POTD #257 GTO Duel vs Stephen Chidwick

Octopi's George, puts me in a tough spot.

We have a long post today, so I will keep this introduction to the introduction brief. Today is March 24th, which is the official Punt-o-versary (surely I can come up with something better than this). It was March 24th last year when I first published POTD #1, a post that was, by current POTD standards, very short and formatted oddly. I’d like to thank all unpaid, paid, and premium subscribers who signed up for the journey, and if you are one of the early birds whose subscription is about to lapse, you can re-up using the discount code https://www.puntoftheday.com/2026 for 26% off an annual subscription. If you are still on the fence about paying, even at this LOW LOW PRICE! ACT NOW!, it still helps me and POTD to become an unpaid subscriber, which you can do at the link above or by clicking the button below. On to today’s post, which is free for all to read as a Punt-o-versary (I’ll work on something better for 2027) gift for all.

On February 5th, I engaged in a GTO Duel on Octopi Poker where I faced off against Octopi pro Stephen Chidwick. We played vs. “George,” Octopi’s solver, and alternated decisions. For every error we made we would lose points, and the player who survived the longest would win the duel. I’ll link to the entire video below and you can watch for yourself to see who won, but for today I will be reviewing the biggest mistake I made in the match. But first, I’d like to talk about playing vs. trainers and how you should do it.

I’ve written about using a randomizer to systematically randomize in POTD #108, and while it’s not a requirement for all decisions or at most stakes, I’ll reiterate a point I’ve made in that essay: Every poker player randomizes. Middle position opens and you have AJo in the CO, what do you do? Most people would call sometimes and three-bet sometimes— they are playing a mixed strategy where they chose two options that they’re indifferent between. Against top players, I think it can help to randomize systematically, but it’s not a requirement, especially if they’re trying to play exploitative poker versus you.

When you are playing vs. a trainer, I think it is mandatory you play with an RNG on. The interface in most solvers makes using an RNG easy, and you don’t need to worry about wasting mental energy being too precise or causing yourself to act too slow, concerns when playing actual poker that fade when practicing vs. a solver. When you are training, you are aiming to play the most perfect precise poker vs. the toughest opponent possible. In real life, you may be tired, or there will be financial stress, or you will be 12-tabling and need to act immediately. There will be enough challenges that having an RNG spit out a number at you may impede your thought process; in actual play, thinking clearly is more important than nailing the exact solver frequencies.

The goal of preparing for a competition is to make practice more challenging than playing, so you’re overprepared during the actual competition. Like a baseball hitter putting a donut on their bat in the on-deck circle, you want to use an RNG when playing vs. a solver so you’re not just focussing on finding a correct play, but also finding the correct frequencies. The other reason using an RNG is necessary is that if you just try to minimize EV loss vs. the solver, there are a lot of tricks you can use that will help you in training runs, but not real poker. If you take the passive route at every decision, the pot will be smaller so it will be less likely you’ll make a -5bb mistake on the river. Similarly, if your goal is to minimize EV loss and not worry about frequencies, you will be incentivized to always pick the safe play. Let’s say Octopi’s George raises UTG, you defend the BB with A8. The flop comes AhKh2c, you check, and George bets 75% pot and you roll a 90. Well, did you know that weak top pair will raise the flop and fold to a shove? If you did, you’ll also likely know the safe play is to call, and it won’t lose EV. If you practice without using an RNG, you’ll always call, and you’ll never find these sorts of weird plays that may or may not be ones you want to implement into your game when you’re playing vs. flesh-and-blood humans. If you’re unwilling to try tricky and unconventional plays vs. a trainer because you’d rather not lose EV, then I doubt you’d try them with real money on the line. So enjoy watching Stevie and I play vs. a trainer and try to nail not just the correct plays, but the correct frequencies. It was a fun challenge that Justin Bonomo called “the best free strategy content I’ve ever seen.”

GTO Duel #5 Stephen Chidwick vs Sam Greenwood

80bbs deep LJ vs BB single raised pot. 2.3x open size. I have A♥️K♠️

Flop (6.1BB) A♠️5♥️3♠️: George checks, I bet 1.525bbs, George calls
Turn (9.15BB) 2♠️: George checks, I bet 2.75bbs, George calls.
River (14.7bbs) Q♥️: George leads 11bb, I call.

