Can ChatGPT Help You With Poker? (ft. Tim Adams) | Upswing Poker Level-Up #65

This article is a transcription of the Level-Up Podcast, hosted by Upswing VP Mike Brady. You can watch or listen to the entire episode via the links above or read on if you prefer a written version.

Mike Brady (00:00):

Can AI chatbots help you with your poker game? No, not really. In fact, it might make you even worse. I’m Mike Brady and my friend Stu has been on a bad stretch playing tournaments lately, so bad in fact that he turned to ChatGPT to analyze his hands. He ran the hands and analysis by me and I instantly knew that this had to be shared with you here on the Upswing Poker channel, but I’m not alone. To add some proper credibility to our analysis today, I’ve brought out a big gun. Two-time Triton Main Event champion, two-time Super High Roller Bowl champion, the lead coach of the new modern tournament mastery course coming out soon on upswingpoker.com. This guy is truly a top pro. Thanks for joining us, Tim Adams.

Tim Adams (00:46):

Thanks for having me. I’m excited. Let’s see what we have in store.

Mike Brady (00:49):

Yeah, the robots cooked up some good stuff for us. So here’s how this will go. I’ll walk you through a hand that my friend Stu played then read ChatGPT’s analysis. From there, Tim will critique the bots analysis and break down the hand properly. As usual, I’ll ask follow-ups on behalf of you, the viewer. Ready to do this, Tim?

Tim Adams (01:11):

I’m ready. Let’s do it.

Mike Brady (01:13):

Our firsthand blinds are three hundred six hundred. This is a live tournament in Los Angeles and our hero raises under the gun with Ace Seven of clubs to thirteen hundred chips, the cutoff and the big blind who is noted as a fish, calls. ChatGPT just had one word to say about the preflop play here. Fine. What do you think, Tim?

Tim Adams (01:33):

I mean, one thing that I would just nitpick about is when you’re telling a hand history, like, already I’m not sure of our hero’s stack size, which would be important to know. We have the cutoff calling. I also don’t know the stack size and the big blind overcalling. So maybe this is being nitpicky, but if you’re sharing a hand history for the reader of the hand history to conceptually understand what the stack to pot ratio is, tell the stack sizes, if it’s in equal stacks, maybe tell what the cutoff is flatting off of. It’s going to look a lot different if it’s thirty blinds or fifteen blinds. Our ranges are going to change a lot. And when you say under the gun, you have to describe what that means. Is that under the gun eighth-handed or six-handed? That’s going to be the difference between a twenty-four percent opening range to an eighteen percent opening range.

(02:23):

So it’s going to matter a lot. And then finally, for ease for the reader, I would always show the size of the pot per street, again, just to make it really easy. I basically bitched to my friends about this type of thing who sent me a hand history, they asked for opinion, and then I’m spending a couple of minutes trying to figure out what’s the size of the pot, how big did the fish in the spot lead. So just to already get going and talk about that where I’m being a bit of a baby, but yeah, just streamline this for me.

Mike Brady (02:55):

Yeah. I think that’s super common. I get this a lot when people ask me questions about hands or friends will run hands by me. They’ll kind of give me just a little piece of the puzzle and I feel like I need to, like you said, extract all of the rest of the info. And I think it also indicates a potential lack of thinking about the things that are important. Are you caring about how deep the cutoff is here? Are you caring about how deep the big blind is here? Because all that stuff will matter. And I think Ace Seven of clubs, regardless of all this stuff, is a fine open, it’s a fine open size. There’s not too much to talk about there, but the lack of detail is what’s potentially something to work on for our hero here in this hand. We can move on to the flop from here though.

(03:42):

It comes Ace Jack Nine with two spades and the fish leads out for twenty-two hundred. There’s something like five thousand, fifty-five hundred in the pot. So it’s like, let’s call it a forty percent pot lead. Both Stu and the cutoff call, what ChatGPT had to say about this one? The fish donks into two players on Ace Jack Nine. Calling is best, raising accomplishes little. Once again, our hero has Ace Seven. The board is Ace Jack Nine with two spades. What do you think about the flop play in general here, Tim?

Tim Adams (04:12):

Well, for the big blind leading, well, definitely not a favorable board for the big blind. Overall, so this is like Thirty BB effective. There just wouldn’t be that much leading in theory even on low and connected boards. Globally, you’re probably leading less than one percent. So just very little leading. You have the weakest range from the big blind. Even if you do connect well with those, say low and connected boards, like this board for sure, there’s no leading. Essentially, you just have the weakest range, you have a bunch of misses. Overall, there just should not be much leading from the big blind. As played, since this is what the VIP or the casual player or the fish, whatever you want to call it, did do. I think as the IP player, as our hero here, just calling and going from there, you do have another live hand behind you.

(05:02):

So I think just calling makes the most sense.

Mike Brady (05:04):

What’s your threshold for raising here? Obviously Ace Seven pretty dumpy kicker, but I assume once we get into the Ace Queen, Ace King territory, we want to start mixing in at least a good bit of raising. How are you thinking about that?

Tim Adams (05:17):

These spots are so sensitive to the player you’re up against. If this is the type of fish who leads and overvalues the strength of a hand like a weak Ace, then yes, always raising a strong Ace. If it’s someone who I’m kind of readless against, I like even with good Ace x taking it slow and seeing where it goes. But again, it’s so, so player dependent that it’s hard to just pinpoint an exact strategy that I would have.

