Jeremy Maletz

0:02 Bloomberg Audio Studios.

0:04 Podcasts.

0:05 Radio.

0:06 News.

0:09 Hello and welcome to the Money Stuff Podcast. I'm Matt Levine. I write the Money Stuff column for Bloomberg Opinion. And today we have a guest, who's Jeremy Mallitz, who is the head of goat markets at Susquehanna International Group.

0:22 You could say that. Generally goes a little better on business cards if you say head of prediction markets, but goat markets works too. Head of prediction markets. I've been head of macro for a long time at Susquehanna. And in a lot of ways, the prediction markets kind of grew out of that because as head of macro, we had quite a bit of focus around elections during election years.

0:47 We got involved in trading the election over in Europe where you could trade their prediction market style contracts on say Betfair. And we would also use some of the products that exist, say like predicted, even if you can't trade on it for informational purposes. So I'd known about it for a long time. The macro kind of transitioned into the prediction markets, which started out as a side project and now is a very large business for us.

1:12 My model is that there are all these prop trading firms that had 23-year-old interns who got into crypto as a hobby and then that became their crypto business. And it might be natural to assume that many prop trading firms also get into prediction markets as a hobby because of sports betting. Is there any of that at Sales Call Hunter, or is it strictly like, oh, it's election markets?

1:38 So when we first started the prediction market journey, sports weren't a part of predictions in that way. But that being said, there's quite a lot of overlap. And there are certainly a lot of people that are passionate about sports, myself included, the other head of prediction markets. We really got to know each other through sports as some of the DraftKings-style places became more prevalent, trying to understand the some of the opportunities and just being passionate about, you know, all forms of prediction. And so I wouldn't say that it was necessarily directly connected. Like the genesis of our business came from the macro angle, but it's kind of hard to disconnect them because, you know, people at Susquehanna

2:19 are just very into predicting anything that we can. And there's certainly very large markets around sports betting in this world.

2:26 When you think about like your overall book of business in the prediction markets, I'm going to ask you the split of sports and non-sports.

2:35 There's obviously different ways to kind of measure it. Certainly sports is more than half. You know, it's probably sitting somewhere in the, you know, it's not 90%, but it kind of fluctuates by season two. Like it's probably more now with football going on. It's a little less in the summer. It's on this World Cup. But certainly sports is, you know, a very large part of the predictions business overall.

2:56 I think there's some good reasons why that's the case. The biggest ones being that There was already a market for sports through all the regulated operators in certain states that existed. And this huge amount of sort of latent demand that sort of found a product that's a better product in a lot of ways. The other stuff is new, so it's growing. So I can tell you that the other non-sports businesses are actually growing faster than the sports business.

3:22 But the sports business is obviously coming from a much higher overall level.

3:27 Okay. So here's how I think about your business. And tell me if this is wrong. I think of there being like three components. There's classic like flow market making where you are making markets on sports elections, some third category, and you're, you know, quoting a bid and an ask and trading with, let's say, retail customers and hoping to earn the spread. There's parlays, which I don't know how big of your business that is, but I think it's really interesting that like traditional sports makes make a lot of money on parlays and to compete with that Kalshi has to offer parlays, which are called like combos or whatever.

4:04 And I gather that you're in that business. It just seems like interestingly different from the classic market making business. And then the third thing is what I started talking about, which is the goat hedging, which is like you have a block trade OTC-ish business where you will like. do bespoke hedges for corporate and institutional clients on prediction market events?

4:24 Are those kind of the three areas?

4:26 Kind of. I would probably lump the first two together. They're different market structures to market make, but the fundamental principles are the same in that it's still flow-based. We're responding in a different mechanism, but we have our model, we have our pricing, it's largely automated. They function the same from kind of a principles perspective, even if the- They're.

4:47 Interestingly different to me, but I can see why from your business, they're both flow automated trading.

4:51 Exactly. The way that we actually handle something, like the details are different, but like philosophically, I think they're kind of the same. And the main thing being that like everything is happening electronically and, you know, it's in sort of this all to all screen type ecosystem. The block business, that institutional business is meaningfully different. And I do believe that's the next phase. And it's where a lot of my personal time is going into now.

5:14 What's the split there? I assume the volume is much bigger in the float business. Much bigger. I assume you are charging wider spreads on the OTC-ish business?

5:23 Generally, although spread is a little bit of a weirder concept in that OTC business.

5:28 Because you're not on both sides. Right.

5:30 Most of those are positions that we're going to hold for the entirety. It's almost a little bit of a hybrid in some ways between sort of financial markets and insurance in that we have to do underwriting. We use the prediction market's to help with that price discovery and that underwriting. So it's a valuable input to help us to price these trades. But the actual market that exists there is not as relevant outside of the price. Like whether there's $ 1, 000 of liquidity or $ 50, 000 of liquidity on the screen, it doesn't make a difference if the customer wants to do $ 50 million of size. And so what we've really been setting up is to try to take the really big risks on and risks that the current

6:09 insurance industry doesn't cover. to basically sort of bridge those gaps in a lot of the emergent risks. Okay.

6:14 So let's talk about the goat hedge. I read about this like in August, there's stories about this guy who owns a goat herding business in California. Yes. And California law basically was, looks like it was going to require him to pay minimum wage to his goat herders 24 hours a day. And that would have been bad for business.

6:33 And he like wanted to draw attention to his plight, I guess, and or actually insure against it. And so he found an insurance broker who found you, or found Susquehanna, and he ended up buying a contract where he paid $ 50, 000 for $ 500, 000 of coverage against the risk that California didn't do something to change this law. Right. And that trade printed to Kalshi. You did a, I say OTC like over the counter, but you did a- A block trade on Kalshi. You did a bilateral trade with him that instead of just signing a contract and putting it in a drawer, it printed to Kalshi. Kalshi published the terms of the bet, basically, which is fairly complicated terms.

