Internal Market Making & Simulation Elitism w/ Louie from NoVig | Ep 135

Even if you take a point estimate and wrap a distribution around it, you're still doing a lot of post-talk work to try and turn one output into many markets. With a play-by-play sim, everything comes from the same underlying process. Mainlines, derivatives, player props, SGPs, they're all being priced off one coherent simulation. All right, what's up everybody, GPSP, and making his podcast debut with us, Lui from NoVig. Welcome to the show, and thank you very much for doing your first pod with us. Yeah, thank you guys for having me, listen to every episode, so happy to make this one be my debut. We're glad, and this is what we're always looking for, is people who are interesting that people want to hear from, and haven't heard from yet, to come on the pod. It's kind of like the sweet spot. Not that it's not great to have Shipper do his 13th podcast with us, but it's always good to hear from someone new as well. Yeah, and it's like evidence that the show's working well, is that we attract people like Lui who have not been on any other show, it's interesting enough to get those people on. Lui, why don't we start with anybody who's not familiar with you, how you got started with in the betting industry, I don't know all too much about you, I think you started at Fandall, is that correct? That is correct. So walk me through, was that right out of school, and then maybe just talk a little bit about what types of things you were doing at Fandall and what that looked like early on for you? Yeah, sure. So unlike a lot of people in this space, I don't really have much of a gambling impulse. I'm actually pretty allergic to knowingly playing any type of minus EV games. I mean, I've been to casinos plenty of times, but basically only to use the sportsbook kiosks. I came into this more from a math and systems mindset. I went to college, planning to study business and computer science, with the idea that I'd end up in traditional quant finance or something like that. And toward the end of my freshman year, my best friend from high school called me and told me he was booking bets at Michigan. And the money he was making was absolutely eye-opening, but what really stuck out to me was that he had one player who wasn't winning every week, but also wasn't losing quite like everyone else. Over a couple months, the player was positive, but they were betting the same live tennis markets over and over. He basically asked me, "Hey, can you look into why this is happening?" At the time, I genuinely believe sportsbooks were like these perfect pricing machines, and that winning long-term just wasn't really possible and assumed he was just on a heater, but once I started digging into it, I realized I had it completely wrong. Games are trying to perfectly predict games, they're mostly trying to manage risk efficiently. And once that clicked, I was kind of hooked because I realized this meant they were beatable at certain spots. So I started scraping odds immediately, building really simple point estimate models, and just looking for spots where prices moved slower than the game itself. And a couple of years later, through another friend who was also booking, I got my first paperhead account and started betting into those lives, but nothing fancy, just like being faster where their software was slow to react. And after I graduated, I spent about six months or so doing that full time. It paid all my bills, not that I had that many as a fresh college grad, but more importantly, it taught me how fragile edges really are. So eventually, my friend cut me off, and not as a friend, but from being on his book. And not long after that, I was in a pretty serious accident that completely told my car. I barrel rolled three times over off the highway, and it was really a mess. I walked away completely fine and unscathed, which was very fortunate. But it forced a real reset in terms of what I wanted to do long term with my life. And that's when I decided I wanted to understand this business from the inside out at scale. So a few days after the accident, I started applying to sportsbook trading roles, I landed at Fandall a few weeks after that, and the rest has sort of been history. Well, I mean, that's a crazy intro to get to Fandall, but it's interesting is I feel like Fandall is like the best place to land. I know this was a couple of years ago, right? So like three and a half years ago, or something like that. So I'm thinking of now, if you're like, at least of the traditional rec books, Fandall's kind of like the one I think that gets a lot of respect, and people would want to work there. At the time, was that like your first choice? Yeah, it was certainly my first choice, but I mean, I applied to a few other books at the same time, probably most notably was Bet365, and Fandall was very quick to move along in the hiring process. Granted, it was like five rounds, you know, trader take home tests, all these things. And I was on the fifth round when Bet365 first reached out about my application. And you know, so I told the Bet365 representative, I was like, you know, just in the spirit of your transparency, like I'm also applying to Fandall, I'm on their final round. I don't know how long this process might take with you guys, but you know, if I don't want to stop someone else from maybe getting a fair look, if I end up getting this job within the next few days, you know, obviously going to pull out of the process here. And kind of to my surprise, because like I totally could have just been making that up on our phone call, the 365 rep was like, oh, you're in the final process with Fandall. Like, we should just hire you, like why don't you come to the office tomorrow? So I was like, okay, I mean, I'll come check it out. And you know, long story short, in talking to me, the guy seemed to think that I knew, I knew enough about what I was doing. And they offered me the role on the spot when I went to the office that day. And I told them I was like, you know, respectfully, I'm going to like wait and see how the Fandall process goes as well. And this kind of worked out really well for me, where, you know, I was able to sort of pit the two offers against each other and have them kind of negotiate against themselves and raise the starting salary and you know, bonus and whatever. And it got to a point where after going back like three times, the 365 rep, who I'm actually still friendly with today and is, you know, have a good relationship with, he's like their head trader over there, he told me he's like, you know, off the record. And if we were to match this offer, I think it'd be better for your career to go work at Fandall. And like that was just like a kind of blew me away that like someone would negotiate against themselves like that. But you know, I respected him for saying that. And like you said, I thought Fandall was, you know, the clear, the clear place to go just off of like name recognition. Like if I went home and told my mom, like, you know, I got a job at Fandall as a trader. Like she sees the commercials like, if I said I got bet 365, she'd be like, like what the fuck are you talking about? Like, is that a real company? Well, it's funny that bet 365 was top down hiring. They just, you know, they just chase Fandall as the sharp side and saying, if this person made it through this many rounds at Fandall, it must be good. So I want the next question I want to ask is like, what, what did actually look like at Fandall? One of the things, we have some questions on this later on from some listeners and whatnot. But I think it's really interesting for people who've only ever been like on the betting side to hear about any, anything that can be shared from your experience as a Fandall and really like what the day to day look like for you. But maybe before that, you know, what did, what did, what are, what, you mentioned a trader take home test? Like what, what are they doing? So what is a trader take home test for these companies for entry level traders? So well, at the time at Fandall, for some context, when I first started at Fandall globally, because you know, the Flutter organization has a ton of brands and the traders all control the odds across every platform. So there's not like a, you know, the Fandall traders are moving every Flutter price. And at the time when I started, we were 30 traders globally, which was a Jersey City, a Dublin and a Melbourne location. And by my last day there, we were 130 traders. So when we were just 30 traders when I first started, there wasn't really a distinction between the role you were just hired as a trader. There wasn't really like an entry level trader, whereas now they certainly have like a junior trader position and things of the sort. But to answer your question about what the take home test looked like, there's a few variations of them. And you know, over the years that I worked there, I got to see the other ones that I didn't take, just as new applicants were applying. And they're typically like a 10 to 20 question test. Obviously, the 10 question one is some of them are a bit more involved in, you know, the 20 question one has a few more like quick hitting questions. But more or less, they're like simple like game theory and like EV type questions. Things like, you know, if you if you roll a dice, you know, 10 times like what's the EV on the payout or things like that, or they might ask, they'll give you an American odds, a three way money line and say, can you tell me what the not the no push, just the two way market should be, you know, you have to devigate and whatever. And something else that they were really keen on at Fandall, because Flutter is a European brand. All of the backend systems for Fandall use decimal odds horrifying. Right. Well, what I'm about to say next in today's exchange environment makes a lot more sense. But you know, three and a half, four years ago, I got a lot of shit for it was. I kept all of my odd screens in probability. Everything was always in percentage. And that was just because like I was, you know, building my own models, so I was working with percentages anyway. I just didn't want to convert them. And I remember pretty distinctly my first day on the job at Fandall, the trader who was supposed to like skill me up and onboard me was sitting next to me. And he's like, oh, can you open your odd screen? So I think I've changed the tab. And he looked at the percentages and he was like, what the fuck is wrong with you? Like, you were absolutely a psychopath. And I was like, what are you talking about? I was like, am I supposed to use decimal? And he goes, oh no, God forbid, no, like we obviously use American odds. And I was like, well, our tooling is decimal odds. And he's like, yeah, well, I mean, we could just do that conversion. And I was like, well, can I just convert the percentage to decimal the same way? And he's like, that's, that's actually a psychopath. Like I've never seen anyone use percentage. So a big part of like the take home test was also about like just general odds conversion. And for all of the answers, I wrote all my answers in probability. And I remember like after I got hired, like, he obviously seen my odd screen. It was me and one other trader who got hired in that round. And they were making jokes being like, oh, well, because they didn't see whose name was on the test when they were reviewing it outside the hiring manager. And they were like, oh, well, we know whose test this was like, we've never seen someone just work in probability. That was kind of like a fun story about the take home test. So you get to Fandall, everyone knows you're crazy because you have your odd screen, something that no one's ever done before. What is it? You know, where do you start? What are you doing? Like, I think one thing that most people on the show didn't, haven't seen the other side of the counter. And a lot of the time we're guessing as to what's going on over there, what you know, you as you're doing. So, you know, what, walk us through maybe like what you were responsible for, you know, how you're moving the numbers, how you're setting openers, et cetera, just to get people inside into, all right, yeah, who their counterpart is. Sure. So the day-to-day at Fandall was actually pretty straightforward. In any given sport, there were usually four or five traders working at once and the slate would get split up evenly. So if there were eight NBA games and four basketball traders on shift, you'd each be responsible for two games and you'd lock into those as hard as you could. Most of the job was reacting, you're watching straight bets come in in real time through the bet feed, keeping one eye on, you know, sharp books like Penny and Chris and another eye on a handful of tagged, sharp customers that you respect. You're not really ever trying to originate anything yourself. You're really just, you're managing price discovery as the event start time gets closer and closer. If you were scheduled to trade a game live, the day naturally split into, you know, pre-match and in-play responsibilities, pre-match was a lot of profiling, watching news alerts, you know, Twitter, injury stuff, just trying to absorb as much information as possible. So once the game went live, it was typically one trader per game and your entire job for the next, you know, two and a half hours was to keep the odd screen bet feed and whatever low latency data provider we were using open and react as fast as possible to sharp flow and market moves. Inplayed profiling was pretty limited outside of stoppages, but it was certainly encouraged and extremely high ROI for the company. The traders that could manage profiling and moving the odds in play, outside of live trading, every trader also spent time, you know, practicing openers for main lines and player props in their sport. We'd all send in our numbers for the next day, slay in a slack chat and the opening lines basically just awaited average. ADR and, you know, Benson have alluded to this before, but there's really nothing magical about it. traders aren't spending time to post the most correct line, they're trying to post the least incorrect one as fast as possible and start writing bets because you can always just take a click and move the number. One thing people underestimate is how important communication was in the role. I'm not the best communicator, but we kept shared notes on why prices moved, which customers bet what and at what price compared to market and whether we received the click first or about steam chased into. That was a pretty key distinction that we had to make. A lot of that was to make handovers, smoother between Jersey City, Dublin and Melbourne teams, just for, you know, fluid 24/7 coverage, but for the more technical traders, there's maybe three of us. There was also time carved out for modeling or programming projects, in my case, a big one was building infrastructure to automatically reprice lead futures off a season sim. So instead of manually nudging a dozen different markets, one sharp bet could auto move the wind total division price, playoffs, championships, all in one shot. Overall, the role was way less about, you know, setting smart lines and way more about getting to the true price as quickly and as cheaply as possible and then writing as many bets as you could. There's a bunch of different aspects I wanted to drill down on there. First part, so when you, you said you might be managing like two games as a trader, how are many traders you have, you know, and how are many games? Is that like all aspects of the game? So that's, you know, obviously the main markets, derivatives, props, is that everything? Yeah, that's everything. And, you know, like the books really aren't good at pricing median outcomes, especially early in the odd cycle. I mean, as you get closer to tip, you know, the market just sort of converges and that's not really to the credit of, like say a fandal or a draft games that they have the same line as penny, they're just copying it. But what these rec books do really well, and I think fandal in particular, is there really just one giant derivative calculator. So it'd be on the trader to manage like the median line inputs. So like the parent line, you'd be managing like the main spread, total money line, first half spread, first half total, first half money line, and then you'd be managing like the median projection for every player prop. And then the models that fandal had weren't models to