Mike Brady (05:45):

Sure. Makes sense. Well, we’ll get a little bit more info about this player as the hand goes on, but of course he didn’t have it in the moment. The turn is another Ace, doesn’t bring any extra flush draws. So our hero now has trip Aces on Ace Jack Nine Ace. The fish leads again for four thousand. That’s four thousand in a roughly, let’s call it twelve thousand. Stu now has pretty interesting spot. I think he’s got about 20K behind or something like that. He raises to fifteen thousand. The cutoff gets out of the way and the fish calls all in with Ten Eight offsuit for a flopped open-ender. That is a not super great shape, especially against Stu’s hand. His Seven isn’t even live. Now let’s go over to ChatGPT’s analysis. Once the Ace pairs and he bets again, your jam is pretty clearly exploitative rather than solver driven.

(06:36):

And against this profile, I think that’s good. This is getting into some juicy bot analysis here. “Why”, the bot asks? Because population donk bet call ranges here are weak Ace x, draws, pair plus draw, and random spaz. Interesting that ChatGPT is using the non-politically correct spaz there. Your stack to pot ratio is awkward enough that jamming denies equity and gets looked up by worse Ace x, which it’s worth noting that a kicker wouldn’t play here, surprisingly often. And the result is a reminder of why live poker is beautiful and horrifying simultaneously. You got called by Ten Eight, drawing dead except to the Queen. One Queen, actually it says. That’s not a bad beat story. That’s an advertisement for why these tournaments are profitable. Absolute chef’s kiss of a ChatGPT sentence at the end there. The classic it’s not X, it’s Y sentence structure. First of all, let’s talk about the hand in general and then we could talk about kind of what the bot was saying.

(07:37):

So we faced this four k bet on the turn. It’s like a third pot. How are you looking at this spot as Stu when you turn trips?

Tim Adams (07:45):

Again, we’re out in the woods here with this type of node since it’s again so player dependent since in theory the node doesn’t exist. But there are upsides for both plays. If you do call and say the cutoff folds, there’s going to be half pot on the river. You have to ask yourself, is your opponent the type of player to follow through with bluffs? Maybe there’s an upside to calling. Another upside is if you just call and say the player behind you jams, you’re getting a lot more information, maybe the big blind then jams and you can get away. Even if the cutoff jams and it’s back to you like the big blind folds, like you’re probably not going to be in great shape potentially. These things are all, again, you’re out in the woods. If you do jam, you are … Well, you’re not really denying equity from the big blind because we already know the outcome of the hand.

(08:36):

He did bet-call like a terrible draw, but you are denying equity from the cutoff. So say they have a good flush draw or something. So there’s a bonus there. So again, it’s really, really player dependent. I really think you can go either way. In this particular spot, I’m kind of giving credit to our hero here because it seems like he pinpointed his villain and basically got the villain to play the hand terribly.

Mike Brady (08:58):

Yeah. It definitely results wise worked out really well for Stu and he clearly had a beat on his opponent. Moving on to what the bot ChatGPT was saying about this hand, this is kind of going to be a theme throughout this video where it just says terms and words all over the place. I actually find it really fascinating how often it does this where it just uses words, like random phrases that it read on upswingpoker.com or some other site. It just generates random text in a way. So at one point it says the jam is pretty clearly exploitative rather than solver driven. Based on your analysis, you agree with that sentiment to a degree, right?

Tim Adams (09:39):

Well, again, for it to be a solver-esque hand, it has to be probably in our node that exists. This is definitely far, far and away from a node that would ever exist in the solver world because, well, the big blind would never ever get leads on the flops to begin with. So this is just an obscure hand where, yeah, at this point jamming the turn, is it solver driven or explo driven? Well, it’s an exploit hand. So I guess yes, it’s an exploit jam, but it makes a lot of sense anyways, I would say to rip in this particular spot even versus someone pretty reasonable because the pot is getting so big and kind of protecting your hand, winning what’s already in the middle. There are real benefits to that.

Mike Brady (10:23):

Certainly. It goes on to say that the donk bet/call ranges here. So the hands that the player is going to lead out with and then call a jam are weak Ace x, pair plus draw, which I guess that would be like Nine x of spades specifically and a random spaz. I don’t know why a random spaz would bet and then call off unless you’re calling this hand a random spaz. I don’t know. Stack to pot ratio is awkward. That’s actually kind of true. I don’t totally disagree with that. The part about jamming denies equity and gets looked up by worse Ace x, this is something that I’ve noticed the bots really don’t understand about poker. And for those who don’t know how LLMs like ChatGPT work, it’s not understanding poker in a holistic sense. It’s generating text. It’s trying to predict what you want to hear essentially.

(11:09):

And that’s why it’s able to say something like, you can get looked up by worse Ace x when of course if you understand poker even in a basic way, you know that three Aces with a Jack Nine kicker, those are your kickers. The kickers don’t play. If the guy has Ace Five, the Five and or Seven are not playing. So the bot’s just not even able to understand that it’s just generating text. I do actually agree with the end where this is a reminder of why live poker is beautiful and horrifying simultaneously. It’s very true, but I’m going to give the bots analysis what, maybe a D+ on this one. What do you think?

Tim Adams (11:46):

Yeah, I mean, it’s a complete word salad. A lot of it’s like speaking another language really, right? I’m trying to decipher what … There’s a few things that make sense and then there’s a bunch of other things that make absolutely no sense. So it’s really hard to wrap your head around it.

Mike Brady (12:00):

Get ready for more of that, Tim. By the way, before we move on to the next hand, Tim’s new course Modern Tournament Mastery comes out this upcoming Monday. Tim has been working his ass off on this thing for the last, I don’t know, eight, nine months or so and he even tapped a couple of his close friends to help make it even more valuable. Tim, could you tell our listeners and viewers a little bit about what you’ve been working on?

