d like to become a Premium Subscriber and see what you get for $300/year. I am rerunning some of those speciality posts this week. I am starting with one of the most discussed poker hands of the year, Andrew Robl folding the second nut flush to Justin Gavri. If you already read my analysis the first time around and don’t want read it again, scroll to the bottom for the premium content.
In case you missed the most recent episode of High Stakes Poker Andrew Robl folded the second nuts to Justin Gavri. The board was not KQJT2, this was not PLO, it was a two card queen high flush vs a two card king high flush on an unpaired board with the ace of trump on the flop and no straight flushes possible. If you have yet to see this hand, The video with the time stamp is embedded below and I will transcribe the hand history later in this post.
Moments after the fold, a very impressed Nick Schulman said “that’s the best fold I’ve ever seen in my life. That’s the best fold of all time. I don’t give a fuck what anybody has to say, that’s the greatest fold of all time in a cash game. That’s unprecedented.” Others were less impressed, Fedor Holz called the fold “kind of trivial” and Len Ashby wrote that “I’ve made probably a hundred bigger folds in my life”. I of course, rushed to my computer to broadcast that Robl’s river size was suboptimal, we all play our parts. So what was this fold: The greatest fold of all time or a trivial fold that anyone could find so long as their mind was sufficiently primed? Let’s dig through some solver and exploitative analysis and see if we can make sense of it.
(500//1k/1k) (SB/BB/BBA). 6 Handed.
Brown folds LJ, Eric Wasserson (430k) straddles to 2k in the HJ, Gavri (325k) makes it 5k in the CO with Kh9h, Madden (176k) folds on the button, Senor Tilt (166k) calls in the SB with QcJs, Robl calls in the BB with Qh7h, Wasserson folds.
Flop (18k) AhTc2h: Tilt checks, Robl checks, Gavri bets 8k, Tilt calls, Robl raises to 35k, Gavri calls, Tilt folds.
Turn (98k) 9s: Robl checks, Gavri checks
River (98k) 6h: Robl bets 125k, Gavri shoves for 290k total, Robl folds.
What I Think (No Cheating)
The optimal strategy in these games do not involve perfectly executing solver strategies, however when I look at any poker hand I do like thinking about how the hand would play if it were played by three experts. My guess is that, Tilt and Robl’s preflop calls are too loose, Gavri’s c-bet size is fine, but probably should be a little bigger or a little smaller. Tilt should fold to the c-bet and Robl’s check-raise is fine. Once we get to the turn either player can take an aggressive action, betting a flush draw after check-raising the flop is always a fine play from Robl. Gavri’s turn decision is closer, but he should be able to get Robl to fold some Ax or Tx, while potentially getting Robl to stick it in with a worse draw like Qh8h.
Once we get to the river, I think Robl has a polarized range he rarely checks the turn with two pair or better and he rarely checkraises the flop with one pair. His river range should have a lot of flushes, but also a lot of missed straight draws. When your range has a lot of air and flushes betting with flushes and bluffs makes sense. However I don’t care for Robl’s size because he has a good enough hand that he wants to play for stacks, but he picks a size that makes it very hard to do that. His size is so big that he makes it less likely Gavri raises worse for value or bluff raises himself. Robl should pick a size that’s small enough that Gavri will always raise the river with 5h4h or just try to beat Gavri to the punch and go all-in himself. I don’t think the solver would ever overbet shove the river here in Robl’s spot because Gavri should have a flush a lot, but Gavri doesn’t know that. If I think Gavri would never fold a flush or bluff raise over a block, I’d shove myself. If I think he might fold a flush vs a jam, I’d block to try and induce a raise. I am not tempted to check-shove because I think Gavri’s average bet size facing a check will be too small and I want to let him call with top pair.
Once Robl bets his size and get’s quickly shoved on, it’s to put it mildly a pretty sick spot. The first question you need to ask is, does Gavri ever value shove worse? If he does, it makes folding rather difficult. Gavri has at most seven combos of the nuts and if he has three combos of worse flushes, you need to call. The second question is, does Gavri ever bluff? It might seem impossible, but AK,KQ,KJ with the nut blocker all seem like totally reasonable bluffing combos. It’s a daring fold, one that I would not have made myself, but I can understand the logic behind it, especially if you had a live or timing read on Gavri.
What The Solver Thinks
Preflop is too loose from Tilt and Robl. Gavri’s c-bet is fine, but he’s supposed to only c-bet 30% of the time with range, and he’s almost certinly c-betting more often than that. Tilt’s flop call is too loose, Robl’s check-raise is fine. Ranges are already supposed to be tight enough that Robl is supposed to start folding hands like 6h5h and As9s to a bet and a call on the flop. When Robl check-raises the flop Gavri is supposed to fold 60% of his top pair combos including certain AK combos such as AcKs. Gavri almost always continues with KQ and KJ with the Kh.