7:18 It's like if any of these 12 things happened that would relieve his pain, then the bet paid off for you. And if none of them happened, then he got his $ 500, 000. And that printed to Kalshi. Then anyone who wanted to could trade it, which I'm very curious who traded it. But we'll talk about that later. So you've talked about when people come to you for a $ 50 million trade or whatever, Like you can look at cow sheep prices and even though you're not like, you know, just laying off the risk immediately, like you have a sense of what the market implied probability is. But here there was no market implied probability because the contract didn't exist before you did the block trade. How'd you price the goat risk?

7:59 So we had a guy that got on the phone and talked to a lot of people. And he literally talked to goat herders.

8:05 Sure.

8:06 He talked to lobbyists. He talked to all the people who would have an opinion, plus some of our other internal research. We have our own research. We basically canvassed as much information as we possibly could to come up with an opinion on what we thought the fair probability was.

8:21 Did you have pre-existing goat research?

8:23 We did not have pre-existing goat research, but we certainly have a lot of kind of political research. In general... We are in the business of trying to price random things. So we have networks that we understand where to go to, to figure out how to price random things.

8:51 What is the history of Susquehanna being in the business of knowing how to price random things? Is that like, because you've gotten into prediction markets? Or is that like from the time Jeff Yass was a baby, he was taking any bets that people came to him with?

9:03 I mean, I think that is probably true. I don't know, baby, but like six, maybe? Yeah, no, I think that it's always been very much in our culture. I mean, I came to Susquehanna through a recruiting poker tournament 20 years ago. You had some questions earlier about kind of like market making, like I think there's different forms of flow market making, some that are trying to strictly balance and they're using all the different flow signals, and some that are trying to say, we're going to try to come up with a fair price on this and use all the information we have to constantly come up with a fair price and then take the risk wherever it is. We've always been very much in the latter category. It's much more

9:42 about we're trying to figure out what's fair as opposed to we're just trying to balance the order flow. Obviously, you do both. But I think that mentality has been in our culture of like try to price a thing, figure out what's fair. And we have a lot of random businesses at Susquehanna that are pricing random things. We're in the insurance business. like the official non-predictions insurance business as well. And a lot of that has been pricing random risks that people have. The predictions angle helps us to take that to the next level, but it's always something that we've done.

10:13 So why, when the good guys showed up, Why couldn't you be like, go see the insurance people? What makes something an insurance product versus a prediction markets block trade?

10:21 Insurance in general is very good at understanding. We have this large data set of things that have happened in the past. We know how to pool all this risk. We've got great procedures, et cetera. So you're looking for a standard property product. They're going to totally understand how to do that. They're not in the business of here's some new thing that just happened and we're going to figure it out. Like that's just not really how insurance works.

10:43 Not even at Susquehanna?

10:44 Well, at Susquehanna, that's how it works. But, you know, we are kind of in the insurance business sort of, but I wouldn't say we have like an insurance ethos.

10:50 Okay. I'm not saying ethos. I'm saying like the product, right? Like you have an ethos that is product agnostic, like we'll take a bet, right? Yeah. And then the question is, is the bet expressed as a calci block trade or as a contract called insurance? Like, is it just like, You get better capital treatment for that? Why print this to Kelsey rather than to the insurer?

11:09 Oh, so that is a good question. There are kind of three mechanisms really for which we can take on one of these trades right now. We can do a contract that's printed to a prediction market. We can do a swap and we can do a regulated insurance product. There are kind of advantages and disadvantages to all of them.

11:26 If you print it on an exchange- You have the most transparency. It's on exchange. It's cleared. All of those things are valuable to certain people. You do it as a swap, then it has more favorable capital treatment sometimes, but it's not on exchange. We might link it to an exchange. You do it as an insurance product. There are some advantages sometimes to being a regulated insurance product.

11:46 There's also some fees and some other drawbacks. So our goal when we talk to one of these people isn't really to push them in one direction or another. Our goal is to understand what makes the most sense for them. There are some players out there that say, we really want to do this on an exchange. It's important to us.

12:02 There are some that say, we aren't anywhere near close to being able to connect to one of these exchanges. We've got a million things we've got to figure out on our compliance side, but we want you to take this risk. Can you do it in a different way? And our goal is just to figure out how to take the risk. We're kind of execution platform agnostic.

12:17 I assume if you face a hedge fund, they want to do a swap. And I assume if you face a goat herder, he probably can't do a swap. He's probably... not allowed to do.

12:26 That's oftentimes more true, although there are exceptions in terms of how some of this goes. But yeah, for someone like a goat herder, it's way easier to do this audit exchange for sure.

12:34 So tell me more about like, you're a hedge fund, you want to do a bet. You come to you, you're like, hey, I want to do a bet. You like write up the bet. Like you have yourself and good lawyers. The hedge fund has their trader and good lawyers. You can like hash out the payoff mechanism of the bet. The goat herder bet, is like a tri-party bet. There's you, there's the goat herder, and you have lawyers, and maybe he has lawyers, and then there's Kalshi. The contract that you enter into is not like you sign it and the goat herder signs it. It's like you buy and sell a product on the exchange that is listed, approved, run by the CFTC by Kalshi. How does that work?

13:15 Who writes that contract and who has input into that contract?

13:18 Ultimately, in that scenario, Kalshi writes the contract, but It's a discussion.

13:23 But you wouldn't trade it with a goat herder unless you were both happy with the terms, right?

13:26 Right, right. So they ultimately are the ones that sign off on it. It's on their exchange. But we have meaningful input, right? We really do all work together. We are all discussing what the different terms are. And at the end of the day, everyone's kind of incentivized to get terms that cover the actual risk, right? We want to cover the goat herder's risk. They want to cover their risk. Cal Street understands that this market is in many ways being created by to help facilitate this specific risk here, this trade. And that's kind of part of the beauty of prediction markets is how easy it is to create these new markets now, like the fact that it can be done in the form of a day.

14:00 So we're all kind of incentivized to work together. So I would say it's a collaborative process. Like Kalshi has a lot of experience writing these contracts. We have a lot of experience doing it as well. The goat herder may be a little bit less, but he knows what his risks are. He knows the risks better than anybody.