Tim Adams (12:25):

Yeah, I’m really excited to share the course we built. I was lucky enough to get Daniel Dvoress to jump along and join me in the project. In my opinion, he’s the best coach out there. Steven Chidwick also joins as a guest coach. So yeah, I was thrilled by that. He’s obviously a legend. I think he’s regarded as the best of the best. Essentially, the course just takes you through how I study, how I prepare, early stages, the mid stages, which can always be tricky, end game. And yeah, to wrap it up, we do a little bonus course on super high roller mechanics. So we hope the course is fully complete and gives you a lot to work with. At the end of the day, poker is about just gaining more knowledge and I hope that’s kind of what we provided to you guys.

Mike Brady (13:10):

I’ve looked through the course a lot, reviewed a lot of the content. The outline looks absolutely stupendous. It really covers all of the things that a top level pro like Tim finds important. He goes really in depth on those different parts of the game. And I think whether you play five hundred plus average buy-in online or three hundred dollar live tournaments, there is going to be something in this course that will really help you make more money over the rest of your poker journey. I would definitely say that, would you agree, Tim, that the people who are going to get the most value out of this, probably people who play average buy-in two hundred or more online or maybe like one ks live. It’s not to say that lower stakes players won’t get value, but that’s really who you’re mainly talking to throughout the course.

Tim Adams (13:55):

When you’re making content for a really broad group of people, it’s always this thing where some people are going to just get a little more out of it than others. And I think you nailed it. I think it’s mostly people playing two hundred dollar ABI online and like you said, maybe one k live where their goal is to move up and play higher or feel just more comfortable playing five ks or ten ks. There is a lot of information in there and it is pretty dense with theory, but we wanted to also make it interactive. So there’s a big stream review that Stevie and I do, which it’s a stream where we both final table the tournament. I do a lot of drilling versus the bot because that’s how I actually study in a lot of scenarios where I try to get an idea of the theory and then I actually try to put myself in scenarios where I’m in difficult situations versus an unexploitable perfect player and I just try to relay not only my knowledge, but how I feel about certain spots.

(14:50):

We’re always looking through solves with an exploitative lens so we’re not just fully robotic the way we look at poker. There’s lots of things available to you when you make a poker decision and that’s what we’re trying to get at, that there’s just so many factors and I hope the course in a general sense encompasses all those things that may make you a better player and that’s our hope.

Mike Brady (15:11):

Yeah. We truly guarantee that this will help your game. So make sure you check it out this Monday coming to Upswing Poker. And if you get it during launch week, you’re also going to get Tim’s bonus course, Super High Roller Mechanics, some super juicy, valuable stuff in there totally for free. That’s going to cost three hundred dollars basically starting next Saturday, but you get it for free if you get the course during launch week. That launch date is May 25th. That’s a Monday. If you want to pre-register and get notified when the course goes live, check out the link in the description. Moving on to hand number two. Blinds are five hundred a thousand. There is something that carries over from the previous advice that you gave Tim where we don’t know what stage of the tournament this is and when the blinds are five hundred a thousand, we are potentially approaching the money or some sort of important consideration like that.

(16:01):

But we do get the stack size this time, the small blind who is noted as a fish, limps in off of a thirty k stack and Stu checks in the big blind with Eight Seven suited. The ChatGPT bot did not comment on the preflop play. It only said, “This is probably my favorite hand.” I guess we’ll see why shortly. But I don’t think the preflop decision is necessarily trivial. What do you think about this spot in the big blind? Do we ever go for an iso raise with Eight Seven suited, pure check this hand? What do you think?

Tim Adams (16:27):

So with Eight Seven suited in particular, I would never raise this. You want to be very polar at this depth because getting limp-jammed on is a real possibility. So I would mostly structure my raises for say bluffs because I don’t want to be very linear. It would be very one high, one low, like Ten Two off. If there were suited hands, it’d be more like Six Two suited, Five Two suited type of hands. But Eight Seven suited is really pretty. I want to see a flop. It retains equity really well in a limp check. So definitely not interested in raising this one. I want to be very polar and this is more of like a middling hand that I don’t want to get blown off of.

Mike Brady (17:02):

Yeah. It’s such a disaster if we raise to three big blinds here, they jam and we have to fold a hand that probably has what, thirty-five, forty percent equity against the jamming range depending on what they’re doing it with. Just pretty much a disaster. So we’d much rather take a flop. But then I gather deeper stacked, you would consider raising this hand. If we started the hand a hundred blinds deep where you can actually raise and then potentially call a three bet, now it gets into the territory of being able to maybe raise with Eight Seven suited and similar suited connectors, right?

Tim Adams (17:29):

Right, exactly. So when we’re deeper, of course we’re not getting blown off of our equity. That’s a big part of this is at a hundred, if you face a limp and you definitely want to raise this probably pure, you can comfortably call a limp re-raise. So that’s not a problem. It’s just when your decision turns binary where you actually are getting limp jammed on on the shallower adept. So yes, you don’t want to be raising this particular hand.

Mike Brady (17:57):

Moving on to the flop then as Stu does check, we get the King Six Five rainbow with the King of diamonds. So we have a nice open-ended straight draw with a backdoor flush draw. The fish bets twenty-five hundred, so that’s roughly a pot size bet or near it. And Stu goes for a raise to seventy-five hundred, that’s a 3x raise and the fish snap calls. By the way, for anyone listening on podcast platforms, if you’re having trouble following along on the hands, we’re going to put the hand histories and maybe the ChatGPT analysis as well in the description of the podcast so you can check that out if you’re struggling to follow along at all. Just please don’t do it while you’re driving. fish bets twenty-five hundred into roughly a three k chip pot on that King Six Five rainbow board. And Stu with some backdoor draws, he decides to raise it up to seventy-five hundred, three times the bet, and the fish “snap calls”.

