POTD #179 Nate Silver's NAPT Punt
A Thanksgiving Punt. Who will be Leon Lett? (yes it was technically a FG bear with me)
Yesterday on POTD, we celebrated Maria Konnikova winning a Spadie at NAPT Las Vegas by looking at a hand where she had punted earlier in the festival. Today, we will look at a hand from her Risky Business co-host Nate Silver before determining which podcast host punted the most at Resorts World earlier this month. Maria sent me four hands that she felt were interesting where she might have punted; Nate was kind enough to send me dozens of hands he played throughout the trip, which unfortunately puts him at a bit of a disadvantage as he gave me more grist for my punting mill. However, I also believe Maria spent more time in Vegas and played more hands of poker, so the four hands she picked were from a larger sample of potential punts. If only we had a data scientist and someone who wrote a book on con artists that could help me with unskewing a self-reported dataset or determining if the data is legitimate. Read on below to see who will get anointed with the dunce cap (or crown?) of being the biggest punter of the NAPT.
NAPT High Roller $10,300 Event #42
(400/800/800) (SB/BB/BBA). 50k Starting Stack. Registration is Open.
It folds to Johan Schumacher (59k) in the LJ who makes it 1600, it folds to Nate (90k) in the CO with 9♣️9♦️ who makes it 5500, it folds back to Johan who calls.
Flop (13k) A♠️9♥️5♥️: Johan checks, Nate bets 6k, Johan calls
Turn (25k) T♠️: Johan checks, Nate shoves for 47.5k, Johan folds.
What Nate Was Thinking
Nate was kind enough to share his thoughts with me and they are quoted below. They have been lightly edited for formatting and clarity.
Johan Schumacher opens the LJ to 1600 on 400/800 blinds; most people have been opening to 1800 instead but that is his standard. I have maybe 90K and cover him. I 3-bet to 5500 with 99 from CO. The size is I guess a slight brainfart if we want to go ~3x as most people had been opening to 1800, but I’m not sure it’s a big deal or whether he’ll particularly notice. He checks, I bet a little on the larger side (I think like 6000), he calls. Turn Ts, so now a double flush draw on the board. He checks, I shove for what is maybe 2x pot effective, he thinks for maybe 20 seconds and folds before using a timebank. Solver doesn’t seem to like fastplaying so much, though there are a lot of combo draws on this board that I can sort of be repping. In general, these big overbets for value don’t seem to work for me very often in tourneys (so maybe I should be using this as a bluff line more?)1 whereas they have a somewhat higher hit rate in cash. Not sure I hate it against him specifically as we have a little bit of a history of battling.
After I told Nate that I was going to analyze this hand he messaged me some more thoughts about his line.
I think part of the problem is that my flop sizing is a little larger than the solver recommends. So we end up with this SPR of ~1.9 on the turn, I believe my thinking was like: I’m already fairly polar, and I want to make combo draws indifferent. But if I b70 or something, now there’s only like a third-size to half-size bet left on the river. For some reason, I have an aversion to leaving these small bets behind (note that this isn’t *that* small), whereas solvers seem to have no problem with them.
What I Thought (no cheating outside of knowing the results of the hand)
Three betting pocket nines 75 bbs deep is a rare play, but a necessary one to sometimes make. It gives your range some medium-strength hands that play back vs. a four-bet and some hands that can bluff on high-card boards. The c-bet strategy on this board seems close; you don’t always c-bet ace-high boards because you have a lot of TT-KK type hands, but this board is not that connected and you often pure cbet on dry boards after three betting. I think you play two sizes with range on the flop, a bigger size centered around AK/AQ and some bluffs. Either way, I’d guess 99 pure bets and mostly bets small, because you want to induce action with middle set.
Nate’s size seemed a little big, but not so big that it would lose more than a thimble of EV. The turn is a different scenario— Nate says his shove looks bluffy and represents draws, but my concern is that his shove looks a lot like AK/AQ, hands that are almost always best right now, but aren’t happy on almost half of the deck on the river. Nate’s turn shove is large enough that even a monster draw like QsJs or JhTh would need to fold to a 2x pot turn shove if Nate turned AcQc face up. If Johan has two pair plus, he’s already stacking off, so your strategy with 99 here should be built around is squeeze as much money out of top pair that will check-call or draws that will check-call or check-raise, and shoving does not do that.