14:14 Intuitively, having worked at a bank, If you were doing an ISDA with a hedge fund, you'd probably say to me that everyone wants it to cover all the risks and be fair. But you're getting a little edge in on the contract, right? And they're trying to get a little edge in on the contract. Kalshi, I think, has a genuine incentive to... If something happened where the goat herder was disappointed, where he didn't get... the payoff, things didn't work out for him politically, and also your contract didn't pay out, like that would kind of be a windfall for you, right? And vice versa, like if his contract paid out any other thing you wanted, that'd be a windfall for him. But like for Kalshi, they want

14:57 like the right result in either case, I think, right? Because they're sort of reputationally trying to build this business. So is it like helpful to the business to have like a neutral arbiter, like during the contract and the determination of the payoff?

15:08 So I guess what I would say is, first of all, I don't think that that sort of haggling over terms and trying to win that has been my experience in this so far. Like we are also trying to build the business here and we actually don't want a situation where we win, but the other side like didn't get covered on their risk.

15:26 Have you ever like retroactively, have you ever like written a contract where like in hindsight, the other side didn't get covered and you're like, yeah, we'll just pay you anyway.

15:34 That specific scenario hasn't happened. We've been involved in some contracts where we might have realized there's some form of an edge case where theoretically something could happen, but we've never actually had that edge case happen in a situation that's sort of a relationship type business where we've had a relationship with the other side. There are edge cases in sort of the general screen markets that happen here and there every once in a while. We're making our money on the price at the end of the day.

15:58 We're charging a little bit of edge on the price. We don't want to make our money on the contract terms. The worst thing that can happen is we don't want the next story about the goat herder to be like, oh, he lost his hedge and he lost his business. That's horrible ultimately. And you mentioned in your article, which is fair, part of the reason we wanted to do this trade was because We think there's some degree of edge in the trade. Although to be honest, we priced this one pretty fair. A large reason was because we think this is an important business for predictions overall. And we wanted more proof of concept out there. We want people to see this is what can be done. We'll win in the long run if this business takes off. And our

16:37 role right now is really to bootstrap the liquidity of this business and to show what's possible.

16:41 Okay. So when I read about this trade, it was probably, I don't know, a week after you printed it, something like that? And I looked at the screen for this contract, which, you know, you printed a $ 500, 000 notional trade with this code herder. And when I looked at the screen for the Kelsey contract, it had something like $ 500, 476 of volume.

17:03 So I think that's your trade plus somebody or some people who found this trade after I wrote about it. And like, you know, nine other articles were written about it. Like, I think that picked up and When I looked this morning, there was a total of $ 655, 000 of volume. The contract resolved. You won, which means he also won. When I wrote about it, it had traded tiny, tiny, tiny size in the 80s, which means it traded down from the price you paid, which means that you looked a little like you got picked off by the goat herder, which I think I said to you. But in the long run, you did fine. And it's hard to extrapolate from that little volume. Who traded this thing?

17:46 You know, there are just a lot of people that are passionate.

17:50 Like some money stuff readers, clearly.

17:52 Some money stuff readers. There's a lot of people in the predictions ecosystem that are passionate super forecasters that look over every market and they probably have some tools to help them find ones where there's any form of volume and they try to make a prediction in a trade. And if they think that they have some kind of an opinion, they do it. My guess is that's where this came from, although I can't know for sure. It's also possible that I'm sure some people read your article and saw some of the other publications and were like, oh, I want to get involved in this.

18:19 But the overall- You could buy it at lower than the price Susquehanna paid, which feels like a good deal.

18:23 In theory, you could, yeah.

18:24 Also, I wrote about this. In theory, if instead of $ 400 a volume, there was $ 10 million a volume after you printed the trade, you could have laid off your risk, right? Yeah. Depending on, you might have lost money if it had all been at 85. But also, with the amount of volume there actually was, ultimately, if you just take the beginning and the end of the trade- the law was changed in some way that helps him. I don't actually know the details, but like the bet resolved to the law was changed. So he got $ 0 on his bet and you got your 500,000 back. You didn't pay out the $ 500, 000, but he could have like sold his position, right?

18:59 Like he paid 10 for it. He could have sold it at 15.

19:01 He certainly could have. And, you know, to be clear, like, There were pieces of news that came out in the interim around– I don't know if the nation was following this with bated breath, but– I.

19:12 Should have been. I'm probably the most interested person and I did not follow it.

19:16 There were pieces of certain news that came out around it and if you were in the weeds of it, the probability might have been moving around a little bit here and there and you can interpret things in certain ways. You're right. The volume wasn't super high. In theory, you could have a scenario though where someone puts on a hedge trade– And then decides after the fact like, oh, you know what? Like this is now trading at such a level. Like I'm going to take off some of my hedge, all of my hedge. Maybe I don't need the hedge anymore. Something else happened. Like that's one of the great things about this form of hedging as opposed to say like some just like an insurance contract is it feels like a financial product, right? You can get

19:50 out of your risk.

19:51 It feels like an exchange-traded product.

19:53 Right, like an exchange-traded financial product. The ecosystem needs to scale up and we're trying to bootstrap that scaling, but– Once it gets to that point, our hope would be like, yeah, trades come to us. We put them up. There's an active market in this thing. Other people might have similar risks.

20:09 And ultimately, people can trade in and out of their position. If they want to come back to us and they say, hey, the market's trading here now, would you let us out? Would you let us add? Those are all flexible things that we can do here that wouldn't be possible without the market existing. Okay.

20:24 And let's talk about capital and funding. When you print this trade to Kelshi, if I'm a retail trader on Kelshi, If I want to bet on something that has a 90% probability of happening, I put up $ 0. 90 on the dollar. That's what you do here, right? You're putting up.

20:38 $ 450, 000.

20:39 He's putting up $ 50, 000. It all goes to Kalshi. You get, I guess, Fed funds on your cash or whatever. And when the contract resolves, either you get the $ 500, 000, which is what happened, or he gets the $ 500, 000. I assume that when Susquehanna is trading options, it's more capital efficient.

20:59 Yes, it is. And I think that ultimately that's one of the areas that we hope to see predictions evolve over time.