(18:52):

Moving on to the bot’s analysis, your flop raise is excellent. Strong words, strong equity, fold equity, range advantage from aggression. Live players massively overfold turns after calling flop raises with weak one pair of hands. Talk about word salad. Let’s talk first about what you actually think of this raise here, Tim. Are you okay with kind of turning this hand to like a mergie kind of not bluff, little bit of a value bet, or do you prefer just calling?

Tim Adams (19:21):

Yeah, my default would just be to call this hand, especially versus like a casual player who’s going to be unpredictable. I assume our hero here is raise and then calling versus a jam. Yo definitely can’t raise fold after you’ve put in so much money with so much equity. So yeah, my default game plan would be to call this and then just often raise strong King x and kind of expect my opponent to potentially overplay something if … Again, there’s a lot of nuance in that. But yeah, the small blind is betting close to full pot here. That’s usually a player who isn’t going anywhere. So I don’t like raising if you’re thinking about doing something crazy like raise folding, but I don’t assume that he’s doing that.

Mike Brady (20:04):

Yeah, interesting. I think against, this is just speculation on my part. I think against live players, raise-calling off is going to do really, really poorly here because they’re just going to have a King way too often, way too often. So that would actually make me lean even more towards what you’re saying and like really want to call because if we raise and get jammed on, yeah, we’re arguably forced to call off and I think we’re going to be in relatively poor shape there.

Tim Adams (20:30):

Yeah. Then I hate it because if you’re thinking about raising and folding, then that’s quite problematic. Yeah, I would always just call this then. And yeah, for bluff raising, I mean it’s limp check. You have so many hands to choose from. You can use Four Two offsuit, you can use even Nine Seven offsuit, but yeah, this one has way too much equity to ever think about like putting this amount of money in and then folding.

Mike Brady (20:57):

Yeah. And then looking at ChatGPT’s analysis, it really just is a word salad. I love the phrase “range advantage from aggression.” Really, really wonderful combination of words there. Yeah. I mean, live players massively overfold turns after calling flop raises with week one pair hands. I mean, I think everyone kind of folds turn a lot when you call the flop with a weak one pair hand. That’s kind of how poker works.

Tim Adams (21:22):

Yeah… Sorry to cut you off, but exactly if the villain snap called, usually that’s leaning defensive, a little more weak. So yeah, those type of things make sense where maybe the range is just a tad bit weaker. But at the end of the day, even though it’s a limp pot, a lot of money has now gone in. So like I said, this small blind that close to full pot on the flop got raised a relatively big size and called even with casual players or fish or whatever you want to call them, usually they have okay hands, I would imagine. But yes, the snap call leans me to think it’s a little more defensive.

Mike Brady (21:58):

Moving on to the turn, we get the Ten of diamonds, pretty beautiful cards. So we pick up a flush draw to go along with our open-ended straight draw. The fish checks and Stu goes for a twenty k bet. It’s basically putting the opponent all-in. ChatGPT goes on to say, “Then once the diamond arrives, your twenty k sizing is brutal and correct. A lot of players check because “we got there”, but against fish, big turn pressure prints because they hate bluff catching, they fear flushes disproportionately and they call flop two wide, very nice pressure application”. So before I toss it over to you for real analysis as … Tim is actually cringing at what he just heard me say. This is another interesting instance of the LLM not understanding what is happening holistically. So it’s simultaneously saying that we’re applying pressure, we’re going to get him to fold a lot, but he’s also saying they fear flushes disproportionately and “we got there”.

(22:56):

So the LLM is not able to distinguish between what we’ve turned in reality, which is a flush draw and a flush. Again, they don’t know the hand rankings even. It is just generating text.That’s why you got to be careful trusting these LLMs with important information. Doug did a really funny series about this on his channel. A bunch of LLMs played each other in this big heads up bracket tournament thing. It was put on by Kaggle. It’s like a division of Google. If you haven’t checked out those videos, definitely check them out there. They’re quite fun, but Doug is able to query the LLMs and ask like, “Hey, why did you make this play?” And there was one hand in particular, I might get the cards a litle bit wrong. Sorry for this tangent everyone, but I promise it’s arguably worth it. The board is like Jack Nine Three with two clubs, one diamond.

(23:42):

The LLMs get it in. It’s like ChatGPT’s bot versus like the Deep Seek bot or something and they get in with Ace Ten with the Ace of Clubs versus Ace King of Diamonds. So it’s two backdoor flush draws. One of them says, “I raised and gutted in because I had the nut flush draw.” Not true, backdoor nut flush draw. The other one said, “I raised and got it in because I had the nut flush.” He did not. It had either three or four to the suit. So that’s another example of it literally does not understand even what you have. Tim, now that we got the bots analysis out of the way, what do you think of this play?

Tim Adams (24:19):

Yeah, I think you can go either way. You can obviously check and realize your equity or I mean, there’s a big upside to jamming, especially if you think the player you’re up against is someone who’s going to fast play the flop with their King x. If you did get defensive vibes on the flop and you think they’re more centered around some Six x, some Five x that you can now get them off of, obviously jamming makes a lot of sense. There’s always, with this type of player you’re up against, there’s always random misfactors, right? In one direction they may do something crazy like fold King Two or something, right? Or they may do something the other direction, like they may call something crazy like some naked Six x, but either way it goes okay since, well, if you get a King to fold magically, that’s great. If some Six x calls you, well, you’re going to have a bunch of equity with your flush draws, straight draw plus maybe your Eight and your Seven and have live cards.