Robl pure bets his hand on the turn for 75% pot and it’s enough to get bare nut flush draws to start folding, JhTh and AK to start folding. Robl begins the turn with a set or two pair just over 40% of the time and once he bets he has one almost 60% of the time. Gavri can make some very big folds here if Robl bets, but he did not. Robl checked and Gavri made a good turn check behind. Gavri doesn’t get Robl to fold a better hand enough to make betting the turn appealing, which brings us to the river. The solver either picks a small block size (30% pot in my sim) or goes all-in for 3x pot. The all-in range is exclusively nut flushes and nut blocker hands like KhQ,KhJ,KhT. A 3x pot shove makes some non nut flushes indifferent to calling. Some flushes like 5h4h pure call, but some queen and jack high flushes as well as AK,A9,A8 with a heart mix calling the river and are neutral EV. Shoving a queen high flush is greedy from Robl and might lose EV if Gavri misses hero calls with one pair or hero folds too often with a flush. The block size is very appealing to me because Gavri is supposed to fold top pair vs a block quite a bit and Gavri is supposed to just call a block with hands as strong as QhTh and Jh4h. If you think Gavri will always raise a flush and might bluff raise KhQ, KhJ or AcKh over a block, I find it hard to believe that blocking isn’t the best exploitative play, in addition to being the solver play. If you think Gavri might save you some money by raising smaller than all-in with a king high flush or if you trust yourself to know when you can go for extra value and three bet shoving over his river raise, even more reason to block. I think too much analysis of this hand has revolved around a simple framework of “you have the second nuts vs an unsophisticated player you should bet big”, but the solver shows us that betting a large size is almost certainly not the best way to get max value versus an unsophisticated player who might always raise a small bet with a flush.
If you don’t give Robl a river bet size larger than 125k, Gavri responds by jamming the nuts and some bluffs with KhQx, KhJx and AxKh. However there is a fundamental problem with the river solution, Robl only picks this size with the Kh in his hand and Gavri only shoves with the Kh in his hand. So this river betting sequence should happen… never. When I did some messing around to make Robl bet this size with hands in between nuts and bluffs and Gavri shoves, he has the nuts 78% of the time. Robl is still supposed to call and calling is making himself 13k, blocking two of Gavri’s flush combos and none of his bluffs is still too valuable.
What Sam Thinks
I gave the technical breakdown above, that is (I hope) interesting to read about it, but in this section we get down to brass tacks. The question we’re asking here is not: does the solver call a river shove with second nuts, it’s should Andrew Robl call vs Justin Gavri? Normally, when I analyze a hand I do it chronologically because the decisions made on early streets create the ranges that shape river strategies, but the only question here is does Justin ever bluff and will he ever value shove worse? Robl has played with Gavri a lot, if he rarely played with him I doubt he’d have the confidence to make such a big fold. When playing an unknown you need to account for uncertainty in your ability to range your opponent. In this case I’ll defer to Robl’s expertise and even though there is pretty big chasm between never bluffing and bluffing 9%1 of the time you have the nut blocker in your hand. I do believe that results matter quite a bit when making a big hero fold, except that the results don’t tell you quite enough here because if Gavri has the nuts here 75% of the time, Robl would still need to call with a set of twos given his pot odds.
The people who have chimed in about how this is an easy fold, seem to be making a pretty common mistake that happens in the discussion of viral poker hands. The goal in a poker hand is to make as much money as possible, the goal on social media is to be right. If this fold is -EV, you’ll be losing money, but you’ll usually be right, there are not pot odds in the discourse. This is why I think Nick Schulman’s claims of it being the greatest fold of all time are much more in line with the reality of the play than Fedor’s claim of it being “trivial”. This hand is not just about having the discipline to fold the second nuts when your opponent has the nuts, but having the confidence to know he won’t shove a jack high flush or bluff in a spot that’s pretty easy to bluff.2
This hand begins with a preflop call from Tilt that is too loose, it’s followed by a preflop play from Robl that is too loose. Gavri likely c-bets too often, Tilt makes a losing flop peel. The next couple plays, the flop check-raise, the call of the check-raise the turn checking through are all fine plays. However the meta in these games involves extremely loose play preflop, on the flop, and on the turn. On the river many people replied to me saying they like Robl’s bet size because they think he’ll get looked up by top pair too often. Gavri is an experienced poker player to understand the third nuts is not worth raising to this action, but people seem so confident he will massively overplay top pair? The worst type of poker player you can be is one that is too loose on early streets and too tight on later streets and it feels like everyone is suggesting this is the way to play. Viewers of HSP see that everyone’s strategy is too loose everywhere, until the last bet goes in on the river when everyone turns into Allen Kessler, I’m not buying it. This is not my first day analyzing a poker hand, I understand there is a class of player who is too loose everywhere, but will never bluff raise the river. Gavri might not be Alan Keating and I don’t think either of these bluffs are similar to bluff raising the river for practically a minraise over an overbet, but he’s shown that he has some moves in his arsenal, even if it’s not this exact one.