21:06 I've heard recently that Kashi is rolling out leverage for at least some institutional. Right.

21:11 And I think that'll be very helpful, getting forms of margin and leverage. Realistically, what helps the most here in this form of business is something that looks a little bit like portfolio margin. Say that we have 100 of these trades on where all of them are us selling 5% to 10% of something that's not super likely to happen happening.

21:32 That's what the insurance business looks like. A lot of risks are people wanting to cover some not very likely thing that happens, but if it happens, it's catastrophic. You have 100 of those things on. They're all uncorrelated. In theory, you shouldn't actually have to fund 95 cents on the dollar on all of those, right? That's not super efficient.

21:53 For now, we're kind of just saying, we're going to figure out how to fund these things.

21:58 How do you find, like, is that like Jeff Yass's equity or like, is someone giving you that portfolio margin in some form, you know?

22:04 So in traditional options business, the prime broker is going to give you a portfolio margin. Here, there is no portfolio margin yet, right? So we're just funding the cash, right? And it makes the trades a little bit more expensive. It means that if you have a really long dated trade, so let's say, for example, someone wanted to insure against the Olympics being canceled in LA, which is years away, that becomes very challenging to do on an exchange in the current setup because then your cost of capital for having your 90 plus percent on for that long is meaningful, right?

22:36 A shorter dated trade is much easier.

22:54 Okay. So I want to talk about like your general OTC book. I say OTC, I feel like technically the goat herder trade is not over the counter because you printed on an exchange.

23:03 Institutional risk hedging is kind of the term that we use internally to.

23:06 Talk about it. In that book, first of all, like how many trades do you have? Is it like hundreds? Is it like Dozens?

23:12 It sort of fluctuates from time to time, but it's more in the dozens than the hundreds now. We're obviously hoping to grow it.

23:18 What's the split between true OTC swaps and on exchange?

23:24 I don't know exactly in terms of the number of trades. In terms of the size, the larger trades tend to be more likely to be a swap.

23:30 And what's the margining on that? Is that like unsecured? Is that like you're posting daily collateral?

23:37 The margin is generally just negotiated. There are certainly some of those trades where we put up zero. In the GoHerder trade, if we did it on swap, we would probably say, I think you can probably trust we're good for $ 450, 000. There could be larger trades where.

23:51 We- You'd take his 50 there. Probably.

23:53 But when you start talking about larger types of situations, we might have to put up margin. The other side might be given an ability to do margin. When you're on swap, it's all case by case basis. It also provides us with quite a lot of flexibility that we may not have with a pure binary prediction market. To give you an example, we've had parties that have come to us interested in saying, we're really concerned about a moratorium happening on data centers in our state or in our county or whatnot. But if that happens, what we're really exposed to is now the depreciation of the chips.

24:27 Can we come up with something that has a trigger that's like a prediction market-based trigger, binary trigger? Does this thing happen? But then pays out based on the price of an asset, whether that's the price of chips or the price of compute or sometimes the price of power. And we're pretty flexible at being able to do that because, again, if we have a reasonable index to settle to- that value is, that's kind of our bread and butter, right?

24:50 Moving into a financial asset and doing that as a firm. So that's the thing that we're good at and it covers their risk more directly, all the better, right? So we're very happy to be creative when it comes to that swap and that can go to the margin. We're going to look at all the components of it and try to figure out what's the best way to get this thing done that sort of fits all the parties.

25:09 And again, we're looking at this business long-term. So we do want them to be ultimately be happy as well with what's happening because we want repeat customers and Some of these are small communities. We want one data center to tell another, like, oh, we got this thing and it was helpful. So that is our goal.

25:24 One thing I always thought about is the Cal Shoes will tell you, like, we got into this business because people want to hedge election risk. And, like, there's the story of Jane Street getting the election right and, like, losing a bucket of money because they got the directional impact of it wrong. And I always thought that was very, like, illustrative because, like, there are people in the world who are like, I think that Trump will be bad for the S & P. And so I want to buy...

25:47 a hedge on Trump getting elected because I have all this S and P exposure, but it's totally the wrong hedge because like, you know, the S and P goes up, right? Like when you have people hedging the election, is that the thing you talked about where it's like, there's an election trigger and that a financial asset payoff, or is it like people who are like, I'm in a very regulated industry. I'm going to be shut down if the Democrats win. What is the intuition behind people doing election hedges?

26:13 I think those are more just binary. It's just who's going to win. And I do think that generally- Why is that a hedge?

26:17 Who's hedging what there?

26:19 So I think there's a few different angles. One of them is like you said, this would be just very bad for my business. And I do think that the things that are the real businesses, like this is bad for my business, that's probably the thing that's a little bit slower to pick up because it takes longer to get someone who has a real money business to start doing this. That's a totally new product. It makes sense why the financial players get involved first.

26:42 Well, but also like if you have a real money business, like if you sell ball bearings, you're not going to get shut down if like one side wins the election. I could name some firms that might get shut down if one side wins the election. Right.

26:54 So oftentimes it might be a situation where it's not like you're going to get shut down, but there could certainly be economic impact to your business, right? And you could want to sort of hedge out some of the economic impact.

27:05 I'm just wondering like how modelable that is, right? Is this like Politically committed CEOs were like, oh, if the Democrats win, it's going to be a disaster for me. But like, that's not empirically true. I feel like people's intuitions about the effect, even Jane Street's intuitions about the effect of an election on like their business or asset prices are unreliable. And it's a strange way to hedge.

27:28 I can give you an example. Oftentimes, I think if it's a real business, it's much more targeted. It's much less like vibes, this party is going to crush me, but more say, again, people just to go back to the data centers, because it's the thing we're hearing more about than everything else combined. And I wouldn't say it's like the majority of the trades we've done to date, but it's by far the majority of the interest that we're getting. If you're a data center operator in Texas, you've got a governor election, data centers are 100% on the ballot. Now they could come and they could say to us something like, we actually want to hedge the moratorium risk, but you could also get a pretty reasonable

28:02 proxy for just hedging the governor race. And those two things wind up being extremely correlated. So I think with the real money business, it comes more down to like, what is the actual policy that's going to impact me?