What We Got Wrong
Nate wrote that his preflop size was a little large and that he wanted to go ~3x, but most people are opening to 1800 and he had a brainfart, but this deep it does seem like the solver prefers going somewhere closer to 3.5x, so Nate’s size is appropriate. My guess from reading Nate’s other hands is he was probably three-betting 99 a little often here, and I was probably three-betting it not quite enough, but combined we probably would be close to the 25-30% three-bet frequency the solver wants with 99.
The cutoff does not c-bet 100% here; they check some obvious middle of range hands like KK and weak top pair and some obvious traps like AA. 99 rarely checks and 55 never checks. You’d think that, given we mix checks on the flop, our betting range is polar and we would pick a bigger size occasionally, but we do not. It’s small size or check, and Nate’s size loses around 1/4 of a BB in EQ.
The turn shove is simply overkill. It’s a large enough shove that hands like top pair with a flush draw and AQ should be indifferent. So our shove is simply not getting called enough, but there is another consideration here. Johan is already supposed to play very tight on the turn; he’s already supposed to start folding a lot of top pair, including hands as strong as AQ, to a 60% pot bet. So while Nate thought his shove might look weak and be a good exploit, I actually think this is a spot where the correct exploit is to consider that Johan is probably not folding top pair nearly enough on the turn, and making him put 1/3rd of his remaining stack in the pot drawing dead 100% of the time is higher EV than occasionally getting him to hero call off drawing dead.
What Sam Thinks and Grade
Nate has played a lot of poker and has strong instincts about his table image and how to leverage it, but I think today he failed to marry those instincts with some technical details about the hand. Whether it’s preflop or postflop, one of the classic lines before making a big hero call with a weak pair is “I put them on AK.” The problem is, that’s not what you want your opponent to think on a connected ace-high board, because if they beat AK, you’re probably getting all of their money anyways. If the board were JsTs5h4h and Nate had 55, I bet the solver would not shove 55, but I believe Johan would almost always call with a pair plus draw, and he might even hero call with a hand like AT if he put Nate on AK. On a board like that, I could get behind Nate trying to exploit someone by making his hand look bluffy, but I don’t think this is the right board, and I think he’s more likely to induce a hero fold than a hero call. Nate’s flop size loses around 0.3bbs in EQ, but the turn shove loses around 3.3bbs. That’s a lot of EV to give up vs. a strong player; you’d need an airtight read to deviate in that fashion. I remember Ben “Sauce123” Sulsky told me his advice to beginner poker players was, “bet big when you have a great hand.” Like most of Sauce’s advice, it’s wise, but sometimes you can bet too big and it can end up costing you a lot more than a bad hero call or an ill-advised bluff. Today is one of those hands.
Picking a Winner
Both Nate and Maria made small flop mistakes and big turn mistakes. I initially thought Maria’s flop check-call lost EV, but it turns out I was wrong and it’s the less common play, but doesn’t lose EV. I was off on Nate’s flop strategy and his actual flop size loses a tiny bit of EV, but both of their plays are fine. They both made a similar mistake on the turn, they put too much money in the pot with their hand, Nate had such a good hand he needed to squeeze max value, while Maria’s range was too weak to raise at all.
We’re at Punt #179 and I have started developing an internal taxonomy of punts and my grading usually hews to a principle that putting in too much money with a bad hand is the worst kind of punt. Overplaying draws is something I have personally done several times on POTD and you might think that I would reprimand Maria for making the same mistake I keep making. Instead, I will show herself and myself some empathy: when I first read the hand she sent me I thought “This turn lead seems silly. Let’s raise him up. Sure why not?” it was only open further reflection that I concluded her turn raise doesn’t make much sense. When looking through Nate’s hands, I visibly recoiled when I read that he shoved a set for almost 2x pot. My overall instinct is to punish hands where we put in too much money with a bad hand, but in this case I will allow my primal instincts to guide me. If I visibly recoil and ask “Why would you do that?” aloud to no one, you get a C- and are awarded the Punt of the NAPT.
Happy Thanksgiving and thank you to Nate and Maria for participating. If you are subscribed to this Substack, you are likely subscribed to their Substacks, but in case you aren’t. Nate recently wrote about the politics of poker players
and Maria wrote about the use of AI in the NO LIMIT documentary
I think one thing worth noting here is that in general overbets *should* work a lot, so when they work a lot, it doesn’t necessarily mean your opponents are over-folding and you should be doing it more.