try and like get what the spread or the total should be. They would be given these parent inputs, what's the 10 plus 11 plus, what's the, you know, what's the all-time given three extra points on the spread, like they're purely just like derivative calculators, but they're really fucking good at it. Which is why, you know, you don't often see people win betting into SGPs or things like that. Um, where that methodology obviously falls short is you, as a trading team, you get really comfortable with the fact that, oh, if my parent line looks close to market, I'm going to assume every bet on the derivative is a good bet for the book because you're just trusting the data science work. And that leaves you, you know, vulnerable to a ton of different things like the most egregious being if your main line got faked, now you're going to write essentially infinite bets on the alt line until something moves your main line again, because you have no sense of what that alt pricing should be. You're just trusting the sort of derivative calculations done by the data science team. Do you ever get to a point where you're moving the main line on the halts? Because like it was clear you're just getting, I mean, you just said that like if you're getting faked, right? So I could see that being a situation where you could be like, wow, we just moved this main on 5K action and we took like, you know, 50 K across the board and all, I mean, I don't know if that's real, it's a pretty realistic scenario. Yeah. And especially for a sport like basketball, um, you may or may not know this, but, um, like basketball line up. We know ball. We know ball. Basketball line ups are not static, but they're pretty consistent throughout the regular season where things like a quarter derivatives are largely based on the rotations that a team runs. Um, so if you have this, you know, grand derivative calculator, that's powering all of your alt lines, but, you know, just thinking back like a few years, like, if a, if a JoL and B is playing the entire first quarter, you need to price those alt differently because you know he's sitting for the first handful of minutes in the second quarter or someone like Shay or, you know, LeBron for a while would check out say at like the six minute mark and it's like all of these cannot just be priced off one static conversion tool given the parent main line. Um, so there were definitely cases where like some spots were like, I thought that myself and a few others would like shine as basketball traders was like having that domain knowledge to be able to be like, Oh, if the second quarter under is being hit, it doesn't necessarily mean that the full game is off. If you know, it's one of these scenarios where like the best player is playing the entire first quarter, um, but then also having the ability to recognize when, Hey, this, this second quarter is being bet and it actually is indicative that the full game line is actually off. Um, and like that's like a maybe like a trading 201 or 301 type of deal where the fresh traders usually just kind of skip over those things, but they're certainly like high ROI, um, moves that you can make as a trader for a book, not only in just moving the price, but, you know, profiling the customers that are doing that, um, yeah. So we, uh, honestly, we could probably do two hours on just Fandall and some of these questions, um, maybe I'll ask just a couple more, um, when, maybe we could start to move into some of your own betting. So like you're at Fandall, are you, are you betting your own stuff as well? Are you starting to, like, how does that, that process start with you? So when you work at a regulated sports book, there are pretty tight restrictions around where you can and can't bet. You're generally not allowed to bet at any shops that share data providers or anything like that. So while I was at Fandall, I actually didn't place a single bet, um, that was just the reality of the role and I was pretty fine with it at the time. Um, it would be pretty difficult to get down if you're only out for, you know, MGM and Caesars, uh, with real penalties where you, you lose your gaming license if you were playing offshore and, you know, there were no exchanges or sweepstakes at the time, um, gaming in the sweepstakes and exchange world now. That's obviously very different. We have a lot more freedom to place bets pretty much anywhere with the obvious exception of our own platform, you know, outside of internal market making, uh, we can get into more of what that looks like when we talk about Novig, but structurally, it's just a completely different environment. So what did it, what did it look like? Maybe transitioning out of Fandall, you, you, so how long were you at Fandall? And then what ultimately sort of, uh, led you to, to explore, um, I don't know if you went directly to Novig after that or there was a gap in between or sort of walked me through how you sort of transition off of Fandall and why you made that decision. Sure. So, uh, I was at Fandall for two and a half, maybe two and three quarters years. Um, there were definitely a few things that had me thinking about leaving Fandall, but to be clear, I didn't actually leave my choice. I was fired. Um, some people who have been around gambling Twitter for a while might remember me being pretty vocal about some frustrations I had with the Fandall trading team earlier this year, uh, as a better. We don't need to get into all that here or else, you know, straight to jail for me. Uh, obviously it, uh, it wasn't a clean plan transition. It was more of a hard stop and then figuring out what was next. In hindsight, it, it probably accelerated a transition that was going to happen anyway. And it's worked out really well for me, so I don't have many complaints, um, as far as going to NoVig. Yeah, so the first call I made after telling my girlfriend I had been fired was actually to flop. Uh, at the same time he was working, I think, in, uh, he was still in traditional finance and betting on his own, but we built a real relationship through some consulting work. Um, about a month earlier, we'd actually all met up me, Chris and Henry, kind of by accident, uh, a month before I had been fired, uh, Chris double booked dinner with both of us and figured we should probably just all go together rather than rescheduling either me or Henry. Um, I had DMed Henry on Twitter before, but actually on the back of, uh, him talking with GP and going the internal trading tools, um, but that was the first time Henry and I had really sat down and talked and that dinner ended up being pretty important in hindsight. So when I called Chris after getting fired, literally like less than an hour after being fired, I was still standing outside the fandal office. He basically said, you should talk to Henry, see if there's anything there with NoVig. So I reached out, we talked and pretty quickly things moved forward. I think, you know, within maybe 10 to 12 days of leaving fandal, I started my first day at NoVig on the trading team, which at the time was just myself and Henry and that really came down to just, you know, relationships and reputation, uh, because of how I operate as a trader and handle myself in conversation with, you know, Chris and Henry in the past, there was already a level of trust. So the transition was pretty smooth. You know, I need to miss a pay period. I stepped right into the role and, you know, Henry kind of just let me get to it. Did you consider betting your own stuff at that point? Yeah, I definitely considered it. And honestly, if your only goal is to extract EV in the short term, going the pure betting route makes a lot of sense from me though, it didn't feel like the highest upside path, you know, betting alone caps your learning rate pretty quickly. You're reacting to markets, but you don't really get to see why prices move, who moved them, how liquidity behaves or, you know, or where things actually break under size, uh, what appealed me about going to Novig was that I could still bet anywhere, and actually Henry encouraged me to bet anywhere. So I, you know, I could finally start betting into my former co-workers lines at Fandle, which are, you know, how those are, uh, but also still sit closer to, like, the plumbing of it all inside a trading team, um, and that kind of exposure just compounds way faster than grinding accounts. I also think purely betting can be pretty fragile, um, I mean, depending on your edge, edges come and go, accounts get limited, uh, regimes change and, like, what's allowed, what's not allowed. And being involved in building markets felt more durable and more aligned with how I think. So I mean, I thought about just betting, but for me, building and, and learning at a company I believed in and working with someone who I also had a lot of respect for and Henry was definitely the, the better long term bet, and, you know, that could start up equity never hurts either. Yeah, you've spun that up a bit. I'm sure, uh, after seeing some of the last raises and whatnot, so yeah, that's a, uh, that's a nice thing to have on, on the back, on the back burner too. So okay, let's talk no big. So I was kind of surprised to know that you, in Henry, it was kind of like a two-man shop at first. Like as far as the trading team goes, you guys were just around the clock, like, yep. Talk, I guess, like, what was the early days of, of, of trading at Novik, like, the early days of trading at Novik were not much trading at all. It was a lot of putting out fires, being available at two, three a.m. because a system crashed. It was a lot of, you know, people were using the platform and I guess that's sort of the beauty of an exchange is that the product could operate even if the internal trading team wasn't providing liquidity, you know, naturally that's a worse off user experience, but, and then we were still able to be live and test things as it happened. And really the first, maybe three or four months were really just about building things. And I give major credit to Henry, you know, for a ton of things, but for being willing to, like, bring me on board and really defer a lot to the trading experience I had from a Fandall and let that inform, you know, how we, how we try to profile counter parties without information, what the trading UI and tool system should look like, how they should react. But yeah, the, the first, maybe three or four months were really just building. It wasn't until June of this year that the trading team actually changed from me and Henry to adding two more and then recently within the past month and half adding Chris. And we're increasingly becoming more and more trading and then we were sort of trading software where we were largely just building for a long time and like now we're starting to actually, you know, get our trading in. But yeah, my role and responsibility at Novig is changed from day one to now, a ton and that's sort of the environment of a startup where things are very fluid and you wear many hats, but my role at Novig was extremely different from what I was doing at Fandall. You know, the biggest similarity is that on the side of the counter, you're almost always playing defense and as a better, you're trying to play offense, you're trying to, you know, snipe mistakes, get down as much as possible and maximize your P&L. On the trading side, whether it's a sports book or an exchange, the job is really about risk management across the slate. At Novig, the trading team's responsible now for, you know, creating and settling markets, making sure liquidity is consistently available throughout the day and then scaling that liquidity appropriately is events to get closer to start. But beyond that, a big part of my role is working closely with the trading engineers, like I said, and at first that was just Henry and that team is, the engineer and team has grown dramatically. But, you know, every day I'm writing code to improve the trading UI, ingest new data sources, pipe things like injury or underdog, Twitter alerts directly into our feeds and just generally writing software to make, quoting RFQs or straight trades as efficient as possible. And Henry has a line he likes to use, which is, if we all got hit by a bus tomorrow, could the system still run and people sometimes laugh at that and think it means that the traders are automating themselves out of their jobs, but that's not really the point. The goal is to automate the defensive side of trading as much as possible, you know, the boring but critical stuff so that the internal trading team can actually pick their spots to play offense like a better. Like the human judgment still matters, you know, deciding when to lean into risk or when to step back or how to intelligently engage with counterparties on the exchange. So I, you know, the concept of the internal market maker has been in the news, like I feel like several iterations, the gambling Twitter news, several iterations. So I think this will be probably a topic we stick on for a while, just because I don't feel like it's, it's talked too much about and it's certainly the people who are a part of one of those teams don't, don't talk a whole lot. So maybe, maybe just to tee that up a little bit more, like what, what do you view, like what is sort of the role of the internal market maker on Novig today? Like are you all still sort of like primarily in a phase of, you know, seeding markets, making sure there's liquidity, all of that is just not even the right dichotomy to think through the, the sort of roles and responsibilities or is it more of like a, like you said, like attacking phase, like, because I guess for someone who's, you know, not well informed on this topic, it'd be like, when your exchange is new, I would think there need to be much more focus on sort of like making sure everything's seeding, you're probably losing, like if you think of like cow's sheet trading or whatever, like probably losing, you're seeding these things nobody else wants to seed, but eventually, hopefully you get to, to scale that, that flip switches a little bit, like we're, we're on the, the sort of life cycle of that, are you all? I would say, you know, it depends, as a trading theme, we're definitely skewing more now towards that playing offense type of role. And I think from an outsider looking in, it probably makes sense of why we brought on someone like Flux, who's, you know, particularly known, well, depending on how well you know him, he's especially good at getting down and being ruthless about attacking a market, and just bringing in someone with that mindset helped sort of change the culture a bit from defensive trading to more playing offense with trading, but it's certainly not the role of every trader every day. Like, for example, when we're in NBA season, I'm focused not zero, but a very small amount of my attention on a daily basis is going to ensuring that there's liquidity in markets. And I'm primarily looking at where can I be profitable, like actually trading these, given that I modeled a sport, and you know, it's, it's my main thing. If it's something like, you know, the, the dog days of summer where you can only reasonably trade so much WMBA, then my focus might change to, you know, ensuring we actually have liquidity across the board doing some more special and niche markets or spending the day just working on trading software. So I think it depends on which trader you would be talking to, but the general shift is to be playing more offense now that the automated system is particularly good at playing defense. So if you guys were to go, I assume the goal is to get onto the CFTC, um, you would be essentially like a separate entity at that point, which would be interesting like it would be like a culture trading rate. And you know, I think there is some discourse around like what these entities should look like, um, should they be profit seeking or not, you know, as long as there is a Chinese wall, does it matter? What do you kind of thoughts on, you know, what you, if you were to split off under like the CFTC structure, or you, I guess you can just apply it to like what culture trading is now as far as like people from the outside looking in might be like, oh, it's kind of weird to have an exchange market make on their own exchange. Uh, what would you say to like those people or about that structure in general? So I think the being welled off is the key component is as long as you're welled off, you effectively are the same as, you know, any other large market