POTD #59 Jason Koon vs Espen Jorstad in the Triton Montenegro 150k

Reviewing an interesting poker hand and twitter thread.

Around a month ago, WSOP Main Event Winner and GTOWizard Pro Espen Jorstad had a tweet thread about an “insane” spot from a $25k SCOOP Final Table. I flippantly replied that the sim was “broken” and the output was insane because the sim was bad. I don’t stand by my word choice, but I do think the sim had some problems with it, mostly that it didn’t consider FGS and that it didn’t have a large enough three bet non all-in size from Kahle, and that ICM is a pretty simple model and not the gospel. There are also potentially problems with the postflop game tree, but the sim is preflop only, and I do not know what the whole game tree looks like. Espen respectfully responded and called my response "lazy”, while CoinPoker Pro and Men’s Rights Activist Bencb said my response was “The problem these days”.

Last week, Espen posted another Twitter thread, this time about a hand he played vs. Jason Koon in Triton Montenegro. I had some quibbles with his thread, and I thought of a quote from Jean Luc-Godard, who said “In order to criticize a movie, you have to make another movie.” So today, in order to criticize his sim, I am going to make another sim, and my criticism is not even going to be about either of the actual hands they had, which I think were pretty well-played. Instead, we’re doing a deep dive into one sentence at the of Espen’s thread. Buckle up for a thrilling adventure.

The Hand

Triton Montenegro - Event #12 $150K NLH 8-Handed
500/1k/1k (SB/BB/BBA)

It folds to Espen (185k) in the HJ who raises to 2.5k with J♣️9♣️, Jason Koon (185k) calls on the button.

Flop (7.5k) 9♦️6♠️5♠️ Espen checks, Jason bets 2.5k, Espen calls.
Turn (12.5k) 9♠️ Espen checks, Jason bets 8k, Espen calls
River (28.5k) A♦️ Espen checks, Jason bets 28.5k, Espen calls and loses to Jason’s 55

What Espen Was Thinking

Espen thoroughly laid out his thought process in the Twitter thread linked above. He believed that the flop was pure check with range, as many low-to-middling boards where a player can flop a straight are. This deep, he wasn’t interested in raising mediocre top pair on the flop. On the turn, he didn’t think he would raise 9x when a flush filled, and especially not a without better kicker or flush or straight draws. On the river, he thought a 9 blocked some value bets and his hand was strong enough that he could not fold vs. this action, so he called and was shown 555.

What Sam Thinks (Some Cheating)
Normally, I try to give my comments on a hand without looking at outputs, as it’s unfair to the player I am criticizing. However, I became aware of this hand through Espen’s Twitter thread, so I have seen the outputs he wrote about. I think this is a well-played hand and mostly a cooler. Jason could check the flop or turn with 55 from time to time, Jason should probably bet a little smaller on the turn, and Espen can pick his own adventure on the river, including leading the first time around, but it’s all fine. What irked me about Espen’s thread is something he said early on:

In general you want to be cautious of building a pot OOP when playing very deep on draw heavy boards. I decided I was gonna check my range.

…

On the flop the solver did actually want a small betting range, but as I decided in-game to simplify and play only checks, I have nodelocked the solver to do the same.

We all make mistakes, and I believe Espen is telling the truth that he played this flop as pure check when it is not, but the problem is, he’s node-locking the solver to do that when there is no reason to believe Jason will think Espen is pure checking the flop. Jason is a very good poker player and respects Espen’s game; Jason might not know what OOP’s exact flop strategy should look like, but I’d bet as he was playing the hand, he didn’t give Espen credit for dark checking.

Why does this matter? They both played their hand fine; is this worth dedicating an entire blog describing a “punt”? Well, there are two things going on here: One is selfish, but I believe in self-improvement, and I believe it's a much better use of my time to turn my energy into a blog post instead of griping on social media. Two, I think Espen’s thread shows the perils of node-locking, and that is useful information for my subscribers.

The Individual Detail That Got Me To Write This Post

In Espen’s final tweet he writes

Jason is of course a super strong player and much better than your average opponent playing this deep, so I won't comment on what his river strategy would be in this spot. I do however doubt that population would check back Ks8s almost pure, as the solver did (IP range vs check in screenshot below). That's a rabbit hole I'll avoid going down today - but I think the main take-away from this solution is just how careful you're supposed to be playing 200 bb deep. I think for most MTT players, it's going to feel very unnatural to check back strong flushes on this river

So let’s go down this rabbit hole. If Espen plays bets on the flop, as the solver does and as most poker players would, the solver never checks back a flush on the river. In Espen’s sim, the smallest river lead size was quarter pot, and OOP leads the river 5% of the time. In my sim, I also put in a 10% pot lead, and the lead frequency increases to 14% (11% b10, 3% b25). The OOP leading range consists of Ax, 9x, some flushes, and some bluffs like KsQx or 2s2x. If OOP never leads on the river, we start seeing check backs with exactly QTss and JTss (Ts9 is a high frequency call for OOP) and 43ss.

I think Espen’s advice that you’re supposed to play careful OOP 200bbs deep is sound, but it’s not the only lesson to take from the hand. He “punted” by dark checking the flop, and to me the lesson of the hand is, when you node-lock someone to a flop strategy that is not right, the rest of the sim might not be quite right. I think node-locking the river to pure check is a fine use of a solver, because it does not have implications down the rest of the game tree, it only affects the river strategy. I also think it’s a more realistic assumption that OOP isn’t finding a pretty unintuitive river lead on a relatively blank river than to assume that they’re pure checking the flop, that IP knows this, and that both players play accordingly.

Grade

I think Espen’s range strategy was off on the flop, in a way that wasn’t costly and he played the rest of the hand fine, but it sounds like his river range strategy was a little off. Missing a potential lead block, but also not giving enough creedence to folding or shoving. If Jason were IP and checked back Ks8s, i’d have given him a C, but for Espen’s side of the hand, I will give him a

B-