I think it’s easy to look at Gavri’s timing and conclude that he’d think a little longer with a worse flush and that he’d never bluff so quickly. I think the first point is valid, but it’s hard for me to say because I’ve never played with him. The second point is tough to say, some people like making big bluffs quickly because it looks strong, other’s do it because they don’t want to give themselves a chance to chicken out, so they shove before they can have second thoughts about it. Ultimately, I think the timing is enough that I’ll defer to Robl’s expertise playing with Gavri and the fact that Robl was right and conlclude, Gavri probably had the nuts enough here that Robl’s fold was the right one.
Final Thoughts and Grade
A normal pot on HSP involves several players making extremely loose, unsolver approved play. I tend to not write about these sorts of hands because while I am interested by the drama of these hands, I am not that interested in them on a strategical level that would fit what I’m going for in POTD. What drew me into writing about this hand was that I was shocked by the confidence of to use Nick Schulman’s term the “civilians” that this was fold was easy. In games like the ones on HSP that are too loose at every single node, the idea that folding the two card second nuts would ever be obvious or easy is crazy to me. If you think folding the second nuts in a spot like this is easy, you should probably follow Nick's advice and start bluffing more yourself. You also might want to consider the possibility that other’s in your game are constantly bluffing you and are shocked by all the huge folds you’re making. To me Robl did make one of the best folds of all time, he also picked the wrong size on the river. So I’ll give him an A+ on the river fold, combined with a B- in the rest of the hand, which averages out to what I believe is the first A in POTD history an
A-
POTD #265 #onemorething
So the solver analysis I used today is admittedly a bit of a Frankenstein solution. The hand is CO vs SB vs BB, but given QJo and Q7s are supposed to pure fold preflop and these games play very loose preflop. I made is BTN vs SB vs BB so we’d get something a little closer to the actual ranges being used in the hand. I still don’t think these are close to the “correct” ranges in the hand, but it’s closer. Even shifting preflop positions one spot over, Q7s does ’t sneak into Robl’s range, but I added a little bit of it to see how the solver would want to play it. The solver plays it very aggressively, but this is in part because it’s such a low reach hand for Robl. If I gave him all the Q7,Q8 and Q9s combos I suspect we see some overcalling on the flop with Qh7h.
To make the tree smaller, I only ran the flop with the sizes actually used, which I felt were incorrect. I thought that Gavri would have a small c-bet size, but upon further reflection it makes more sense he c-bets rarely and pretty polar. This is not like a PFR vs SB vs BB hand, but closer to PFR vs SB vs SB. Robl isn’t peeling J2s here preflop. Gavri has position and a range advantage, but he’s against two pretty specific ranges and can’t just auto bet. So it makes sense he’d bet rarely and pick a larger c-bet size. When Gavri bets 2/3rds pot, Tilt starts folding a lot of top pair in the SB. I didn’t test to see if Gavri would want to bet bigger, but folding to pair to one bet is very nice.
Similarly, I thought Robl’s checkraise size was too large, but open further reflection it makes a lot of sense. His bluffs are gutshots, flush draws and Tx, his value range is two pair and sets. He wants to pick a size that is big enough that it can get hands like flush draws and top pair to fold right away. A polar check-raise that targets top pair folds, should be on the big size. The turn and river solve did not require nearly as much guess work, but the outputs are only as good as preflop and flop ranges. I entered the Rocket Solver ranges into PIO and ran a vanilla turn sim, I suspect if Gavri is c-betting too much top pair on the flop and calling a checkraise with it too often that betting the turn works out very well for Robl. On the river, I needed to run a sim where I only give Robl one size to force some volume into that size that isn’t just nuts or nut blocker, but the general shape of the solution remains the same, Robl only gets shoved on by the nuts and but blockers, but still needs to call a jam with a flush because he blocks too much value.
This is assuming Gavri c-bets the flop, calls a check-raise and checks the turn with various hands with the Kh in them at the same frequency the solver does. I don’t think Gavri is playing these exact frequencies, but I think these estimates are reasonable proxies.
I made this point when I wrote about Jesse Lonis’s big fold, but getting to the river with an offsuit hand with the Kh in it should happen relatively often for Gavri. Occasionally recognizing that you block the nuts and going for a big bluff is not that sophisticated a play. I don’t know if Gavri will bluff here very often, but I am sure he’s aware of the fact that it’s a play one can make.