maker on the platform. You're not getting any special information, where it can get dicey is if a spun off trading team of a platform is offered a rebate structure or any type of incentive program that would only be offered to them. Like you can, you can create some without making it exclusive to your spin off company or your spin off trading team. You can make some trading thresholds that are effectively impossible for a third party to qualify for knowing that your spin off trading team can and effectively route back, you know, rebates or any type of liquidity pool to them, which can help aid them in being profitable. Um, but I think I think the biggest thing is, as long as the trading team is welled off, it really should be a non issue. Um, as far as no big potentially, you know, doing something like that in the future, I don't think it would change too much about how we operate right now. Um, we don't have any information that an API user doesn't have at the moment. Um, so effectively, we'd be operating the same way. The only thing is instead of our API being, um, open on a profit share or a rev share agreement, it would just be open for all users. Um, so sure that might increase the competition, um, but effectively the way in which we would operate and price things and trade wouldn't really change. Well, you probably trade on other exchanges. I don't know if you do that now or if that's something you can talk about, but like if you are like a truly walled off entity, you could theoretically be on call share or whatnot. But it does like make me wonder, like if you have these exchanges and they all have their trading arm, they're not going to give the market making deals to the trading harms of the other exchanges, right? So it's like, it's going to be an interesting, it's going to be interesting. But to that point, um, not getting the same market deals as the trading arm, you know, it doesn't prevent someone like you from market making a shit load of volume on call sheet. No, I agree. And that's, well, I hope to think like it is one size fits all for that, that deal. Right. And I think it is. Um, I think the call sheet thing isn't like less interesting now because they're clearly, have, they make money not through that, like they make money through overall the volume that call sheet trading probably like puts on to the platform, but they're, they're definitely not playing offense. I think we all would agree with that for now, but they probably will at that some point in like, absolutely. So actually in the RFQ space, right, right, right, which we will get to. That's a teaser. That's a good job. That's, that's good, um, narrative structure there, where we're to tease the true podcast profession. That's right. Um, we have a more question. A couple more questions on the internal market making, um, because again, they every time this comes up, I feel like, you know, you get all these scare headlines. So I really want to just dive into some of the aspects you said. So just for full transparency, like you all have the exact same sort of resources, what you can see, what you can't see as any other sort of API user on, on Novig. You, you can't see like user IDs or any of that type of, of things. Right. And I think that's the, the only way to operate in this space is to be like that. Um, otherwise, you're effectively a sports book. Right. And so that's, that's the reality of it. I guess. And I don't know to what extent you all have talked about it. I think the, the challenges is the PR of it because there's like a truth of, this is no different than any other trader. And there's the reality that there's how people feel of like, I'm being, you know, I'm against, I'm actually against an entity like, I'm against these, these power users who work at Novig. Like, I don't know if you, you've, I, because I just have seen it with, with Kalshi and I've seen it with some of these other companies. Like, I don't know if that's, that's anything you're thinking about at this point, obviously, relatively early in the journey. But like the PR of that or how to get your users comfortable around the, the idea that sometimes your trade is taken by, um, Novig itself. Yeah. Yeah. Right. No, I mean, I was like, the call, I had that one Kalshi like thread around, not throughout. I was like, in somebody's thread, but around that. And someone was like, well, if, like, it was truly no advantage, like, would it be a problem? And I was like, well, I mean, I guess not. But like, I don't know if you're ever going to be able to convince anybody that, of that, right? Or like, convince most of the people of that. I mean, the only entity that really needs to be convinced of is the CFTC or else they won't allow you to operate. But even with that being said, there's, there's still a bunch of misconceptions around IMMs, you know, internal market makers. Like, probably the dumbest one, but, you know, people from the outside. And this is where a lot of like, the scare, tactic, headwinds generate is people feel like an internal market maker is like, omniscient or always right. And, you know, that's just not really true at all. Like, internal teams are not optimized to be predicting outcomes better than you. They're optimized to prevent going bust while maintaining liquidity. Like, their biggest advantage is infrastructure around order flow. You know, they can see how different types of money behave and coordinate pricing across a lot of markets at once. And that makes them especially good at risk management and at not getting run over. But the flip side is that internal teams can be really slow to discover any type of like new truths in the market or detect any signal. You know, there's more process, more guardrails, and just generally more incentive to avoid tail risk than to lean into any kind of uncertainty. And, you know, that's a distinct design choice by IMMs. I also think there's like a misconception that an internal market maker, and like you said, is like somehow unfair or predatory. Like, in practice, they're mostly playing defense. Like, how you mentioned calcium trading, like, they are mostly playing defense. And, you know, they're reacting to information. The market is giving them just like any other internal market maker and trying to keep things in order. So like, I think the tradeoff is pretty clear. Like, internal teams are great at stability and scale, but that same structure makes it really hard for them to be early or creative in any type of trading that they do. You know, neither is inherently better. They just they just solve different problems. One more question that I'm not sure you're going to be able to answer. And if you answer, and we need to cut it, that's fine too. But what is what do you have an answer on what the current sort of like make take split is for for something like no big trading? As far as if what we're making versus taking. Yes. But we're like upwards of 90% making. Okay. I had assumed it was it was going to be a high number. I just I think I'll like, I'm just trying to put myself in a a sort of like normal recreational customer because I agree with everything you've said about it. Like, this is this is a dynamic that sort of needs like in my opinion to have proper liquidity, especially early on at any of these exchanges. And it's just like a reality of the business, I think. I think people would find it like less offensive or feel less bamboozled if if it was generally like the the internal trading is making. And maybe that's a wrong take. But I feel like people would be more like upset if they're like, oh, these these they're taking my bet or whatever. And maybe it doesn't even matter just because like recreational people never make orders in the first place. So, but yeah, they still get pissed. They still changes retaking. I just wanted to ask that maybe we can move to the the modeling and betting specific stuff because I feel like that's a lot of what your questions are geared towards. So I'm curious like maybe before we get into some more of the specific stuff, like, are you doing a lot, you know, we've mostly been talking about I would say like trading and not necessarily like the modeling, the origination, all these types of things. Like, are you doing a lot of of that stuff as part of your job at at Novig now? Yeah, in fact, you know, the the NBA just decided to shut down an API endpoint a few weeks into the season with zero notice. And, you know, we use like linear to like track tickets at Novig and like, you know, for engineering work. And my main one right now is like get back up and running because like a big part of my job during the NBA season is to be like, like, it kind of goes back to what you said about making versus taking like, it's more valuable for the exchange to if markets minus 110 minus 110 on a prop. And I think fair is minus 130. If I can go out and offer some plus 115s out there, like that is a net positive for the exchange. Like it drives price discovery. It gives us more flow than we would typically get. So like, I'm very much incentivized by both Chris and Henry to, you know, be modeling basketball or really any sport that I can, but primarily basketball and, you know, put original prices out there. Okay, that that is interesting. I think that's that's cool because so you're basically betting possible. I mean, that's kind of the use case of that's kind of the classic use case right on these exchanges is like sharp better offers are guests filled for infinity exactly. And like, why not do that if you're also sharp better making on the exchange? I mean, another cool part about exchanges is, you know, you never know who your counterparty is, but that also means that you can see things one sided and the market doesn't actually appear one sided. You can just be participating on one side of a market. So even if you like something maybe at big free and you don't have the strongest, you know, lean on it directionally, like you could just be buying one side of a market because you're making. You don't actually need to be seeding both sides and, you know, the automated trading systems that we have are always going to be seeding both sides, but like, that's one of those spots where like a trader can come in and play some offense a bit indirectly by being like, you know, I don't know where this price is good to, but I certainly know the other side is wrong. And it's a lot easier to be directionally correct that way than to just put your, you know, I think it's good to minus 130. So I'm going to start buying a ton of minus 115's. Like, you might not know if minus 115 is good or not, but you know if you could take plus 110 the opposite way, like you feel very confident there. So you might just make things one sided. So let's maybe take a step back and get into how you got into this. So is your your background? I know you said you were sort of like quantitative business type of route. Like, do you have, you have like a math background? Is that what you sort of came from in school? And have you been sort of like, because it sounds like when you describe Fando was very like much, again, like how I described trading focused and not modeling focused. Maybe that's a not a totally fair characterization as you gave, you gave an example with the season sim, I think. But have you been basically like doing this since, since you graduated? And so this is like a strong suit of years to the modeling? Yeah, I think, I think my best skill is definitely on like the modeling side. Not so much like, you know, I'm serviceable to code things for like, you know, UI or systems, especially with the help of LLMs these days. But you know, if you if I had to write code by hand, like I could write models by hand and not need the assistance of, you know, an LLM or the auto complete in a coding ID or something like that. So I think definitely like the math and stats background is very strong. And when did you start modeling? Like what was your first model you built? The third model I built was in the middle of college, this is where I started betting on, you know, that second friend's paper had with. And it was it was using whatever the main line was at close to build a live model that depending on what the score was of a game and what the, you know, the closing line was for spread and total. Calculate what's the probability for like wind margins and total bands. And the idea behind it was at the time, I thought like, okay, these prices, they start off super huge. Like if you're going to do an exacta of wind by six to 10 in the NBA, you're going to get something like, you know, plus 800 odds or longer. And throughout the game, if you had any type of like trading acumen, you could effectively build out positions across all these different bands. And, you know, you'd end the game up, you'd end the game with either a healthy arb or a very strong plus EV position, like as you dwindle down the fourth quarter. And this was the kind of bet that looked really good on sportsbooks. And it certainly played a role in me getting hired at Fandall that in the NBA, they used to allow you to SGP alt spreads and alt totals, which they don't allow you to do anymore. This partially why. But you could, you could effectively create your own bands outside of like this typical, you know, six to 10 or 11 to 15. You would just take one team's minus spread, the other teams plus alt spread. You can create any range of outcomes that you wanted. And if you did that, the live models weren't particularly good at that time. They were slow to react to places like pinnacle. It was pretty easy to just like paint yourself an arb or a very healthy middle throughout the game. So your first model was like a, that's not most people's first. I added two numbers together and multiplied a third. Oh, wow. I mean, it was a lot of like, depending, because I was still coming from the mindset, which isn't entirely true even today. But I think for anyone starting off, this is the right way to think about it. Like I was like the closing line is the best source of truth that I will ever have. And across sports, that's not always true, particularly in things like CBB. But for the NBA, I felt very confident that like the closing spread and total were good. And I have back tests from scraping the odds and seeing that like, okay, this is more efficient than the opener, which obviously it is. But I needed that validation at that time. And I just thought to myself like, okay, if I can use this like, what are all the downstream sort of derivatives that come from this closing line? And then I felt like the best way to action it was by like, because I didn't have like a real type of bank role. So I thought, you know, what can I do? That's risk-free. That isn't just straight arbitrage because surely like, if I know what arbitrage is and I just started looking into sports betting, the books definitely know it. And I had already been cut off from my friend's PPH account. And I was like, well, what if a fandal were to cut me off? Like, then what would I do? Like, and it didn't even occur to me that like, oh, you just get another fandal. But I was very like conscious if I want these bets to look bad from the beginning. And yeah, I think like in my interviews with fandal, like bringing this up and I'm going through my PNL on the market types I was doing. Because when you when you create those, you know, win margins and total bands, like you're betting live SGP. So you'd look like an absolute fish to the book. And so they realize like, well, actually our live model isn't that great. So all the derivatives from this live model also aren't great. And they compounded to be increasingly bad. So it was pretty easy to win doing that when, you know, bonuses in particular were pretty egregious at the time. So seeing me be like an 80% plus of my handle dump on live SGP. Now I was getting profit boost deposit matches, you know, all those crazy types of benefits left and right that it really didn't take much skill to win. Wow. That's honestly like probably the most impressive like first model slash intro to betting that feels like four or five years worth of learning and betting. But um, okay, let's talk. I think something we've talked about here. And you know, we did a like basic model episode. And then I think kind of did distribute. I think we did maybe three of those. But, you know, what let's talk about taking like a, you know, taking like a meaner median and a distribution versus sim. So, you know, you talked maybe about potentially a future sims at a fandible. But