(25:11):

So you’re going to have a bunch of equity. So I do like the jam versus this player type, yes.

Mike Brady (25:16):

Nice. Well, it did work out. The fish got out of the way. We win the pot with Eight high and we can move on to the next one. Hey, real quick, if you’re getting value from this video, do us a quick favor, click that like button, subscribe, rate the podcast five stars. Any positive signals you can send to our algorithmic overlords, we’d really appreciate it. It allows us to justify continuing to make valuable content like this to help your poker game. Thanks. Moving on, I believe this is the same tournament. The blinds are now six hundred twelve hundred. Our hero raises from middle position with Ace Jack offsuit to twenty-five hundred off of a thirty-five k chip stack, the small blind and the big blind call. Anything to say here on the preflop play, Tim, seems pretty standard.

Tim Adams (25:58):

Very standard. Again, MP, I assume that means hijack, but again, these things matter. So I would always go, for ease, I always call under the gun eight-handed, under the gun eight, seven-handed, well, under the gun seven, lowjack, hijack, cutoff, button. That’s how I personally think of it. And the positions are just so important to get right. So I assume our hero in this hand means MP is hijack.

Mike Brady (26:27):

Yeah, something like that. It could be lojack as well. I kind of hear it used interchangeably. ChatGPT agrees, just said preflop is a standard open, so it finally got something right there. We did forget to grade it on the last one. Let’s just give it, I don’t know, a C-. We can move on though. Flop King Jack Four rainbow, both the small blind and the big blind check and Stu goes for a two thousand chip continuation bet. Very small bet, something like ten k in the pot. So he’s going about twenty percent and only the small blind calls. Here’s what the bot had to say. Into two players, this two k c-bet bet is good. The small sizing fits your range advantage and protects your checking range. Not sure how a bet can protect the checking range. When the small blind calls, his range is heavily King x, Jack x, stubborn pocket pairs, and occasional floats.

(27:18):

So kind of two things to jump onto there, Tim. First of all, what do you think about Stu’s play going for this small c-bet with this pretty solidly medium strength to maybe slightly strong hand middle pair top kicker. And do you agree with sort of ChatGPTs ranging here, King x, Jack x, stubborn pocket pairs, occasional floats?

Tim Adams (27:40):

Yeah. I mean, these three-way spots can be tricky. This type of board, well, it should hit the small blind quite hard since they’re probably pretty heavy with offsuit broadway flats. So I don’t think this is like a pure bet board for the IP player like it would be say in a single-raise pot if you raise first in and the big blind called. So you clearly have a middling hand with Ace Jack. To be honest, this hang can go either way, check or bet at some frequency with a little bit of a preference to check. But I do like the flop size. You don’t need to bet overly big just to target a bunch of pocket pairs. It is quite a good board for you anyways. So like I said, you can really go either way and that’s going to be the case in a lot of spots

Mike Brady (28:27):

Yeah, that’s poker for you, especially when you have these kind of hands that clearly this hand middle pair top kicker is worth twenty percent of the pot at this point, but it’s also kind of middling so it kind of wants to check. So it just makes sense. Betting small and checking are very similar actions so it makes sense that can kind of go other way and like this. We get the Nine of diamonds on the turn. We’re heads up here. So it’s King Jack Four Nine. Queen Ten is kind of the obvious thing that changes on the turn and gets there for the nut straight and the action goes check check. ChatGPT likes the check, says you’re not getting three streets from worse often and you keep his weaker Jack x and bluffs alive. Do you like this check on the turn? Would you maybe try to eke out a bet here to check the river?

(29:08):

How are you approaching it?

Tim Adams (29:09):

Yeah. I mean, I think this hand predominantly checks. I wouldn’t be surprised. Again, coming from like a solver background where there may be some mergie bets here where you can possibly target some King Seven suited and get a hand like Queen Jack, Jack Ten to call you again. So betting does have some upside, but yeah, I would predominantly check this.

Mike Brady (29:31):

Stu got that one right and we go to the river, which completes the board with a Five. So the board is finally King Jack Four Nine Five, no flush draws at any point and the small blind bets out for four thousand. Fairly small bet. It’s like four thousand into something like twelve thousand. So we’re getting a really, really good price. Stu decides to pay it off. ChatGPT says this is “a key spot”. I got to say, I disagree, facing twenty-five percent bets on the river it’s not going to make or break your poker career. And it says population massively under bluffs (Says the word river donks, it’s using a term wrong) especially in this line “Honestly, I think this is probably a fold exploitatively.” What do you think about this one?

Tim Adams (30:16):

Yeah, I mean, I get it. ChatGPT has some okay points. Again, coming from an equilibrium background, you would have to call Ace Jack because like you said, it is only a small bet. It’s not impossible that you can chop versus Ace Jack yourself or you could potentially beat Queen Jack. Not that much money has gone into the pot. So yeah, facing, what is it like a third pot? Yeah, I think Ace Jack is just a clear call. I get it exploitatively. I see some medium mid stakes players or lower stakes players kind of go down this slippery slope here where when they’re up against a milky size that they don’t want to pay it. And I get it because it feels like the general player pool is kind of only using that size as a milky value bet. But in theory, they’re supposed to have it with the small size.

(31:09):

If they had fifty value combos, they only need roughly ten bluffs or something to create this. So you’re supposed to call and lose a bunch. So yeah, you just kind of create this bias where you may think, “oh, I always call and they have it”, but they’re kind of just supposed to have it. That’s the way a value to bluff ratio works. And it can be, again, like a slippery slope if you’re going to go down this explode route of trying to own people by overfolding because there is always a randomness factor that your opponent may be up to. You don’t always know. So oftentimes when I’m the spot and I feel like I’m getting milked or whatever and I call and I win, I try to remind myself not to have that bias because there is a randomness factor. There is this confirmation bias thing where you call and lose so often that you’re like, “Oh, I should have got away from it.