28:14 Okay. So like you're a data center developer and you're like, if the election goes one way, there'll be a moratorium. If the election goes the other way, there won't be a moratorium. There's a moratorium contract that you can buy from Susquehanna and there's an election contract. There's, let's say, no basis risk in the moratorium contract, right? If there's a moratorium, you get paid. If there's no moratorium, you don't get paid. There's somewhat obviously... basis risk and the election contract. Like they're highly correlated, but there's like obvious basis risk. Why are you buying the election contract? Is it cheaper? Is it more fun? Like what's, is it, is it the more holistically a threat than like a more,

28:49 like what's going on there?

28:51 I think the way that the conversation kind of happens is the party might not actually even know at the beginning that there is a moratorium market. Yeah, but you can tell them. Which is why I think that realistically.

29:00 Why are you selling them the election one?

29:02 In that scenario, I don't think we've done a lot of sort of like real money business trading the election. The people who want to trade the election itself are much more likely to be a firm that's on the financial side, where they're looking at the financial impact, like a hedge fund, for example. They're looking at their portfolio risk and all the ways that their portfolio is going to move based on that election.

29:22 So they're like, my portfolio has some basis risk to the moratorium contract, some basis risk to the election contract. So neither one is a perfect hedge to my overall portfolio. Right.

29:30 And I think that most hedge funds that have portfolios are are used to trying to break down factors. And election is absolutely an important factor, and you've probably seen some of the baskets that banks put out of, here's all these different products that you should put on to bet on Republicans winning or Democrats winning.

29:48 And those are- Your portfolio looks like one of those baskets you're calling Susquehanna for a hedge. Right.

29:52 And you can hedge that by putting on that basket. But again, that basket is actually oftentimes a dirtier hedge. You have the ability to, you could close some of your risk, but a lot of these firms might like those positions. They don't want to close the risk. So if what you're worried about is what's the impact going to be on this portfolio, you can hedge the election directly.

30:09 And I think that's where you're more likely to see the election. Plus, to be honest, people have a lot of opinions on the election. I think you have a lot of people who just think one side is going to win or not. It's a market that obviously trades a lot. If you're a real money business, I think in general, you're probably going to go narrower. And that's what we see. Like the real money businesses, they're looking for things that are much narrower the firms that are closer to financial markets are often looking for things that are broader.

30:32 And then is that because they're hedging a broader portfolio or because they're taking directional risk with you?

30:37 I think it can be both. Obviously, when we are doing this institutional business, we're really trying to focus on the hedging trades in most markets. Now, in an election market, especially a big one like, say, Senate control, those are liquid enough markets with things going back and forth that we're happy to take on the other side of someone else having an opinion and it's their alpha trade. If someone wanted a very, very bespoke thing. Like if some hedge fund came to us and they were like, you want to put on this goat herder hedge, we would be less likely to take it on.

31:04 I wrote about, you did a hedge for a Spanish soccer team getting relegated. That is like a blazing red alarm adverse selection risk, possibly, right? And I assume, and I wrote this, that I assume that your due diligence involved like modeling how much money they would lose by getting relegated and making sure that you were hedging something less than 100% of that. 100%. Yes. How much adverse selection due diligence is there in these markets? Are people coming to try to pick you off? You know what I mean? Are politicians betting against themselves? Is getting adversely selected by people who know more about these markets a big risk on the OTC side, or is it all pretty up and up?

31:46 To this point, we haven't really experienced that on the OTC side. If we're going to show a big size and a tight price... We want to have an understanding of the story, right? And we actually really want to work with the other side to understand.

31:58 You say a tight price. I assume you're not showing people two-sided markets on these things.

32:04 There are times that we do. There are times that we will show a two-sided market. A lot of the times it is one-sided, but there are some parties that actually want a two-sided market and will make that.

32:12 And tell me, so you said you want to understand the story, but what's the story of a two-sided market? It's just the hedge fund wants to speculate or.

32:19 So sometimes it's coming through a broker, right? And brokers are just used to- We want to get a two-sided market. We want to show the width. That's sort of part of their process. In that case, we're happy to do it. Usually, if it's a broker, they're more likely to ask for a two-sided market. Not always, but more likely.

32:33 You say brokers. A broker comes to you with, we have an anonymous client who wants to take a position on the Texas election. Can you show us a two-sided market? So you don't really have a sense of the story there.

32:42 So if it's something that's kind of more general, right? If it's the Texas election, there's a lot of information out there.

32:48 You're not likely to be adversely selected by someone.

32:50 We're not that worried about adverse selection. And so it's fine to throw a two-sided market. They could have an opinion and that's okay, right? If they're like, we want you to build this market from scratch and it's this very specific thing, go back to the goat trade. We've obviously talked about that one a good amount today.

33:05 It's like the best paradigm. Right.

33:07 No, it is great. Obviously, we're going to need to know the story, right? You can't design the trade that you have the opinion on to get your alpha trade against us. We're not going to be able to get out of the risk in some very random specific thing. So it really comes down to the specificity. If it's a really narrow trade, we want to have a conversation.

33:23 We want to understand why you want to do it. And part of that is so we don't get adversely selected. Part of that is so that we make sure we actually design the right product that actually takes your risks on. If you come in in a liquid market, I want to predict who's going to control the Senate. That's fine.

33:36 We'll make you a two-sided market in size. We feel comfortable. There's enough trading in that market that we could probably get out of a position in that. And, you know, we know enough about it that we're not that concerned that someone's going to adverse select us.

33:47 In crude numbers, this book of RTC trades, like, are you laying off like 10% of it or like 90% of it or like?

33:54 Probably closer to the 10% than the 90%.

33:55 That's what I thought, yeah.

33:56 This is a business where we are the vast majority of the time holding all of the risk.

34:02 I assume if this business is like 100x where it is today, you'll be laying off 90% of the risk.

34:08 Yeah, for sure.

34:09 If everything gets super liquid, you won't be holding. Absolutely. You won't be like a giant insurance company. Right.