what, uh, what's your breakdown? You all sims or you all, you know, are you all point plus distributions or what, you know, what do you think's better and do you think there's a good use case for both? So there's kind of a running joke on gambling Twitter recently about the coastal elites perpetuated by peanut better. And this is probably where I lean into that a bit. Um, I'm pretty strongly in the camp that full play by play simulation is the end game and would absolutely categorize myself as a sim elitist. Um, it's not that single point models are useless. Like they're actually very good at specific problems. But, but once you care about more than one number or one market at a time, they start to break down. You know, even if you take a point estimate and wrap a distribution around it, you're still doing a lot of post-talk work to try and turn one output into many markets with a play by play sim. Everything comes from the same underlying process, you know, main lines, derivatives, player props, SGPs. They're all being priced off one coherent simulation of how the game unfolds. And obviously it's way more work. You're chaining together a bunch of, you know, sub models to generate the actual play by play and maintaining that is hard, especially like I said, one that MBA decides to shut down an API endpoint a few weeks into the season with no notice. But, but even if a sim is only directionally correct, you're, you're already in a much better place because correlation and derivatives are like built in by construction. Like if you have a single point estimate model, it might tell you the game total is, you know, 232.2. But then you still have to decide how to translate that into a line, a price and a distribution. At that point, you're basically recreating pieces of a simulation anyway, just in a way more fragmented way. So for me, like once you're you're thinking about pricing a league holistically instead of one market at a time, play by play sims just dominate. Like it's not always necessary, but it's hard to beat once you're trying to originate like full slate set scale. Like you kind of just need to play by play sim. Like otherwise you have what? half a dozen to 10 different models running at once. None of them communicating with each other like to build like a pipeline where they do communicate like you might as well just dip your toes into the simulation. So it's interesting. This this reminds me a lot. So we did a whole show on this. I think it was the simulations episode that we did. I was. Yeah. And so the the way I think about it, I'm curious, your your take on this is I sort of view it like that sort of a point model. So just so everybody knows what we're on the same terms with everybody. What we're sort of talking about is you have some model that spits out. I think LeBron is going to have 20 points and then the lines 21. How do you price that? That's like a point estimate model. The other way would be you have like sort of a full simulation. So you say 70% of these, you know, Sims, he has over 21 and a half points, whatever, 70% of time. So you have a direct answer. And what I would what I would sort of liken it to is like a a point estimate model is like a pair of scissors. It's not as powerful, but it can be useful like in the right place. But and you can't really hurt yourself with it. Whereas like a a play by play Sim or whatever is like a gun, like obviously you could do more damage with it. But you can also like really screw something up and hurt yourself. Because I the way I view it like a play by play Sim or anything like the errors or imperfections compound in a way that they will not in like a point estimate. So like when you were talking about the correlation parts of it, like I agree with you, it's the holy grail. But if I was like to ever think about this, something like like football, and I know you primarily do basketball, I believe, like the correlations compound in a way that's like really complicated. And if you screw something up in like the first quarter, then your second quarter is going to be messed up. Then your fourth, you know, it's just it's just a much harder problem. So like do you agree with that sort of split and how do you how do you sort of manage some of the the challenges that come with like compounding mistakes or imperfections is your answer. Just don't make any mistakes and have everything dialed in or how do you deal with that? So I mean, I wish that was the answer. I would have a lot more money than I have right now. But I also think that your analogy of the decision and gun is really accurate. I've never really thought about it like that, but I completely agree. I do think where you're discounting though is it's much easier to be fooled by a point estimate model. You might it's it's very difficult if you're just predicting say like LeBron points to be that far away. I mean, not this season, but like historically, he's like a, you know, somewhere between 24 and a half and 27 and a half points. Like you can quite easily even just using averages get to a number that makes sense to your eye. Whereas if you're deriving this output from a play by place in anytime one of those things goes wrong and it compounds your end output is not going to make sense to market. So like if you're if you're main lines look relatively close to market and they've come from simulating the whole game. It's not a perfectly safe assumption, but it's a pretty safe assumption to be like, well, if I got to the market total here, I probably have most of the player stuff cracked as well. Or even the other way around where you could be like, you know, LeBron's points look good here. I'm probably simulating like the scoring of this match properly because it's like they're accumulating over each play. You're not actually modeling it directly. So for you to come out any bit of alignment to the market, you've probably done more things right than wrong and it won't be too dangerous. Yeah, I think what you said about the mod like if you're pricing this is what we talked about last week. Honestly like having a sim or having that like more robust model. And these days if you're a market maker is really valuable. And what you said about like the models communicating with each other. It's like it's so in a sim you change a rate or an assumption. And like the rest of your your player points and your total, you know, quarter totals and all of that will react to that change. But if you like think of something new and you have 12 different models, you have to go into each of those models and figure out like, how does this affect my assumptions in like this? You know, maybe in the distribution or in the actual, you know, mean or whatever. And it's just like it just starts to get my anxiety. It just starts to flare up even thinking about having to go change all those different things. 100%. I mean the other part I kind of alluded to but like you just have so many more points to gauge calibration with a full play by place in. But you might have zero intention to bet the quarter derivatives. But you can go check to see if those make sense. You might not have any intention to bet SGPs, but you can go query like, you know, LeBron and Luca both getting 20 plus like it's the amount of correlation my sim has is that aligned with the markets implied correlation. You just have so many more places to gauge whether you're right or wrong. As opposed to just a point estimate model, you're just going to be like, because like you want to bet where you have the largest discrepancy to the market. And I think with a point estimate model, it really encourages you to regress to the market. Whereas with a play by place in, you're more encouraged to actually calibrate everything correctly. Like you have more, I have more faith when I show a large edge for my play by place in when I can go point at, you know, 12 other markets that's coming from the sim and be like, well, these are all accurate. How can just this one market be wrong? Like I have a lot of faith that this ROI is close to true. Whereas if I just had, if I modeled it individually and I got some, you know, something that was two and a half points off market, which is not like unfeasible for an NBA player prop and points, you might have some doubt whether that's actually true. And that's how they're going to lead you to like not hammering your best plays enough or passing up on plays that show smaller ROI because you just can't trust it. I think it's a fantastic point about the number of validation points because I would like in it to, if you, if you have a full slate of, you know, NFL player props or NBA player props and you're like perfectly aligned on market for all of them, all 100 players, except one, you probably would feel pretty good about betting that, you know, assuming there's not some news or whatever. If there was only two props on the board and you were off market on one, like probably wouldn't feel as good about betting that. And so like the more touch points you can create to sort of check your work. You could even, you know, like I've, we talked about about like checking to public projections and we're going to get into some of some of that stuff. But like the more contact points you can have to check your work, you know, and obviously if you have full play by play, and you should be able to price like everything. That's obviously why it's like so much more powerful is you can answer like any question, derivatives, props, core, to me, it's always been like the attractive part is like the correlation part. Like to me, I'm generally thinking about the decision that way is like, am I going to gain some, because you can't. It's really, really hard to do a correlation without sort of like a build up Sim. And you know, when you're, I've always, you know, I'm primarily thinking about football and I wanted to ask you about basketball in general because basketball, I guess my question is going to be like basketball seems like a nightmare sport in some respects to bet just because of like the, especially the NBA like lineup stuff and whatnot. But I do think what would basketball has going for it is like, there's a very finite set of outcomes of any potential possession, like it's not nearly as diverse as set of outcomes as what can happen on a football play. And so it's almost like, I remember when we talked about the simulation, on the simulation episode, we said like the perfect sport to do like play by play Sim was tennis. In some respect, I feel like basketball shares, shares some similarities without all the, the injury nonsense that goes on in it. You know, I agree. I think, you know, this doesn't cover every sport, but like broadly, I think that spectrum of like well suited for Sim versus not is tennis is, I think by far, the most well suited for simulation followed by baseball, then basketball, then probably hockey soccer, and then like way off is football. And I think a lot of that two has to do with just like, not sample size in like the generic sense, but like sports that have more plays per game or more like possessions or more events per game are generally easier to build a Sim for, where I think like football is like at the extreme of that where not only do you have so few games a season, like there's really not that many plays per game and they don't vary as much within game. Like, you know, I'm a no football expert, but like typically if a team comes out and is like, oh, like they're running the ball a lot tonight or this is clearly like just a passing offense tonight given the matchup. Like you only have so many games that you can learn that teams, like true rates on all those plays and all those players rates within those plays that for a sport where you can get so locked into like one possession type. It becomes really hard to like make any meaningful type of, you know, pull meant any meaningful type of assumption out of it. I also think like coaching like has the largest effect in a sport like football where football is just complicated. You might have a team team and then change your offensive coordinator and like now basically everything from the year before is close to being garbage in terms of like how they're going to perform the following season. If you're doing a play by placing. I mean, you miss golf, golf's a good sense for it. Everyone knows you're good. I don't want to get real sports keeping. I guess I get a lot of shit within the no big slack chats because I say every sport is fake except for basketball and that start pro basketball. NFL is as foolish as it comes. Baseball, another one, like all these are fake sports boxing. Don't even get me started on Jake Paul, Anthony Joshua. It's all these are all fake sports. Just get this out of the way. You have anything on the fight. This is going to come out after we don't want it. We're now putting them on. I don't know the ability to pass post it. No, I've got an argument where I people were throwing around that this is a rort. And I just think that's like materially incorrect usage of the term. Like not because you just have a capped ROI, like you know, you're betting something that's like such a big favor. You only stand to make like so many percentage points like on your risk. And I don't necessarily think that like a rort has to be defined by what the ROI is because it can be balanced by like how much money you can get down. And on this fight, you can get a huge amount of money down. So like it checks that box off. But I actually and you know, I was talking with Maple, who I know is a friend of the pod about it. I think the and you know, go down on the limb here. It's much it's much easier to say that Joshua is a lock when he's, you know, fucking minus a thousand everywhere. Like do we really know if the fair price should have been, you know, minus 2000, minus 10,000 if he goes out and wins? Like it's going to be very hard to like update what that there should be. But I do think the biggest edge in the fight and it's moved across a lot of books. So I guess to a sense I'm past posting. And I still think it's good in most places. Is the over two and a half rounds. I kind of like, I feel like everybody likes that. I've been getting a lot of hate for that. I bet a ton of it at a minus 120 minus 122. It's the only real take with conviction I have about the fight. Okay. But yeah. Okay. Just say I think that, you know, the minus a thousand is going to win. But yeah, because we posted the questions on Twitter, but we're recording this. It's the Friday before the fight. So these are when Louis comes down, loses his bet. It doesn't mean that he didn't watch the fight and he's just a total idiot. It's a really different record before the bet. It still might be a number total idiot, but yeah. Okay. Brief pause, but maybe to cap off the modeling betting specific stuff. Is there anything else you wanted to share as it related to modeling any like philosophies ideas around any of this we talked about that we didn't touch on? Yeah. So I think the, with a point estimate model, you can be greatly rewarded for being directionally correct, whereas with a sim, like perfect being the enemy of good comes up a lot. Like you have so many validation points to check if your sim is calibrated on that you can really sort of get lost in the sauce of always trying to improve. And, you know, for some more like basketball, like I might be, I might be tweaking to be like, oh my, my offensive rebound rate off a mist free throw following and ones is off base. Like how much value is that really adding to my ability to win at betting? Probably close to zero. But the perfectionist in me is going to get lost and trying to fix that. So like I actually followed victim to this all the time. Like I regularly leave money on the table tweaking sims trying to get them fully, you know, quote, unquote, right? Instead of grabbing the lower hanging fruit earlier in the odd cycle. And, you know, the irony is that like the market really just doesn't care if your model is perfect. Like it cares whether you're earlier to the information or not. A model or a sim that's directionally correct at 9 a.m. is usually way more valuable than a beautiful one at 5 p.m. You know, unless you're really a crusher and then you can, you know, win at post. But for the vast majority of people, like you'd rather be directionally correct early. And I think this is where a lot of technically minded people struggle. Like we're trained to want like completeness and elegance, but betting rewards speed and signal overall that. So