(31:56):

It’s so obvious.” But oftentimes you just call and you win. So I would always play, especially for these small bets, I don’t like doing anything fancy. I kind of just pay. You don’t have to be right all that often. And like I said, it’s not impossible that you beat value. So yes, I think our hero played the hand very well and I give him a high grade.

Mike Brady (32:16):

Yeah, agreed. I think the way you “exploit” someone who you think is maybe too value-heavy here. And again, to your point, they’re supposed to be mostly value-heavy here. Their bluff to value ratios got to be what, seventy-five or eighty to twenty, something like that. Eighty, twenty percent, that is eighty percent in favor of value. So you’re supposed to lose a lot, but that’s okay. It can still be a profitable call. That is tricky to kind of cement in your understanding as a human.You want to win more often than you lose, that’s just not the mechanics of this spot. But I think the way you exploit this player then is you fold like Ten Nine, because that’s a hand that probably pays the four k as well in Solver Land, let’s say. But if you think the guy’s just got a Jack or a King too often, then you fold the Ten Nine, you fold your turned third pair and don’t lose too much sleep.

(33:02):

But I think if you’re folding Ace Jack, I mean, yeah, you beat some value, you beat Queen Jack, you beat Jack Ten, a Nine… I think something like Ten Nine or Queen Nine could definitely be in there for this player. And yeah, you’ll occasionally see a guy show up with pocket Twos or something weird, especially when you’re playing live. So you don’t need that stuff to happen that often to make this call profitable.

Tim Adams (33:24):

Right. And with the Ten Nine, you can definitely get aggressive as well, especially if you think it’s going to overperform by just putting in a river raise, like jamming the river goes quite well with that kind of combo, especially if you have the right suits where you block some King Nine, Jack Nine suited. Those are the type of things I’m thinking about. And especially I’m trying to target that player type who I think is being milky and then really, really values their tournament life and doesn’t want a hero call. So yeah, those are the type of things I’m thinking about.

Mike Brady (33:53):

Yeah. So overall ChatGPT mostly a little too monsters under the bed, I would say on this one. I would go for maybe another D+ for this one.

Tim Adams (34:01):

Right. I would agree. And they kind of have the population tendency thing. They have the idea, but they miss the point a little bit, I think, but not the worst breakdown from ChatGPT. I will give credit.

Mike Brady (34:17):

Moving on. Five hundred a thousand blinds in this next hand. Really short one here. It’s just a preflop spot. The small blind who’s an active non-studied player, that’s our hero’s read. He limps in from the small blind and says, “You can raise if you want”, as Stu is thinking. Stu snap raises after hearing that with pocket fours to twenty-five k, they’re twenty-eight effective. So he’s basically jamming for twenty-eight blinds and the small blind quickly folds. I’m just going to read all of ChatGPT’s analysis here for the most part and then we’ll go from there. This is hilarious and honestly probably good. You had twenty-eight BB effective. He verbally invited aggression. That usually means fake strength, discomfort, or recreational needling. Interesting one there. The key point, most non-studied live players limp fold way too much, blind versus blind. I think that’s undoubtedly true. So your shove leverages maximum fold equity, huge pressure on wheat capped ranges and his likely inability to call off correctly.

(35:16):

Here’s my favorite part of ChatGPT’s analysis. Would the sover jam pocket Fours for twenty-eight big blinds over a limp? Almost certainly not. Would I hate this exploit against a weak recreational player telegraphing uncertainty? Not at all. This is a very live poker hand. The snap matters. Snap shoving projects confidence and removes information leakage. Tank jamming there is much worse. So before I kick it over to you, Tim, here’s kind of my thoughts on this one and why I… So Stu sent me like twenty of these hands. This is a short one, but it really jumped out at me because I thought it was interesting that ChatGPT got arguably the more obvious thing, but maybe the more important thing about being a poker player correct. It got the really obvious population read, right? Which the guy said, “You can raise if you want. ” He limped, he’s a live player.

(36:05):

If we use our logical reasoning here, I would bet ninety-nine percent of the time he’s not playing a game. He didn’t limp with Aces and this is like a really… He’s playing 4D chess to try to get us to jam. He actually limped because he has like Ten Six suited and then he’s looking at us and he’s not feeling great and he’s like, “You can raise if you want.” He’s telling the truth when he says that and ChatGPT correctly gets that population read right. I gather you would largely agree with that bit.

Tim Adams (36:30):

I agree with that. Yeah.

Mike Brady (36:31):

But then we move on to the part where it just totally takes a blindfolded shot in the dark about what a solver would do here with Fours. So it says, “Would a solver jam pocket fours twenty-eight blinds over a limp? Almost certainly not.” I just absolutely love that it pure guesses there. And Tim, you can give the answer. Would a sover ever jam Fours for let’s say twenty-eight or maybe thirty big blinds over a limp?

Tim Adams (36:58):

Yeah. I mean, the small pocket pairs do rip over the limp. It’s like Deuces through Fives and some offsuit Ace x builds the range, but yeah, it is actually the higher EV players to rip. So the chatbot is just talking out of their ass a little bit. The chatbot is like a bit of this friend where they might know a lot of stuff, but they’re just way overconfident it sounds like.