34:13 And I think there's different ways that we ultimately can wind up laying off that risk. Some of it can be the markets are liquid enough that we're just doing it in the markets. That'll probably work for some things. And the broader you get, the easier that is. So I can absolutely envision a scenario that's not so far in the future where there's a liquid enough market in, say, will Texas have a moratorium on data centers that we can trade in and out of that. When you start talking about things that are hyper-specific, you're a lot further from that. Our goal is we really do want to work with people in the insurance industry because they're used to all different forms of syndication and sharing and things

34:49 like that. And so that's probably one of the better ways that we can ultimately wind up sharing some of that, like laying off some of that risk is to potentially share some of it with someone who's in the insurance space. And I think that's probably a much easier way to go for the specific things.

35:04 Let's say a data center moratorium trade is- not that liquid. I don't know if that's true, but Texas governor is really liquid. Are you taking $ 100 million of data center risk and laying off $ 50 million in the governor election and taking some basis? Is there a factor correlation model of hedging this stuff?

35:23 It's more that it influences how we will act. So we're more likely to try to lean into a position in some of those other markets. Most of them, if you're taking a massive trade and you're going to just sort of like, put this massive trade on and have the basis risk. Like there's still not quite the liquidity in like Texas governor to be like, we're going to put on this massive trade. We're probably going to impact that market in the process. And we're probably going to wind up with some degree of basis risk there too.

35:49 So it's more of a, we have this on, we understand the important factors that are our risk factors that we have. We're going to sort of use that both to change what we think is fair, because there can be information in some of these trades and also to potentially try to sort of reduce the factors as we can.

36:04 But that being said, we understand that a lot of these are we need to basically manage our risk budget effectively. And some of it can be done through laying off and all those things. A lot more of it has to be done through being selective in terms of what are the trades that we're going to put on, what trades make sense. And when we look at data centers and just the total size of the risk that's probably uncovered right now, it's in the hundreds of billions of dollars.

36:28 So it's definitely something we have to put a lot of thought into of like, how do we ultimately at scale? And right now it's not like it's big enough to be like it's eaten up all of anybody's risk budget for us. But at scale, we need to understand how this looks when you have some massive risk that the world has and they need someone to be the capacity on the other side. And that's a good part of the reason why we actually do kind of invite competition to a degree in that like, We want other capacity.

36:55 We want to grow the markets. And we think we'll ultimately actually win to that, even if it means we maybe are competing with more forms of insurance syndicates that are doing some of these things too. We actually welcome that.

37:05 I love the idea that prediction markets are basically sports gambling now, and the vision is they're basically society's hedge against data center risk. Yeah.

37:18 You know, things happen in mysterious ways. It makes total sense. I suppose. But I always thought when I first wanted to start building a prediction markets desk, you know, it's four years ago now. I really did envision this institutional hedging case as the reason to do it. I've got a story I've told a few times now about a friend who's CFO of a musical instrument company that was concerned about tariffs, and that was kind of part of the light bulb in my head. That was the original reason we really wanted to do it. The path goes through sports, because sports has sort of built the space.

37:49 It's created a lot of volume, a lot of interest on it. But I do firmly believe that if prediction markets is ultimately successful- the long-term success will be predicated on this institutional hedging. And that's for ecosystem reasons, because you need a robust ecosystem that has a combination of different types of market participants. And hedging is a really important part, right? If you think about traditional financial markets.

38:14 People are not looking for alpha. Exactly. The traditional market, great source of people who are not looking for alpha are retail sports gamblers, but data center hedgers are a much Right.

38:24 You absolutely do. If your entire market is built from people who they're trying to predict, and there's a reason why that's the first people in predictions. It's prediction markets. People are trying to predict the outcome. You can build something there, but it's not as robust because you don't have the type of variety of reasons people are trading.

38:41 It's a much more robust ecosystem if you have the hedging in there as well. I think there's some hedging now, but I'm not going to go and tell you that that's the majority of the order flow that you see in predictions today. But I do honestly think it needs to be if it's going to become what it needs to become.

38:56 Is it the majority of the order flow in your institutional hedging book slash OTC book?

39:03 In the institutional hedging book, yes. The majority is definitely hedging.

39:06 As opposed to hedge fund alpha trades.

39:08 Yes, for sure. But again, that volume relative to the overall predictions volume is still very small as of today.

39:14 Yeah.

39:23 Thank you.

39:30 Sorry to circle back to sports, but like sports hedging. So some sports hedging is like you're doing a promotion and you're whatever. But like how much sports hedging is sports teams hedging themselves?

39:42 I don't think that there's a lot of sports teams.

39:44 Have you done more than one? Because there's one publicly reported Spanish soccer relegation trade. Have you done more than like three?

39:51 So not all of it is in sort of this current era of predictions either. There's been history of different types of OTC transactions where sports teams might hedge some kind of risk or another, whether it's how far they advance or whatnot. Sports teams don't want to put on a trade that's a direct bet against themselves a lot of the time, and there are reasons for that, partially just kind of optics. But then there's a lot of businesses that are intrinsically tied up with the sports teams, right?

40:18 You could say you have some... network that has media rights and they need a team to make the playoffs or even just like they've got some high marking matchup that's going to happen on Christmas or some key day and like they want the teams to be good. You've got the sponsors, you've got all these mouths that are tied up with it, right? And we're probably seeing more of them involved in saying like, we really want to hedge some of the risks that we have on there.

40:41 The different players that have intrinsic risks tied up in the performance of a sports team or even just a player. And then obviously there's the promotion stuff too, which is real. There's a lot of promotions that are based on sports teams and we're seeing more of them start to say, well, we can hedge our promotion risk. Or even we can launch a promotion if we can hedge the risk, right?

41:00 We want to do a promotion where if you pick all the NCAA March Madness winners, you get a billion dollars or something, but you need someone to actually be the one who's going to take that on and Those are things that we're willing to do, right? You need some mechanism to be able to do that.