like you can always update a model as new information comes in. But you can never rewind the market. So like for me, that's always been a constant tension. Like knowing when something is good enough to deploy. First of all, when I'm just polishing for my own satisfaction and and you know, friends like mine like like maple and fluff are pretty quick and frequent to remind me close to every day. How much my intellectual curiosity gets in the way of the amount of money I should actually be earning and call me a moron. I honestly think that was that was well said. Honestly, I do think a lot of people. Oh, sorry. I was just going to say I feel like a lot of people who who enjoy the modeling. That is an easy trap to fall into. Yeah, I got yelled at. Trapped a couple days ago by my partners, some stupid golf optimization I wanted to do. Well, like once you start doing it like the not even just the perfectionist in you, but like someone who is able to build a model that's that good. Once they notice that something is wrong, it's very, very hard not to just. And then like you'll have your betting partners or even just like friends who know your approach being like like you know, I just got down 80k today on these two props. Like how much did you bet? Zero. You're just in the modeling workshop. Like that's the money, man. Like this happens constantly. They get like fluff tells me. I forget if it's I'm either the dumbest smart person that he's ever met where I would classify him as the smartest dumb person I would. Which one is better? I don't know. It's better. That's true. There's no doubt about that. The second is in the IQ sweet spot where they're comfortable just dumping because they know in the long run it'll work out. And if you've year too far to the right, like you're just in a mess and unfortunately that's where I land most of the time, but it's a work of progress. Well, I have I want you to hit the the five P's of sports betting for for the audience to leave me with something they can remember. You want to take us through those? Yeah, so I've definitely made this up myself, but I'm a big believer in it. I call it the five P's of sports betting. It's pricing, priming, partnering, placing, and paying. And the big takeaway is that to be an elite better, you don't need to master all five. You really only need to be exceptional at one. And the best in the world are obviously strong across several, but mastery of just one P's more than enough to make this your primary source of income. Unfortunately for me, pricing is the PM best at and it's probably the least valuable one. You know, people on the outside think if I have the best number, I'll win or I'll win the most money, but the best price means nothing if you can't actually get money down on it. And you know, that's where priming comes in. If you're really good at pricing, garbage, low liquidity markets, you need accounts that are primed so you can actually get meaningful volume down. Otherwise, the edge just doesn't monetize and it's not worth anything. And if you're pricing more liquid markets, you'll eventually accumulate enough CLV that your own accounts get limited. And that's where partnering becomes critical. Some people in this space aren't created pricing at all. Like they have no clue what they're doing, but they're incredible at partnering, you know, sourcing fresh accounts, building networks, keeping things alive. And a lot of cases, that skills way more valuable than the price itself. And then there's placing, you know, you can have the edge and all the accounts in the world. But if you can't execute like, if you can't actually get the bets in cleanly and on time, none of it matters. And you know, that's where movers come in and why placing is its own distinct skill. And then finally, paying. None of the other form matter if you can't get paid. Paying is by far the most important P. A slow pay is always better than no pay. And anyone who's been around log enough in this business learns that less than the hard way. But yeah, you really only need to be good at one of these. But you definitely can't be bad at the last one. You can't be bad at paying. What do you think's the most you think paying is more important than well, okay, paying is like its own thing. I agree. You literally let that's like table. You just can't miss that. Like some people are like elite at paying where like you might have your saddle figure and they don't even care. Like as soon as something wins, they're sending you money. Which is annoying because, you know, you want to maintain, you know, pay pals and like forms of fucking money. You don't want to burn those. I mean, it's going to happen, but you'd like to minimize it. So that's not always the best thing if they're doing that. But like there are other people on the opposite side of the spectrum who, you know, you win a figure and you passed your saddle mark. Right. You know, it's every excuse in the book why they can't give it to you. And then on the flip side of it, you know, there's people who are really bad at collecting. Like I fall into that category where like someone will be trying to send me money. And you know, I've even been accused in the past of like trying to angle the price of Bitcoin like waiting for it to be my favorite to collect. And you know, like I don't care for that. But like I'm just a terrible responder and terrible like organizationally. So like sometimes I don't really know that someone owes me money. And they'll be trying to get in contact with me and like the message just like sits in the dust and they're like the fuck is wrong with this guy like trying to pay you. Well then the number one question we got was the responding. So we will get to that. But before getting to the listener questions, maybe let's do like 20 minutes or so on some of these industry stuff. Sure. That we wanted to that we wanted to talk about. Let's start with you had brought up some talking points you wanted to talk about related to services or tats, whatever, you know, whatever, where you want to use. So I know they're like very big in I would guess their presence is like the biggest in like the NBA props in these days like ones that are taken seriously like things like ETR and unabated stuff like that. So whatever direction you want to go to talk about this, you had sort of provided these as bullet points you want to talk about. But what is sort of like the landscape of that? And how do you think of those services? So I don't think people actually hate tats. I think they hate scammers and losing money. If a towel, if a towel consistently made their subs money, no one would care what they charged. Like the real issue is incentives like most tats aren't paid on performance, they're paid on confidence and marketing. So there's no real accountability when things don't work. And I think a similar issue shows up with some tools in the space like take projections, for example, like mean projections by themselves are basically useless to betters unless you have a way to convert them into prices. Or you're using them as model inputs or as part of a broader regression. Like a projection isn't a bet, a price is. And like this is where unabated has been really influential. Their player prop simulator is what actually lets people turn projections into lines. Like without that conversion layer, most betters wouldn't know what to do with a mean projection at all. And ETR is a good example on the other side of this. Like they're very sharp. I respect them a lot. But they mostly sell mean projections. Like they give out picks as well, but they mostly sell mean projections. And when someone thinks they have an edge and it doesn't pan out, the ETR teams fall back as often. You know, while our sims are different than unabated and it's like from a betters perspective, there's no clean way to like audit or reconcile that. I don't think it's malicious, but I think it highlights a gap. Like if you're selling in the space, you should be clear about the full tool chain needed to go from information, like a mean projection to a price. You could sell projections and conversion tools separately. That's totally fine. But selling projections alone to like a retail audience without a clear pricing framework is at best incomplete. And at worst, in my opinion, borders on being unethical. Yeah, I mean, this is like in my world of more of DFS stuff. I mean, and that's why you should be able to sell the projections, because there's a subset of paying customers who don't care how it converts to a line. But if you know that like upwards of 80% of your sales come from people who are going to be betting these so much so that an ETR's case, like you create an ETR props channel, you call it the MBA player props package. Like you can't just not give a way to convert those. Like charge more if you want, but like you just can't let those sit in the open with like no way to audit them or like reconcile. Like was this a good price or not? But does an ETR gives prop bets though, too? Like actually gives bets. Or is that not right? Yeah, they do give bets as well. But I mean, part of that is just like it's nearly impossible to like get down on them. Like sports folks are in there. Sports book employees are ETR subs. They're unabated subs. They're aware of what these lines are. Like put a poll on Twitter. I think Maple did this. He was like it was something like vote yes or like DM me or reply. Like if you've made a substantial money betting ETR props, because you can bet their projections and have a very healthy earn. But that's because the sports books themselves aren't able to convert ETR's mean or median into a fair line. So everyone is kind of dancing around what this price really should be until they plug it into unabated. And then it's like, well, how accurate is the unabated sim? You know, you could think about an example like the one I always point to is say like Jalen Brown and Jason Tatum. They're not lined at the same median, but there are games where like they virtually have the same median points given the match up and the opponent. In those games like the 50 plus alt for Tatum is catered like there's no way around this. It is a shorter true price than Jalen Brown's 50 plus like he just has a fatter right tail in his scoring distribution and a tool like unabated where you're only inputting like, you know, guard, guard forward like positional level stuff is not going to capture that difference in how it prices the tails. And so like, you know, you can make the argument like how much is someone betting on a 50 plus like, you know, that's just the extreme. Like it becomes pretty prevalent where a guy might be infer, you know, 1.153s and you think it's a massive edge on the over. So you want to take some tail risk and bet the three plus. Like you're just not able to convert those accurately. And people who are using those as source of truth are getting burned on exchanges pretty often to the point that in both the unabated discord and the ETR discord, they're constantly complaining about losing on exchanges. And sure that has to do with like adverse selection and your counterparty on the exchange. But it's also because they're posting arms to things that are not actually real edges and then they're getting burned. And you know, if a place like ETR who, you know, claims to have like a full sim, like they should just be providing you a way to convert your price. Like even if you want to pay while it like to the extreme, like there has to be some level of accountability in my opinion. I think it's an interesting take and I, you know, that is the part that's hard. Like we talked about it with the play by play sim versus the point estimate. That part is the like of the two parts. That's like 90% of the data science hard part is actually turning, especially in a sport like basketball. I would guess like you had said, like you can get a pretty good estimate for points or whatever. But the actual price is as hard. And so whether it's a lack of confidence in what they have, you know, behind it or otherwise, I do think it's a fair criticism because it's, it's setting people like without the right disclosures, it's setting people up to bet. It's like the, I, a while ago, I tweeted out like the putting the what our flex tape on the leak and it's just like Poisson distribution. Like I feel like that's like what a lot of people just do with these things. They're it's like, oh yeah, this, it's even worse probably in like sports like football where, you know, distributions are even wider like receiving yards or something. So like I totally get and I think it's a, it's a good and fair point. Want to go to exchanges or do, do either you have anything else on that? Well, I don't know if I, I don't want to believe at this point or this topic, but like, like I would just hate to see EZR now be like, oh, and this is, this is the whole treatment. I kind of like the fact that you have to work a little bit, but I think that's another reason. I think it's one or two reasons. It's either they can't, they don't feel like they don't want to put it all out there because they don't feel comfortable or they have made the business decision that that would be negative for them, their customers, the industry to put it all out there like that, right? That's, that's always a challenge of these sharp sports betting companies, right? Like, it's very hard to teeter that line. Well, it's like on the flip side, what if unibated just decided like we're not going to offer our player props simulator to the public, like ETR loses a shit ton of customers because there's a handful of people who know they're not fast enough and not glued to the screen enough to get the bets that they release and they rely on the projections and now they won't be able to convert them or they're going to try and convert them on their own and they're just going to be like easy pickings for people who actually have distributions modeled themselves. So like at Bet Bash, I was talking with a captain Jack about this and you know, I was half joking but also very serious. I was like how much would it cost to just like pay you guys to take this tool down? And of course, you know, he uses laughing kind of half entertaining the question, but like I think most NBA bettors would prefer, or I think the biggest winning prop bettors would prefer if the unibated conversion tool didn't exist. And there's a there's a good handful of like the top earners in the space who rely on that tool, but like, you know, if you're that right tail of a winner, like you're going to figure it out, like or you've bet these things enough times that you know like, okay, this projection matches to this line and I can directionally bet it. Whereas like if it's 24.6 on the projection, like the average like rec plus or like shrewd user who's making like 12 to 15k you're betting as a side hustle, they have no idea how to convert that. And like personally, I would love to live in a world where like I can just be taking all of that person's money because they don't know how to convert a price. So and it's probably selfish that I make this point, but I do stand behind it. Yeah, I mean, I've had the same conversation about data graph. It's like how much could we pay a data graph to just cease operations? But yeah, I mean, it's this problem with the whole touting and we can we can we can move on. We can move on from from the touts in the sharp, even like the ones that mean well, but like it comes to this point, like at the end is always here where we're like the best bettors in the space are like, damn, this should just not exist. We should make it less good because of its good that it'll ruin prices faster and you know, all of this. And it's just like for the business, the business wants to be good and have the best outputs and have the best record and all this. And anyway, well, I actually think that like sharp bettors, I think are increasingly coming around to that opinion. And like another opinion, I think they're increasingly coming around to is about like wanting fair limits at sportsbooks. I could vividly remember, especially while I was at Fandall, just a ton of like bitching on Twitter about like, oh, there should be a, you know, a standard house limit on this bet. Like all the all patrons should be able to bet the same amount. And like, I actually don't think people want fair limits at all. Like what they really want is predictable limits. Like as a better, the most important thing isn't like whether the limit is high. Like it's