Mike Brady (37:20):

Right, exactly. Exactly. And again, this video is kind of a cautionary tale on using these LLMs for important stuff. We’re looking at this spot. We have a screenshot if you’re watching the video version of how a solver plays the spot for chips, ChipEV sim. And it’s a thirty big blind sim, pocket Fours, pocket Threes, pocket Twos, one hundred percent jam over the limp. It’s just not close. The EV is just higher, not close. Pocket Fives get in there like half the time too, to Tim’s point about those. But ChatGPT is just, again, it’s generating text, it’s guessing and giving pretty poor advice about that. This is a clear, clear hand to jam. You want to take down this pot preflop. Your hand doesn’t play super well in the single-raised pot line or the limped line because it’s often going to be an underpair. You just perform much better just winning the pot preflop the vast majority of the time.

Tim Adams (38:09):

I’m with you a hundred percent and also with your analysis with ChatGPT. ChatGPT is kind of like a feel player, not really a solver player it seems like.

Mike Brady (38:17):

Yeah. Maybe we should get them in there, see how it performs. Moving on to our fifth and final hand, the blinds are six hundred twelve hundred and the under the gun plus one player raises to twenty-eight hundred. The button calls and the small blind, our hero, calls with King Ten of diamonds and the big blind gets out of the way. ChatGPT says about the preflop play with small blind versus under the gun open and button call with twenty-seven big blinds, calling is fine. Your hand might be dominated, but a suited broadway with that price is playable. What do you think about this one, Tim?

Tim Adams (38:54):

Yeah, I think this combo always wants to jam pre. I mean, we talk about it quite a lot in the course, but these suit of broadways, they’re really aggressive to squeeze jam all-in. Exactly what the bot was just saying, whereas it is often dominated pre by say Ace Ten off, King Queen off and by ripping we can get those hands to fold. So we’re folding out dominating hands. We get to realize all of our equity when we do jam. And being in the small line here multiway, we’re going to obviously have quite bad … Well, we’re going to just negate our positional disadvantage by just ripping and just potentially winning the pot, preflop. And if we do get called, we’ll have okay equity. There’s what, like eight k in the pot. So it’s like a significant part of our stack if we just win it pre.

(39:37):

So yeah, I am not a fan of the overcall. I would like to just rip this pre.

Mike Brady (39:41):

Cool. Yeah. I find myself making this play a lot more when it’s like hijack or cutoff opens, button calls and then my eyes light up because they have so many of those offsuit hands that dominate me that I’m really targeting and getting them to fold preflop. Under the gun one, I’m surprised to hear you say you still like the rip and it’s good to know for my own game because that’s a range that I would be a litle bit more concerned about, but maybe I’m a litle bit too much monsters under the bed.

Tim Adams (40:08):

Yeah. I mean, again, we don’t know if this is under the gun eight or nine or lowjack, but I’m pretty sure on this depth, this is the type of hand class that would always rip, again, getting dominating combos to fold. Even if this U8, they’re still opening King Jack off, King Queen off, Ace Ten off, all hands that are folding versus your jam, they’re folding and then the suited versions, King Jack suited is folding King Queen suited. So yeah, there’s just a huge benefits of ripping this hand class and the button is calling too, right? So the button is going to be relatively wide. They’re going to have a bunch of the pairs. They’re going to have still a bunch of offsuit hands that again dominate you. So you’re just targeting two ranges and it’s not like you have some shitty hand, you have a hand that wants to realize all its equity.

(40:53):

And when you go post, you’re supposed to not realize equity really well as the person in the worst position. So I really, really want a jam here for our hero. Yeah.

Mike Brady (41:04):

And I don’t want to spend too much more time on preflop, but one more kind of … I mean, a theme in this video has been giving all of the details needed to analyze a hand properly. And I think one thing that would matter quite a bit here is how deep under the gun one and button were, because you’ll see in position versus a raise in tournaments some surprisingly strong hands get in there for just calls when you’re like fifty blinds deep. I see this a lot. I’m really surprised by how often like Tens, Jacks or even Queens will just call a raise at fifty big blinds. And I think it’s trying to do two things. You can correct me if I’m wrong, but A, it feels like it doesn’t want to get stacked by like a better hand every single time and by three betting, you just basically guarantee you’re going to get stacked if they have a better hand.

(41:46):

You also allow stuff behind to happen if you just call. So the guy calls on the button here with two Jacks, we jam from the small blind with King Ten suited, under the gun gets out of the way and now he gets like a fist pump call with Jack. So just like some really nice things happen and it also just plays really well in a single-raised pot in position. It’s a really nice hand to have. It’s going to make a lot of strong overpairs. How much would that really play a role here? If under the gun plus one was sixty blinds deep, button was sixty blinds deep, are you more or less likely to make that kind of power jam with King Ten suited?

Tim Adams (42:18):

Yeah, it’s a loaded question. So if it was like the U7 player versus the lojack, you’re going to mostly want to have a pretty aggressive three betting strategy just because there are like a lot of hands behind you. As the button, you’re just going to be a little more polar. Like you’re kind of alluding to anyways, getting in sixty blinds like U7 versus button with Jacks or Queens, it can be a little dicey. You can kind of go either way, especially with Jacks. So yes, you’re making a great point that the effective stacks of the other two players is really, really important, but still when we think of a range and how it’s built and how many combos are in there, yes, if you rip King Ten suited here and you get called by Queens or Jacks, that sucks, but proportionally it doesn’t just make up a ton of the range.

(43:07):

They will still have Deuces and Ace Four suited and just getting anything to fold is great. So yeah, in proportions, I still think I understand where you’re coming from that is a part of it, but the upside of ripping and getting all those other hands to fold is going to be worth it. So even if it was like a sixty BB open, six BB call and you’re on thirty in the small blind, I still think it’s a rip

(43:32):

with this particular hand. Yeah.