If you are an Octopi subscriber you can view the sim here

Since this is the Punt-o-versary we are giving unpaid subscribers two free posts this week. Yesterday’s and today’s. If you aren’t a subscriber and would like to become one, simply click the button below (or sign up using the discount code at the top of page)

What I Was Thinking

I know the ace-wheel-wheel flops are rarely a pure c-bet from the preflop raiser and also tend to be boards that favour a small bet over a big bet. My instinct in this spot is, I am more likely to pick a small size with combos that have more backup. The Ks gives me a backdoor nut flush draw, which is quite a lot of backup, and the Ah blocks backdoor nut flush draws, which will give me a hand that will be a stronger bluff catcher/value bet on heart/heart runouts. All of this led to me to c-bet a small size, and I was correct. It’s a pure bet in the Octopi sim.

The turn is a tricky one, but I was fortunate enough to tag in one of the best players in the world in Stephen Chidwick, and he decided to bet small because we have a range advantage, but not a nut advantage, and because we had a hand that can comfortably continue vs. a check-raise. So he blocked third pot.

On the river, George makes an unconventional lead, and my thought process was that, top pair top kicker should beat all his bluffs, and with the Ks I block the nuts. I have a strong enough hand that it should consider bluff-raising, but one that would never be a pure bluff-raise. However, the RNG was a 4, so I decided I couldn’t bluff on such a passive number, and that I had no other option but to call with a hand that blocked value and beat all his bluffs.

What I Got Wrong

I did not play two streets in this hand: preflop and turn. One of those decisions was made automatically, while the other was aced by Mr. Chidwick, so let’s focus on my decisions. I am a little more inclined to bet the flop with Ks; having a hand that is more willing to put in 80bbs on some runouts makes me more interested in building a pot. The Ks blocks many of the BB’s weakest check-raise bluffs on this board, which hurts our EV a little bit, but we still like betting. Octopi does not use a bigger c-bet size than quarter pot; while my hand mixes checks, there is no decision on an aggro seed.

On the river, I have a nut blocker and I beat all his bluffs, so why is my call so bad? Because George would never play the nuts like this. If you get to the river with the nuts you want to try and get all-in, which in this case would mean check-shoving. Let’s say I had a hand like 5s5x and faced a 75% lead; I am calling every time. If I face a check, I’d bet and still sometimes call a check-raise all-in. So the AxKs is blocking the nuts, but it’s in a situation where my opponent rarely has the nuts. So let’s dig into what hands are in their odd leading range here.

It’s mostly straights, an occasional flush, and then as a bluff it’s almost exclusively KJ, KT, K9, and K8 with the Ks. So I block almost none of his value bets and almost all of his bluffs, and I still lose to all of his value bets. I couldn’t pick a worse hand to call a river bet with here. I’d literally rather call every single other top pair combo before this one. I’d rather call Qd8d, a no-equity bluff that rivers a pair, or pocket nines. Normally, beating all bluffs and blocking the nuts is a good reason to call the river; in today’s hand, the mechanic was reversed, and it cost me my duel vs. Stephen Chidwick

Types of Error
Getting blocker effects wrong

Grade

I think this hand fits the theme of Idiot Week, because if I played this hand vs. a human and they led the river vs. me, my first thought would not be “what should I do here?”, but “that’s a stupid fucking lead.” Obviously when George leads, I know it’s not a stupid fucking lead. In hindsight, I think there are two ways to think about my punt in this hand. The first is that I faced a decision I’d almost never face vs. a human in a real-money poker hand, and I made a blunder that doesn’t really matter because I’d never face this line from a human. The second is, I was thrown into an unfamiliar situation under the pressure of playing a competitive match vs. one of the best in the world, and I was unable to reason my way through something I was unprepared for. Whether you’re playing vs. a trainer or a human, you will always be thrown into unfamiliar situations, and it’s your job to problem-solve and execute. In this hand, the solution is not even that complicated. The BB would never lead out with the nuts; George mostly leads straights, and George mostly bluffs with the Ks. So when I have the Ks in my hand, I have two options: Should I try to bluff George off a straight, or should I fold? I rolled an extreme seed and still called, which meant that folding wasn’t even on my radar. It’s an unusual spot, one I’d rarely face on the felt, but in this hand I was playing vs. George and Stevie, and I lost vs. both of them making a big avoidable blunder.

C-