knowing ahead of time how much you'll be able to get down based on the market type and the time of day. Like if I know I can consistently click say 500 or a thousand dollars on a certain market within an account, like that's way more valuable than occasionally being able to get 10 K down and then getting shut out completely the next time. Like the other thing that people miss and again are increasingly coming around to is that limits are what make betting menus possible. Like if limits were higher across the board, the menus would shrink dramatically like you could see with a shop like circa. They'll let you click big like especially in football. But their menu is tiny like microscopic compared to a fandal or a draft Kings like limits are the trade off that allow books to offer a huge variety of bet types. And for short, better is that variety like actually matters a ton. Like you want access to derivatives to alt lines to player props. You just also want to know how much you can bet on them. Like I mean, there's also the competitive angle. Like if limits are high enough everywhere, you invite much sharper players into the market. And for most winners, like I promise you that's not what you want. You don't want to be competing against the sharpest people in the world. You want to be competing against the dumbest counterparty possible, which is usually just the sports book itself. And like once books start like opening things up and let sharps shape every market, you're effectively trading. It's pinnacle across the entire board. And like again, I promise most bettors just do not want that. I think we both agree. Yeah. That's the topic. I mean, there was a time where this was not to take. Right. I think on all the issues you listed, I don't even think it's I mean, my opinion is not even uniform across like what I'd call gambling Twitter. I think there's still people on both sides of of all of it. I mean like at the end of day, like you know, we everything we said about the the town. It's like we do a podcast here. We face the same challenges of like trying to balance making our show interesting and good to listen to and balancing like not blowing up spots and all this is it's just a really challenging industry to do. Like to me, it's all on a spectrum of like your intent and what you're doing and all this stuff, but it's it's all it, you know, it's all a challenge. So let's go to let's go to some of the prediction market and exchange stuff, though, because you are uniquely, uniquely qualified to talk about some of this stuff. So I really want to focus on maybe two aspects here. First, maybe let's just level set with like what you view the landscape, obviously you work at NoVig currently, you know, sweep stake a land. If I have that right, it's hard to keep up with what changes by week. Okay, currently sweep stake land on, you know, you can share whatever you can share, don't share whatever you cannot share, but obviously there's like a lot of push for for companies to try and get on the CFTC rails. What what do you view? Maybe like just like the broader, we'll talk a little bit about NoVig, we have a question that's NoVig more specific from listening questions, but maybe more broadly, what do you view sort of the sports book, the prediction market, exchange ecosystem looking like go next two to three years? So I actually don't think the space changes that much structurally, and like I think what does change is like who has the best infrastructure? Like edge isn't going away just because like there's more liquidity on these markets, but the biggest winners will be the ones who can move fastest and like operate at scale, which is just like the case now. The methods will just change and like maybe the people doing it will also change, but it'll always be about who can move fastest and who can operate at scale. So like ROI will probably compress over time, but that's relative to the volume traded, right? Like I'd much rather own a smaller piece of a much bigger pie, and like I just don't even think there's a debate there. Like you can always flex like a high ROI, but it's really dollars earned. That's the scoreboard. And in a world where we know we're two to three years from now or three to five, whatever the case may be, the only way in which like edge or roads is because there's just a ton of ton more volume being traded. And in like that regime, like it opens up to be like, well, how can I be the fastest to provide that volume? Because like the volume will never be provided at like negative like X, R, Y. Like people are doing this to make money. And like the edge might thin, but that's just because the volume is increasing. So like it just changes like what your goals are, but you know the overrarch and goal will always just be like, how can I make the most money while mitigating risk? Yeah. I don't disagree. I mean, you know, I'm definitely in pursuit of volume. Sometimes I've at the cost of X, R, Y, from time to time, but what's I want to talk about maybe the because it's like, okay, we talk prediction markets are exchanges. It's like yes, no big and profit or I feel like you've you've both decided like embrace the prediction market thing because you are exchanges. But I think the strategy has been first you were in this, you were earlier, you went sweeps. And then there's a strategy of how a lot of sharp betters have gotten down a lot of money on specifically like props and non-mainlines. And how, you know, how are you able to in this new ecosystem, both using stuff like profit and no big and using, you know, the, I guess like the more traditional exchanges, I don't know, call she in polymarket to get down serious, serious size on stuff like props. Like I think you said, you know, close up to half a million, you could get down in a day on something that like sounds like a pipe dream maybe three years ago. Yeah, absolutely, like, and I mean, it's a real shift even from earlier this year, right? Like exchanges change the problem for sharps of how do I keep accounts alive to how do I access liquidity? And like the meta is evolved, like I said, even from earlier this year, you can't just post fat arms anymore and expect to get filled, you'll get penny jumped or the market will move against you as soon as the sharp side gets hit at whatever sports book you're posting an arb to like, I don't think enough people really understand what they're doing when they post an arb. Like they're your counter party, if they are arbbing, which they almost always are, is taking the bad side of your bet on the exchange you post it. But then they are firing your sharp side into wreck books. And at some point, whether that was a losing customer in the past, a breakeven customer, if they're firing your winning bets into wreck books, like they're going to be reprofiled just as a sharp themselves. Like the book makes no, like there's no discrimination on the book side to be like, oh, well, this guy was a loser. Like they don't care. They just is this sharp betting coming onto our platform. And so like now it's much more about being patient, you got to slice your sides, you got to you got to work iceberg orders, just post a plus EV number to the screen and continuously refresh. Like the real big difference is you're not betting against a single sports book risk team. You're trading against other participants. Like, so now you can actually get real size down without all like the old gymnastics of, you know, priming and getting 40 movers to help you click. Like, that's why you're seeing people do hundreds of thousands a day on props. Now it's not because like the edges are bigger, it's just because the execution is easier. Or at least that's been my experience. So what I know, you know, in the lead up to the show, you had mentioned RFQs and we've referred to it earlier. If anybody's not familiar, like sort of like the parlays on these prediction markets. So I don't know if that's your answer to this, but like what do you view the biggest opportunities? I'm sure we have a lot of people who have just started getting into the prediction markets, maybe come from sports betting or otherwise. Like what do you view the biggest opportunities and where should, where should those people like be focused? Yeah, so I do think it's the RFQs. So like, but more generally, like the big opportunity is always to book a loser's bet. Where in the past, it was, I need to bet a good price into a sports book. Like, exchanges allow you to flip it. I mean, it's essentially the same thing, but like you can flip it now where the goal is just to book as many bad bets as you can. And I think the biggest, like the worst bet that like an uninformed or like a wreck user can make is going to be on a parlay. So if you want to book the worst bets possible, you should try to book parlays. So like, I think the biggest opportunity right now is on like the market making side. If you have the ability to price, you should be trying to price RFQs, you know, SGPs, parlays, you know, the holds don't need to be massive to be sustainable. Like if you're north of mid single digit hold with clean pricing and good execution, that is more than enough at scale. Like anything higher than that usually just reflects like uncertainty, lack of infrastructure or greed because like you will do less volume for short. Like you simply just won't win the order flow if your hold is too high. So like, you could try to just do some type of scraping or API where like I'll quote whatever fandal or I'll quote whatever DK has on this SGP on an RFQ. But it's like they're publishing holds is like, you know, upwards of 12, 15%. Like there are market makers out there who are see how much volume they can do on this and they're like, well, I'll do it for a 10% hold. Like why would I trade 2% markets on straights when I can just, you know, essentially book more on some parlays with a 10% hold and my price will look significantly better than the sports book, whereas like sure, you know, you might be offering minus 0 to on an exchange. So you can take your plus 0 to and have razor thin margins on a plus 100 and you'll make the argument that my minus 102 is so much better than the minus 110 they get clicking on sports books. But like that's also very easy to deduce what the fair price is on the main lines like that. Like they're often just around plus 100 on a lot of these things. So like even your like rec plus or kind of shrewd market maker want to be can go in and be like, well, I'll offer, you know, plus I'll let them take it at plus or sorry minus 102 or I'll let them take it at minus 104. I mean, Penny's taking they're taking plus 105 on both sides. Like I'll offer someone where they can take minus 104 and like I'll do a cent worse than penny or sorry, a cent better than penny. Like where I think people really can get into trouble is offering or betting SGPs without a coherent model of correlation. Like if you don't have a full play by place and you could still do this responsibly just by reverse engineering the correlation from the market. Like you can query up the SGP on Fandall and you can back out what the implied correlation is on that market. Then you can apply it to your own originator or a big free straight prices that you just devig from like market weighted averages. But you need something tying the legs together to account for correlation. Like I think the common thread is that prediction markets reward people who can turn prices into products. Like if you can quote RFQs even mildly manage correlation and then think about hold as like a function of scale instead of greed, there's just a huge opportunity here. And I pitched you both on it off air being like you should be making RFQs. I just got a little a little segment of code to show me all of the RFQ activity. So I haven't I'm not placing I'm not I'm still scared but you your pitch was good. And I think that was also a good pitch there. So I might have to cut this from the podcast. I mean generally like I would prefer if people don't get involved in making RFQs like I've written code to do that like I'm only going to be increasing my little competition. But like you know it's not going to be because I said this on the risk takers pod that people get into RFQs. Maybe hold on. Don't do it. Maybe you're putting the rest some listeners to give it a go. But like ultimately in the long run like like like SIG is going to get into doing it is like how she trading if they're not already doing it. Like there will be plenty of market makers competing for this RFQ flow. That like I think if you can get in while it's good and people are scared. I EGP. Damn you could be drunk. Shout out Brock. Shout out Brock. Yeah Brock is the RFQ guy right now. You know they have to come back on for a third. Yeah. SQ is like well we should get Brock on I was like that would be his third. He's a man that seemingly lived many lives. So we could we could have let's actually met him while I was working at Fandall. I think he was somewhere within the area and like I think on my lunch break one day like I just met him at a cafe across the street. He was like he's like you're free right now. I was like well I've got like 30 minutes on lunch break like let's chop it off and then like I got to go back to the office. You know the Brock is a great guy. Like I helped him is RFQ. He was having trouble to get them to actually accept I mean there was no shots at him but there was a there was a lot of LLM to code to get that process going. But I say that again not to take shots at him just to encourage people that like if you as long as you have like a mediocre understanding of what the code should be doing and you can articulate what you want to an LLM. Like you can just straight up copy and paste like the API documentation from CalShi. Yeah. Into cloud and cursor and like it'll just rip it for you like it it's just really not that hard. Like I think people are discrediting themselves by at least not giving it a try. And even if they fail they'll learn a lot along the way. That sounds like a good good segment to go over to the listener, the listener questions. We got Isaac's question first which is kind of the thing I feel like that comes up with the most. How are you seemingly always online yet unable to respond to a DM? Yeah. I don't know if you've seen his his follow up reply to his own question. No I haven't or maybe I don't know the lines of like like oh thank God that this tweet got a ton of likes. I was starting to think he was only ghost. Oh yeah. Yeah. Yeah. That's right. I didn't see that. That was the last I looked at it. It was like 15 like and I was like is it really 15 people I'm not responding to. But now I'm just the absolute worst like not low responder on the planet. I mean I don't discriminate like it applies to work, business, friends, family, everything. Like I'm also very like bi-modal about it. Like if my phone isn't in my hand I might not reply for days or weeks to a message. Like sometimes even longer. But if you catch me actually in the app like in the in the chat conversation it's the total opposite. Like my read receipts are always on for every platform. It's like instant replies. I mean like I might send you 200 messages in an hour like if I'm actually online. So it's honestly just an area I need to improve on like be more consistent in general. It's probably the highest ROI fix in my life for being real. He's just a man that doesn't want to be honest phone. I respect that in the year 2025. No further questions on that one. From Flup what is your favorite question that you have ever asked on our pod? What cons or let's ask that one and then there's a second part of the question. Sure. So my favorite question I've asked to the pod was about where you incorporate the market. Whether you regress your model output back to the market or use market implied numbers on the input side. And I was quite disappointed in the answer to be honest. I'm very firmly in the input side cam and it seemed you both. Wow. Moving on the regressing the output. Yeah. Every person I have this conversation to says that I thought about leaving it for a hot take but you know Flup asked. Like regressing output to the market just feels backwards to me. You're basically undoing the work you just did. Like I don't want to regress my largest edges. I want to bet them and I want to shrink the feedback loop as much as possible about whether they're true or not. You have a regress into the market is basically like using multiple models. This is how I look at it. In kind