41:15 My impression is that sports markets, first of all, are pretty liquid. And second, there's a lot of correlated trades, right? Like someone who's like, I want to hedge my media rights to some sports team. Like you can lay that off in a lot of places. Yes. Does that work by like you do a big trade and then you try to lay it off? Does it work by like you do a big trade and then like there's some dashboard of like how much Knicks risk you have and you're like, Adjust the prices on your flow market making on the next?

41:41 It's probably more the second. I would say relative to most firms out there, we tend to have more liberal attitudes towards risk. We really focus on the expectancy of the trade. We take on some big trade and we have next risk, right? We're going to tune some of our parameters to try to ultimately over a longer, slower period of time, work out of that risk in the market.

42:03 But we're not going to go and sort of just try to get out of all of it now because that's more expensive, which by the way, is exactly what we do in traditional financial markets too. If we get a big risk put to us, we work out of it slowly. You want to paying less to work out of it if you work out of it over a longer period of time. And we're willing to take the higher return for the longer period of risk. And that's firm policy overall.

42:22 Yeah, I guess it's like in general in the prediction market OTC hedging business, like how much tuning of parameters is there, right? So like the goat herder like goes nowhere, right? Like there's no other correlated trade, right? But like when you're thinking of the risk budget for like Texas elections, these things are all correlated, right? You know, if someone comes to you and is like, I want to do a data center moratorium, you're like, I have so much Texas election risk, like we're going to show a bad price on that because like it would, you know, add too much to our risk. Or like you sell a lot of data center moratoriums and you like, adjust the flow market making price on the Texas governor race. I mean,

42:55 think about it the right way. Is it sort of dashboard of factors and correlations that is kind of like feeding into all of these trades or like are a lot of these trades so one-off that you evaluate them on their own?

43:07 So there's definitely a little of both. Obviously, there's plenty of these things that are one-off and there's plenty of them that are short-dated enough that like by the time you would have gone and done anything, the event is over. Mm-hmm. There are certainly some where we have things for longer periods of time. And I'd like to give you a very clean answer, but it's actually kind of a messy problem.

43:25 Part of it is, okay, we got this position. How did we get it? Is there information in that? Should we actually change our fair values is sort of the first question based on this position that got put to us.

43:35 So the question is, is there information in the trade that should cause our fair values to change? Or are we further from home? You don't care if you're further from home.

43:43 We care about both. But the first question when we have a trade is, is there information in this trade? Obviously, if some of these things we've talked about have been, this seems very much like a hedging trade, but not everything that we do is like, this is definitely a hedging trade on the other side. So the first question is, here's a trade. Let's try to understand how this should impact what the probability is of these things happening. Once we've done that, then we're going to look at our risk and we're going to look at it across all the different factors that we have. And generally, we're going to say, okay, we would need to be compensated a little bit more to put on more risk now that we've got a lot of risk on. And we would

44:15 probably be willing to take risk off for zero expectancy right now to reduce our risk. We would do it for the dead fair price, no bid as spread at all. And that's kind of how it works with most of our risk. And it's the same thing here. Now, the factors and parameters are trickier here, right? Because as you've pointed out, You can't just throw it into sort of some market model to spit out all your correlations and tell you what your risk factors are.

44:36 That's just because it's like there's not second by second time series for 20 years.

44:40 You've got a lot of second order effects that happen. You can't just throw a quant model at this. You have to think really hard about how are these things correlated. For example, how is the Senate race in New Hampshire correlated with this? data center moratorium risk in Texas. They're correlated, right?

44:57 They're less correlated than, you know, with the Texas race. And there's no like long history of data that's going to just give you the answer. So that's part of our process.

45:06 In like 10 years, you'll be able to throw a quant model at it, right? Some of it, sort of.

45:10 It'll certainly be a lot easier to just throw a model at it in 10 years than it is today.

45:14 In part because you'll have correlation of like actual events, which is hard because, you know, Senate elections are every two years or, you know, but like But then there's correlations of like the price time series on Kelshi, right? Which gives you some maybe information?

45:27 It gives you some, yeah. And I think that as you get more history and those markets become more liquid, you will absolutely be able to use that information more to sort of back out what some of the correlations are. All that being said, there's a lot of markets that are new also. So many of these markets exist for a period of time and then they go away. And so you still are going to have to figure out how to do the classification of like, these are the buckets, right?

45:49 This is like a legislation contract. It's in California. How is that related to when we go back eight years, like another legislation contract in California? There's definitely a lot of expertise that's required. And ultimately it's figuring out how to make the quant models understand some of the things that are our intuition about knowing that this product is kind of correlated to this other product for these intrinsic reasons.

46:11 It's really just creating a map of what does causality in the world look like? Because that is the thing about predictions. The breadth is phenomenal, right? In Calci, there's over 100,000 markets as of right now. So obviously we don't trade every one of those markets, but understanding how they're all interconnected. And that is why, why is it useful to have a background as a macro trader? Well, it's because I spent a long time trying to think about how one thing impacts another on like a really broad scale. And that translates well to understanding risk and predictions.

46:41 So like you clearly do elections, do legislation, you do sports. Like, is there like, don't do entertainment, do do entertainment. We'll do it if someone asks.

46:50 We'll do it if someone asks.

46:51 Someone came to you and was like, I need $ 5 million on the Oscar winner. You'd be like, Yeah, let's do it for that.

46:58 Assuming that I understood why they wanted to do the trade. Yeah, I think we would be.

47:03 Able to do it. You'd find some price.

47:05 We'd find a price, yeah. Like, Oster's a good example of, like, there's a pretty decent amount of liquidity in the market. Like, those prices are pretty good. And ultimately, it takes less volume than you'd think to get a good price. price signal out of a market. It's not millions of dollars necessary.

47:21 It's probably tens of thousands of dollars, at which point I'm not saying it's perfect, but you've got a pretty good ballpark for what the fair value is after something like tens of thousands and you get more than that on Oscars. So I think like Oscars where there's that much interest, those prices are probably pretty good. We can make a market around it to a hedger and feel pretty comfortable that we probably are making a market around a pretty fair value. We'll do our own research too, but it's a really good starting place.

47:46 What's the right intuition for what a market around it is? So like, let's say that like someone winning an Oscar is at 25%. Are you making like a 24, 26 market for $ 5 million?