Mike Brady (43:32):

Makes sense. Yeah. Those eight combos of Jacks and Queens not deterring you, which makes sense.

Tim Adams (43:37):

Exactly. They’re mixing too, right? Like you said, eight combos, not a full twelve. So those are things to think about.

Mike Brady (43:47):

Right. Makes sense. All right. Well, with all of that said on the flop or preflop rather, let’s move on to the flop and get the Ten Nine Four with two clubs. So we flop top pair. Checks to the button. So that’s the player who called preflop who goes thirty-five hundred. That’s something like a third pot. Stu decides to just call here and the under the gun player gets out of the way. ChatGPT says the check call is good. You have a top hair and decent kicker, multiway, under the gun still uncap, so no need to blast. I don’t know if it’s so trivial. Do you ever consider a check-raise here?

Tim Adams (44:17):

Yeah, quite a lot. There’s just so much in the pot now. You don’t have that much back. Yes, it can suck getting it in versus a better hand, but a lot of the times you just get it in and both players fold. Just like denying equity. Sometimes you get it in versus a worse Ten. So yeah, I think calling is okay, but I’m leaning to raise pretty frequently. It is a really dynamic board. You are in a positional disadvantage, so much in the pot. I think I lean raise, but I think it’s close in EV.

Mike Brady (44:52):

So Stu does just call and we get the Six of clubs on the turn that makes the board Ten Nine Four, Six now with three clubs and the action goes check, check. Pretty clear check here. I don’t think we want to do any leading and ChatGPT said the same thing. And when the button checks back, he often has medium, one pair showdown value hand like Ten Nine, Ten x and Nine x or maybe some hands with one club that didn’t want to get blown off their equity. Do you agree with the ranging? I mean, obviously we’re going to check to the better in this situation. I assume there’s not really much of a reason to lead or maybe you disagree.

Tim Adams (45:25):

I think there could actually be some sneaky leads on the flush completing turn. Eight seven gets there. Again, the way this hand works is the button, they’re going to be pretty stab happy with a lot of their range. Pocket pairs just betting for kind of this, well, bit of protection, just a lot of small betting as the IP player. And when we call, all of a sudden our range is relatively strong, maybe some Ten x, some strong draws or some decently strong draws. And I think we leverage this situation by often playing a small block on the turn. I could be wrong about this. I can’t say that I’m an expert in this particular three-way pot where it’s like EP opens button calls and small blind calls, but I would not be shocked if you built a sim for this and there would be some small blocking on the turn on the Six of Spades from the small blind.

(46:15):

So I think that is an option for you.

Mike Brady (46:17):

Yeah, makes sense. Six of clubs, by the way, did complete that flush.

Tim Adams (46:19):

Sorry. Yeah, I always think in spades for some reason when I think of flush…

Mike Brady (46:23):

Of course. Spade dominance, classic.

(46:27):

All right. So we do go check, check, and the river is the Jack of heart. So the final board, Ten Nine Four Six Jack with three clubs. Stu goes for a very small bet. He goes like sixteen hundred into about eighteen k. So about a ten percent block bet and he ends up getting raised. He gets raised to ten thousand, which is something like a half pot raise. So not a huge raise by any means. ChatGPT says a lot here and if you’re watching the video version, you can read all of it, but the gist of what it says is very results oriented, frankly. You go sixteen hundred, it’s so small, it can induce exactly what happened. It looks like what your hand is, a hand that doesn’t want a good hand that doesn’t want to face a real bet. Against many players, tiny block bets work. Against anyone with aggression, it gives them a cheap, obvious attack point.

(47:13):

Kind of taking a step back, what do you think about blocking this particular hand? Do you think we’re targeting enough worse hands to block with this hand?

Tim Adams (47:20):

Yeah. I mean, the Jack is a bit dicey. I think it’s fine. It’s kind of like exactly what your hand is worth. Really between checking and playing B ten, both seem fine. Exploitatively, it’s kind of a little bit what ChatGPT is saying, that if you are up against a certain player type who’s a little more passive and likes calling, then this is especially good versus someone with tricks and aggression, this gets a little more worrisome, especially if you’re not prepared to make some difficult decisions like calling or back raising. So I think ChatGPT has at least an okay idea, but in theory, checking or playing B ten seems completely fine and I would have a preference versus certain player types.

Mike Brady (48:02):

The results the button did take this one down. Stu got out of the way and the button showed the two of clubs two of spades. So we had that somewhat relevant two of clubs blocker blocking some of the flushes and our hero gave him a nice fist bump for that one well deserved for showing the heart. And ChatGPT loved it too. “The fist bump was classy.” Thanks bot.

Tim Adams (48:25):

ChatGPT is like a, I don’t even know. What would you describe ChatGPT as like a two five grinder who just has played for forty years or thirty years? It seems like that, right? And has never looked at a preflop chart? That’s what it feels like.

Mike Brady (48:37):

Yeah, maybe two five is even a litle bit of a stretch. I’m not sure. Okay.

Tim Adams (48:40):

Okay.

Mike Brady (48:41):

Give him one three.

Tim Adams (48:42):

Okay, one three.

Mike Brady (48:44):

All right. Thanks a ton for joining us, Tim. Hopefully you enjoyed that really high level analysis mixed in with the extremely low level bot analysis. Make sure to pre-register for Modern Tournament Mastery coming this Monday on upswingpoker.com. If you get that course next week during launch week, you also get Tim’s bonus course Super High Roller Mechanics as a free bonus that’s going to cost three hundred dollars on the weeks that follow. So make sure you jump on that course early. If you want to see more from Tim, check out this video he put out a couple of weeks ago and with all that said, thanks again for watching.