of like when I hear regressing to the market I hear like I make the number on this game six and a half markets at eight and a half. Like I now make the number on this game actually like seven point seven five or something. Like I'm just pulling my output to the market and whether that's like you know you could do some sophisticated things with it. Like I've heard Rufus talk on it and like you know my understanding is they have some sense of like depending on what time of day it is or or how many hours until the event start they're regressing like increasingly heavier to the market number. Which I think like that makes logical sense to me. But it's also like like I don't know are either of you familiar with like unpredictable. No I am loosely you know like they have a market implied like spread and total rating for like every sport. And like essentially what it is is like you know they're backing out from closing lines saying like if this team were to face an average opponent it's basically power rankings right. But they also have it for totals whether like if this team faced the average team this would be their the games point total not the teams point total. And like you could you could do clever things where like you get a market implied number and pass it through on the input side. Which well if it's effectively the same as like if I'm if I'm trying to get a number on an NBA total tonight like I know that I can just open an odd screen to check what the market has. But like because I know I can do that I would rather like either proxy that as best I can before the market opens or just take the number that the market is currently hanging and pass that through on the input side where it's like why should I run my model and then use my brain to try and like balance what I'm seeing in the market and what my model is doing like why don't I just like offload all of that to my model to consider. And obviously it requires you like having like historical odds like intraday odds data but like that's what I was going to ask about because that I think that's one of the biggest challenges is normalizing and you know we answered the question so I think this is good you gave your your cell on that side I think I think what we had talked about this I the part I had brought up was like it sort of depends on whether you're trying to capture like certain interactions with with it or not like my view is that having it outside you have more control that's you that could like that's the plus the downside is you could not be capturing things or it's not as clean like you would set like you said um but yeah I'm I mean I think it's a you to me that decision regressing to the market betting like not full kelly to me they're all the same thing that's the same thing it's it's it's like all it's fundamentally like you're all doing a similar thing it's just like what what problem you're doing and what you think is going to do it the most effectively and I think I go ahead and I was going to say as you said that I actually we actually do that like I get a I we do like a market and pledge player skill for golf and then but I was like yeah I was like I don't know I was like oh yeah I guess technically but the thing about you know regressing to the market um in market making as the week or the day goes on I think is is fair right I think another another point of this too I mean this might just say more about me and like my risk tendencies is like I if if I'm wrong like I need it to be like jammed in my face so that I can see that it's wrong like if I pass on a bet because I like wasn't sure if it was good enough like I'm never going to learn from whatever mistake we're striving that bet like I need to hammer it into the market and then see it get played back or you know lose on it or in the best case find out that I actually have a monster edge on something and um yeah like I think in more cases than not like if you're betting your largest edges like you're going to come out on top and the times where it's blown back in your face you're going to learn a ton about your process to like prevent that in the future. So was Flipp's second part of the question like did he know the answer because his second question was what concept is Louis included in his model RIP um he's most proud of is this is this different or uh yeah I mean like he's asked me a ton about it since like joining the Novig team and that's partially like like you had mentioned like the bottleneck is like getting the intraday hods and like trying to normalize all that like I think working for sports books and now in exchange like I kind of have all that data just out my fingertips where that's definitely like molded my viewpoint on it um but anyway like the there's a few things that I'm pretty proud of um one of them which I is the thing I am most proud of I do not feel comfortable sharing um but I wasn't good I that's why that's why we're the second one that I I'm nearly as proud of and do feel very comfortable sharing is I don't project minutes directly like I don't use minutes as an input at all in my simulation I you know you think about it like a NBA 2k like the computer playing the computer like I just see the starting lineups and mark who's active on the bench and I let the substitution patterns emerge organically through the sim just using an on-off model like at dead ball scenarios and for stable games like where the lineups you know even if the lineup isn't known but like where I can make an educated guess about lineups that's been a huge source of edge for me because minutes then become an output as like a complete distribution conditioned on game state rather than a static assumption so like I think this has made me have much better alt player prop distributions in the market and it's paid off really well but where I still struggle is like the edge cases like you know think about a time when like a player is on an explicit minutes restriction if I know a guy is capped I'll usually just stay out of that game because I haven't fully solved like how to enforce those constraints cleanly inside the sim I mean it's a work in progress and I'd like to think I'll get there at some point but like that's one of the downsides to this but again one of the positives is like I get minutes as a distribution and like you can check if a distribution is calibrated like that's just it's basic data science work and like as opposed to just being like this guy's 26.4 minutes where you know most people who bet NBA props they model like a per minute rate for each stack type then they project minutes or they steal minutes from an ETR or whatever source and then they just multiply the two and then they convert that mean estimate into a line like you know it doesn't take your mean being off by you know you could be off by 0.025 and now all of a sudden what you thought was an edge is no longer an edge and like that can purely be driven by like the minutes projection you were using was off by like you know 0.6 or something so like you can be less than a minute off on that and now your entire play is dead so I never really tried to just like project minutes like I get a minutes projection but like it comes from sin I like it yeah I think I I like it too because I think most people who are betting an NBA probably have like one way to generate edge and that's like have a different opinion on the rate statistics right you have two sources of potential edge those but also minute discrepancies I know people who had like a lot of success in in NBA DFS you know in the past where people who just had like very different opinions on on how much people were actually in a place that that's actually really interesting let's go to Calvin you want to go let it rip all right no big you have any future plans to get back to California in Arizona does the sweepstakes model remain the plan going forward I mean we're obviously always thinking about how to bring no big to more markets you know both in the US and globally that said there aren't any immediate plans to reenter places like California or Arizona right now the sweepstakes model does remain the plan going forward but at the same time like like every serious operator in the space including legacy sports books like we're constantly evaluating other paths you know things like CFTC regulation or some alternative way to expand access like that's for sure an ongoing conversation internally but there's not much more I can share beyond that at the moment and you're definitely you know people should DM a year in charge I believe of these decisions they leave them to you I accidentally pressed the button to pull us out of New York yeah I thought that was pretty funny okay we had trading and PR we had a lot of like at all questions and jokes we'll just we'll we'll acknowledge those but we got to get to the questions questions I think so let's go to friend of the show former guest maple how much alpha did I miss out on by not being a sports book trader for my first gig out of school yeah so I would love to hear I mean I'm not sure it was their first gig out of school but like I would love to hear like a shipper or an ADR talk on this too they haven't passing but you know never really like at least not from what I heard very directly but genuinely I think it's one of the best first roles for someone who knows they want to do this and work in this industry or you know whether you're betting or whatever like you just learn things you can't really pick up anywhere else like you see how limiter are actually configured how parlay blocking works what trips risk systems like what gets an account flagged how priming can really happen in practice like you could try to proxy that knowledge from the outside but it's just not the same as seeing it we're doing it first hand and for me working on the side of the counter was like the highest leverage thing I've arguably ever done like taught me how the market actually works not just how people think it works and it taught me a ton about building play-by-play simulations like probably my biggest takeaway from being a fandal was you know straight up reading all the code of how the play-by-play sums work damn that's the good point I'm going to be putting in a couple applications yeah I heard you had in the past that's right I applied to no big but the first time I applied to no big I I mean it would have been the third employee so wouldn't have made sense for my first job out of school but I also did not get the job yeah they pick Henry I think Henry is the right choice okay let's hit the NBA we have a lot of like your opinion on NBA players type stuff couple good ones so everyone asks do you think SGA is the best player and then we have a kind of give you give us your top 10 NBA players so if you want to hit that hit that quick it doesn't have to I know you have a hot take potentially about SGA so yeah so it'll kind of slip out here but you know I've kind of built a reputation online as a general generational SGA hater which I think is overstated he's incredible he's just not the best player on the planet like that's still in the Koli Yoke Hitch pretty comfortably in my opinion and if we're talking top 10 in in boost order I'd go Yoke Hitch 1 yana and this is assuming health you know Yoke Hitch 1 yana's 2 Wembee 3 SGA 4 luca 5 Tatum 6 Curry 7 Kauai 8 Donovan Mitchell 9 and Kate Cunningham 10 I mean you can argue about the ordering but that's like roughly the tier I'm working from and the big takes are that Yana's is better than SGA and Yoke Hitch is obviously better than all of these players I don't think either of us are the the right people push back on that so if you get hate on internet I apologize for this we'll get a ton of hate I have okay all right well then we don't have to do it for for the internet they'll do it themselves these people don't bother other we'll get we'll why don't we yeah we'll get to SGA in the hot takes to for you to explain a little bit more but one other NBA one we got how many NBA regular season MVP's will be won by an American in the next 10 years this is a really tough question and I was I've talked with Ontario EV bets before and I was a little frustrated he asked me to put me on the spot but you can honestly be 0 like when you look at the top of you right now like most of the truly elite guys are international like I had to put a number on it I'd say maybe like one juiced under like one one and a half would just feel way too high I mean you can like put a converted price but like I feel like it's like one juiced under like the closest to the median that's so crazy it was like if you were to say that like 10 to 15 years ago that okay I mean I might people might think I sound like I'm not saying this now so we'll see uh quick hitters best part of working at Novig over Fandall uh biggest difference is team size at and over we're tiny we're five traders total right now when I started at Fandall we were around 30 globally and by the time I left it was closer to 130 which is just way too many like with a smaller team you have more ownership there's less friction uh if you see something that should change that you can actually just change it yourself and for someone who likes to build like that's huge like Henry gives me a ton of autonomy and like being empowered to just like ship ideas instead of like having to navigate like layers of bureaucracy and process is just easily been the best part all right let's hit the funny question were you as we you know rumbling say you were the trader that failed to put enough vig on the Novig SGP right in this pricing well yeah I just got confused you were called Novig I thought we shouldn't but not seriously uh the short answer is no slightly longer answer is none of the traders were responsible for that and the longer answer than that was it was just an issue tied to like a third party Oz provider and they had a data ingestion error and I guess I just kind of leave it at that sounds good but yeah we can we can uh we won't put you on the spot too much we already covered we covered this in depth contrary to to my joke he is actually not the head of PR so I think we we would have been a failed company already yeah I don't think anyone's hiring you to be the head of PR which is probably on one bad that was uh let's hit blue beards if you have if you're working I was mostly as a joke was that yeah let's hit get I mean they're both jokes but he's definitely serious like I know I mean people can just I guess read the question but like well next let's go next okay okay okay explain your psycho computer and get repo reset workflow yeah so uh Nick knows me in real life and the psycho computer thing is definitely real like my reset workflow it is basically this instead of refactoring a messy code base of which I have a very like high bar for it to not be messy like it especially in the offseason I'll just delete the entire repo and force myself to rebuild it from scratch like I've done that three or four times um it sounds insane but it's my way of avoiding like the if it ain't broke don't fix it track because like if I know I can just win money betting with it still like I'm just gonna work on something else um like a lot of things aren't like technically broken but like I think there's something like structurally wrong and I've actually want to like improve something instead of just polishing it like I just need to nuke it entirely and build it clean which is fucking crazy I mean there's not parts you say like I like you like I have everything you know it's like in a private GitHub repo yeah I can't delete the whole repo delete all the code off my computer like there's no backup like or I mean there are backup but like I delete everything I'm glad we had you on the podcast that's gonna be a unique take I like I kind of I get it I like I otherwise I just won't do it like yeah know yourself right yeah um I think the second half his question was pretty good it's basically like would you be how much more or less profitable would your own betting be if you didn't spend time on Novig but you also didn't have the order flow inflow inflow the order flow info from Novig yeah so I think this is like this is probably one of my favorite questions on the betting side if I if I didn't work at Novig and didn't have any order flow exposure I probably make more money in like pure betting terms over the long run um just like having more time to commit to doing it um but like when you're on this side of the counter like there's kind of an unwritten rule like if someone