47:57 Kind of depends. It's probably a little wider than, than 24, 26, but not going to be 10%.

48:01 Yeah.

48:02 It's tighter than that. It's somewhere in between those. And it sort of depends a little bit on some of the specific dynamics. Like again, you know, we might think they're hedging, but you're never a hundred percent sure. So we're going to talk about all those things and it's sort of a case by case decision, but.

48:15 Yeah.

48:15 Two's a little tight, but might be closer to two than 10. All right.

48:18 What else should we talk about?

48:19 On the institutional front, I do think that probably one of the biggest things that ultimately we will probably see within the next few months is some of these really large trades around some of these big products, like hundreds of millions of dollars. And we actually do see.

48:33 Big products, meaning like data center moratorium?

48:35 Like a data center moratorium, or some of the things that are like, I think you've written about this, like OpenAI and Anthropic have bankruptcy risks that are sort of priced into the market, but nobody can hedge them. Every data center is exposed to one of these massive labs where if one of those- That's interesting.

48:51 To me, that's like OpenAI CDS, right? But there's no CDS, there's no bonds. So there's effectively an OpenAI CDS is you and your prediction markets.

49:02 I mean, yeah. So those markets have existed. I mentioned the moratoriums, but I think that is the two things we hear the most in the data center space. And I know I keep coming back to this, but it's just because it's what we hear over and over again. It's that and then it's the tenant bankruptcy.

49:17 So that's interesting. So like there's trillions of dollars of open AIs debt in the world, but it's not debt, it's leases, right? Right. And the people who are long that debt are data center developers and financiers, right? Right. And they want to lay off some of that debt to like, back when there was a single name CDS market, like, there were people who would get long debt by buying CDS.

49:40 I assume that you are not going to warehouse hundreds of billions of dollars of OpenAI bankruptcy risk.

49:47 Probably not hundreds of billions of dollars.

49:49 I assume that if you're getting into that trade in size, you are somehow intermediating it to someone who wants OpenAI credit exposure and does not own a data center. Is that right?

50:00 I don't know that it's as simple as we're going to find the other side of the trade. It's more likely that we would want to find parties that are willing to partner with us. Like this is one of those situations where there's a massive risk that sort of outsizes everything else that the world wants to hedge. And how can you basically help provide capacity? So we would put on the size that ultimately we're comfortable with, which would be large, but it's not going to be hundreds of billions of dollars.

50:25 Yeah. You say like that the world wants to hedge. Like apparently the world wants to take trillions of dollars of open-ended credit risk, right? Like, well, selling bonds, right? Like someone wants that credit risk. And it's not necessarily that the data center developers want it in the concentration that they have it in, but like, you know.

50:43 Could it be packaged?

50:44 Could it be packaged, yeah.

50:45 It's a great thought. And, you know, we love those types of like innovative things. I don't know that we've gotten to the idea of like figuring out how to package and syndicate and all that yet. We're probably not quite there yet, but absolutely we need to figure out how do we create the structures where we can actually take down some of this risk. Certainly the data centers don't particularly want this. And also I think one of the keys is if you can actually start taking out some of these risks, it makes these projects way more financeable, right?

51:12 Like what we've heard from some of the parties is actually if we can get the right product, we actually wind up getting paid to take off risk, right? Because they get a better financing rate from the banks. Ultimately the banks would be willing to do more of these projects. Like there's a lot of efficiency that ultimately just comes from it. And we've been told legitimately like, If you can get this to work at this right price, it actually will save us money. It's not just that we're taking a risk off the table. So that's where things really start to change is if you can start to sort of create this integrated system where like the banks are saying like, okay, this is what you do. Like they do with other forms of insurance. Like you want

51:48 this loan, you need to have this product that's part of it. And everyone's trying to figure out data centers right now and in AI because you went from nothing to this incredible amount of dollars that's in there. And it's all you hear about in financial circles and it moves the markets more than any other vector.

52:05 I tend to think that like, If the need is to bear $ 10 trillion of Frontier Lab and hyperscaler credit risk, that's like– it's like it'd be Susquehanna.

52:15 No, it's not. It is not going to be us that ultimately is going to take all that risk. It's going to be– It's going to be us that bootstraps it, right? We need to get the proof of concept.

52:23 You're going to run a CDS desk and like– I'm just going to say AIG, but like some insurance company is.

52:31 Going to own– That's absolutely what we'd like. We like taking risk. We'd like to take some of the risk, but what we'd really like to do is to build that market in an efficient way. So we become this cog that helps that market to function efficiently, which is our role, right? We take on risk, but we help the market to function efficiently. And ultimately, yeah, we're working with an AIG or we're working with whichever insurers and other sort of like mechanisms that there are to distribute that risk, right? Like It's funny because a lot of the things in prediction, we're just like, yeah, we'll warehouse it because we're big enough that like most of the prediction things, like we can just be like, yeah, we'll

53:02 take it, whatever. It's millions of dollars. You're big for a goat herder.

53:07 We're big for a goat herder. You're big for a soccer relegation. Right.

53:09 But we're small for the entirety of data center. So we can take down trades that are large, but like I suppose- Gives you a little humility when you're used to feeling like you're the big player until you get involved in data centers. Like, oh, we're looking for $ 10 billion. Like, okay, let's think about how.

53:27 To do this. Jeremy Mallitz, thank you so much for joining us.

53:31 Thank you for having me here. It's been very fun.

53:38 And that was the Money Stuff Podcast. I'm Matt Levine. You can find my work by subscribing to the Money Stuff newsletter on Bloomberg.com. We'd love to hear from you. You can send an email to moneypod at bloomberg.net. Ask us a question and we might answer it on the air. You can also subscribe to our show wherever you're listening right now and leave us a review. It helps more people find the show. The Money Stuff Podcast is produced by Anna Mazarakis and Moses Andon. Our theme music was composed by Blake Naples. Amy Keene is our executive producer and Cheryl Brumley is Bloomberg's head of podcasts. Thanks for listening to the Money Stuff Podcast. We'll be back next week with more stuff.