uncovers something like you just you don't blast the market on top of them like you're obviously leaving Evie on the table knowing that like you can action someone else's information but like it's just an unwritten rule that like even while I was that fandom like you just don't blast someone else's work into the market like you just don't do it um that said I'm very happy like with the trade off like I mean I said a joke before like you know I've got equity there like a very handsome salary and like I get to build infrastructure that's like compounds every day like so for me that's definitely the best of both worlds and like I'm still encouraged to be betting so so we we go to the ending uh the ending part of the show SP yeah I think I think we're good did we was there anything we we didn't cover that you'd you'd like to talk about before the hot take and the the Bayesian update yeah I think uh we could fit in the uh substance abuse section here now um oh a typical section for us right so yeah so yeah what did you want to share with that I prepped pretty strongly for this pod but not for this section and his clichés and may sound I thought it would this would come across better just being like as unscripted as possible and without any thought going into it but generally um for people who who know me um there was say from 2018 there's like a six-year period where um I was like a chronic marijuana consumer like ounces a week of flour like several grams of oil like every two to three days like dabs if people know what those are like um just high 24/7 like I think in that six-year stretch the longest that I was ever sober was maybe like a six-hour period um and that wasn't by choice like I would wake up in the middle of the night and just take like two to three blinkers off like a pen and not even to be high like just because that was a wake like so I had to do that and then I would close my eyes again like I wouldn't even be awake to like experience like being high like it was just it was as chronic as chronic could be and um February 1st 2025 is actually the like from then to now I've consumed zero marijuana zero THC in any form and that that has been like one of the biggest like life changing things for me um and for someone who without like two to my own horn like I'd like to think I'm a I'm a fairly smart guy um like I was able to always like still function while being like super high you know just put in a couple i drops and like no one would ever know the difference outside of me and um you know they might not like hearing this but like you know February 1st is well after I started working at Novig and all of my time at fandal like I was 24/7 just cooked always being there like and again it just doesn't create for like for someone who lacks like self motivation that only just compounded it and I debated whether or not to shout them out by name but you know for the hundreds of things that I am eternally grateful to Big Buck Hunter for um we would actually from December or which is when I decided that I wanted to stop beginning of December up until like the middle of March the following year so 2024 into 2025 we would meet twice a week for an hour before going into the office at a cafe and just be like talking about things and like you know he was just super instrumental in like getting me to quit and then once I did quit like keeping me from going back and like I know GP is like touched on like not drinking is like the biggest plus ROI move you can do like either in a youtube video or I forget what what form of content he mentioned it but um that's just like certainly then the case for me and the hope is kind of that like if anyone is listening to this and there's someone who can be like functionally high or like even a functional drunk like it is like you know it's true people don't have to tell it to you you know it like you would be better off not being that way even if you are functional like you're just not maximizing like your own potential and like there's really just nothing more sad in the world than like someone wasting potential and like this certainly like you know buckets into like me being a terrible responder like I would just be too high to even like pick up the phone like I just didn't even know people were reaching out to me and like I got very comfortable with having like thousands of unread messages at once and like even though I don't smoke now or I'm not like consuming any type of THC now like that habit is still in a second one that's difficult to break where like I know people they get like one notification badge on their phone and they got a race to clear that or like they can't let their head hit the pillow if they have like unresponsive notifications like I just regularly will have thousands of notifications I don't answer and that's not because I'm like you know mr. popular or something it's just they accumulate over time and in the business like betting like responsiveness consistency are two of the the biggest things and I know we didn't quite ask his question but Bluebeards was and he asks this kind of facetiously knowing that I am a shit responder and have shit consistency particularly when I was working with him despite us being friends he said how important is consistency and following through commitments in this industry and you know it matters a lot and like this industry runs on trust but like that said as bad as it sounds like results by you patience like money's green and better is know that better than anyone so like if you can win people will give you more leeway than they probably should but that leash is never infinite like over time like consistency turns those like short term opportunities which can be really fruitful into longer term relationships and like speaking from experience like I've spoiled a lot of like very profitable long term relationships not because of doing poor work when I was there working with them but just for not being responsive and like you know it should no one should ever sour a relationship out of think consistency like let it sour because you actually aren't right for each other like don't don't shoot yourself in the foot yeah I mean uh thanks for sharing that and like you said like I stopped drinking and you know smoking weed too like I smoked a ton of weed um for a while and that I echo everything you say and I think it's great that you share that uh because I think curing from people that I looked up to who had stopped drinking or you know stop smoking weed was really like the one of the big I don't know um yeah for me one of the big parts that got me across the line so you know for the people out there listening to this like hopefully there there's some people there that that take that from you so yeah cheers man that's like what and again shout out Big Buck Hunter yes every year we go no one that is sad on Henry's name and we never do on this podcast I constantly am glazing him 24/7 so sure so I do the same so all right let's hit it let's hit the you know I mean you're a listener so you know the drill you got to give us a couple of hot takes I think you gave a I mean you have two in here one of them was SGA related uh but yeah hit it hit us with your hot takes and yes there's the SGA one like he's just not a top three player in the league um he just gets the benefit of having like a sick roster construction around him like for people who are into the advanced stats so you can check like EPM and Darko like his defensive side of those stats like you know being like five or six years into the league and being like a minus one or something has now become like a plus two like did you really become three points better out of nowhere or they just surrounded him with like these all-in-one metrics for basketball fall victim to like conflation like line-up conflation and it's very hard to parse like who's actually responsible for things and like I just don't think it's coincidence that you like check who are like the defensive leaders and these all-in-one metrics and it's like everyone on the Oklahoma City Thunder like are they all responsible for it like maybe like you know to team sport maybe like the way they mesh is was responsible for it but like it is certainly not that SGA is that good on defense and that like massively inflates his player rating and people like to point to that as why he's you know number one or he has the best EPM whatever the case might be they just don't know the ball I have told it tell me your terrible take on Texas the this one is probably an even worse take um but I fully back it I actually back this more than I back as yet in not being top three um I think I think they should be in one cent American odds like minus one oh one plus a hundred plus one how was it before you always is even possible for the probability all the way you agree this is why it's a hot take all the way up to like you know plus 99999 like everything should tick by one American and the reason I say this is because there's two sides of it like one who does this negatively affect purely engineers like and those people are sophisticated enough to just fucking convert the odds like they don't need to have an underlying like price data structure that we're like convert the American odds it's not that hard and who it benefits like we know like 90 plus percent of volume on PMs is sports like and what does every wreck user do when they open up sports they convert the odds display to American like they should not be like if you're on Novig right now you can jump in order of plus a hundred with like minus a hundred point two and unless someone has precise American odds on like it just looks like plus a hundred like people notice like why am I not top of book it's like well you know technically you were jumped by a hair like this also makes it slightly you know it's more expensive to jump by the median you could jump once once sent American odds and at the tails where the liquidity is a garbage you're encouraging huge amounts of price discovery like if you think it's plus 999 that you're willing to pay and there's a plus 9998 out there or sorry other way around like go ahead and jump it like you should be doing things that encourage people at the tails and like this encourages at the tails it makes the display better for wreck users like in my opinion there's just not a real downside to doing this like everyone should be able to just understand that like oh 101 verse 102 110 verse 101 11 like it's simple we're we're two hours in so GP you give your bottle I'll give my bottle then we can go to the basement I thought your thing was going to be like we need smaller text sizes I I mean these text sizes are too small it's insane at the tails I think it's I mean like but who's making at the tails me and so you don't have a price that you'd be willing to play too I don't want to play the 999998 converted 99999 who ever is going to penny jump me like and it costs nothing wreck users the data on wreck users on these exchanges is that they're very willing to penny jump around the medians they don't want to lay a lot to win a little like I I do accept that I accept that but I don't want any of you if you want to have a pissing contest over like 1.2 versus 1% fair like you guys should you guys should be doing that because that's ultimately better for the user yeah but it's not better for me and Rufus so one of you guys will decide it's not worth your time to engage in it but what what I'm what I'm saying is like like I I just want to make sure that like I the worst person doesn't get a benefit which is the penny jumper so I don't want penny numbers to get a benefit I do think that prices should get should get thinner at the you should basically yes you shouldn't do one cent spreads in the in the 90s and we I've I told this to big block hunter and I agree with this and engulf like the fact that the best the worst odds you can get is 101 is bad but like it does protect you at the median a little bit and I don't think like minus 101 plus 100 is enough at the median because if you're like penny or Chris and you I guess you do know this better than me but like how often are you moving plus 100 to plus 101 on any sharp action like not frequently you'll move like three or four cents at least you don't think there should be a world where like something that's 50 or you know 50% as fair like it shouldn't trade at like oscillating between like minus 101 plus 100 or minus 101 minus 101 but you should have it should be able to get that tight if people are confident it should go that tight I mean it's a different market making problem I mean it's also like in CFTC when I like it doesn't really even matter because when someone takes they're not even going to get that price anyway so it's not like like you're just encouraging makers to quote tighter and tighter but like anyone who's taking is not actually getting like that super tight pricing right but it's just encouraging penny jumpers to makers is what is what I'm saying if they want a penny jump if they penny jump it too far and then you buy it back but you can't because of the taking fee exactly so like you get to a nice equilibrium where it's either no it's not nice but take your fees either have to drop yes no more possibility for price discovery and like I think both of these are like optimal end points like you want take your fees low and you want the prices to get as tight as possible like pie in the sky prediction markets like their goal is to quote things as tight as possible yeah I don't I don't I don't I don't think that that's crazy I don't think that's obviously you're going to move towards efficiency like but I do think that you have to be really careful with making the ticsize smaller because nobody basically once you do it's really hard to make the ticsize bigger I think in just any financial market so you can't be you can't rush into it like it does it actually has to support like every the liquidity has to like be there and be supported by like market makers and and everything where I don't think it is in P out right now yeah I just think the how small ticsizes are around like you know 50 cent outcomes is a function of what the takeer fee is and that doesn't get considered into the ticsizing arguments by most people no I agree and like the reason why you wouldn't want small or ticsome on call string my opinion is if someone penny jumps me I can't take from that exactly so so that that's why that's the be a function of what the takeer fee is yeah yeah yeah I agree that's an okay take Henry oh like most of the tics you talk about it and it's and it's not that hot you know Henry used words that I cannot repeat on air to describe how poor that take was yeah because yeah whatever we've already gone 10 minutes on it I'll just put I think it's a horrible take like we don't we don't need to expand it but the the tail thing is absolutely insane to me so perfect you you was you divided us yeah it was hot um Bayesian update why we've been going long why unless unless GP do you have one and then Louis why I don't got anything good today with it's sort of long already I just said that there needs to be more male representation in Pilates classes you know I think that's the one thing that's keeping men from from Pilates is it's just like there's you don't see a lot of men in there but if you go do it it's all stuff that's like really are you the change I'm trying to be I'm trying to be okay I don't know but okay you know I'm saying like usually maybe there's one other guy there and it's just a lot of stuff that's like really good for people who have desk jobs so I just I think that you know get in there and drive the change all right Louis you want to close this out you got anything you've you've updated your belief for change your mind on recently yeah I think that uh LLM's I I go back and forth with this I think LLM's are a top three human like invention of all time and not nearly enough people are using them appropriately and they should be trying to work with an lLM more where I I also like back and forth between like you know obviously like the average person is never going to understand how to use this properly to I think everyone should use this all the time and right now I'm in the I think everyone should use this all the time can there enough I think that was one of the enders on one of our last maybe two episodes ago but um Louis thanks for coming on thanks for making us your first pod appearance and uh thanks everybody for listening we will see you all in the next episode