Waterhouse VC hosts a discussion with Bernard Marantelli, Founder of White Swan Data. We talk about the tremendous growth in volumes, RFQ parlays, what it takes to be a top-tier market maker, the competitive landscape, and much more.
Timestamps
0:00 Introduction
1:12 What prediction markets are, and how they scaled so quickly
4:06 Who takes the other side of the retail bet
6:40 Singles vs. parlays, and why parlays carry more margin
9:15 The 1,000-to-1 ceiling: tick size and collateral
12:02 White Swan's route in: singles first, then RFQs
13:59 Three pillars of RFQ edge: capital, technology, pricing
16:28 Competitive landscape
21:07 The hardest parts of scaling, and the $100k mapping error
23:40 Audience Q&A: visibility inside an RFQ
27:31 Collateral netting, and the hedge fund threat
34:44 The World Cup: volumes, patterns, and one brutal night
36:52 The NFL opportunity and the Sunday capital spike
42:03 Beyond sport, and hedging out RFQ risk
51:50 Margin compression, bankroll sizing, and the White Swan competition
Transcript
Introduction and disclaimer
Michael Donohue: Welcome everyone to the webinar. Today we're discussing the growth of sports betting on prediction market platforms, and in particular the trading of sports parlays — or multis — which grew significantly over the recent World Cup period and will continue to grow into the NFL season.
I'm joined by Bernard Marantelli, founder of White Swan Data. Bernard has spent more than 20 years in professional betting, finding edges in various parts of the global gambling ecosystem, from sports betting to lotteries and much else. Welcome, Bernard.
Bernard Marantelli: Michael, thanks very much for having me.
Michael: Before we jump into things, I have to read a brief disclaimer. This webinar is general information only. It is not financial advice. While we may talk about some betting strategies today, it is certainly not an invitation to bet on anything or to invest in anything.
We'll speak for about 20 to 30 minutes before opening the discussion to audience questions. There's a chat box in the bottom right-hand corner of your screen — you can send a question at any point and I'll try to slot them into the discussion, otherwise we'll save them for the end. Only your first name will be shown when you ask a question.
What are prediction markets, and how did they scale so quickly?
Michael: Bernard, you've done a lot of interesting things through your career, and many are covered online in podcasts and articles, so I won't rehash the historical material. I want to focus on the current opportunity you're working on. Can you set the scene for us in market making and prediction markets? What are the prediction market platforms, and how have they managed to grow so big, so quickly?
Bernard: The scene is that we're looking at what is effectively an exchange — very similar to a financial exchange, very similar to Betfair. It's quite interesting that my absolute daily focus is back on prediction markets. That was 2004 and 2005 for me with Betfair, building out automated market making and trading algorithms.
Betfair broadly failed as an exchange, although they grew into Flutter — and we can go into the reasons why that might have happened. What we have now is a reincarnation of that, to some extent, with slightly different wrappers, slightly different regulatory paradigms, and a different time in society as well.
At their core, they are an exchange where you can trade an opinion on anything. Will Donald Trump be president? Will so-and-so win the Emmy? Will the Patriots win the football this Saturday? For listeners familiar with Betfair, it's very similar in its actual backend mechanic and what it does, just presented a little differently — they're called prediction markets rather than exchanges, and you generally trade the percentage chance of the outcome happening, which is a UX decision that may be important in their rise. But you're essentially looking at a similar product in a new time frame and a new package.
They're predominantly succeeding in the crypto world and in the USA. Listeners may not be fully aware, but the USA is a very complicated patchwork of regulations, with fights between state and federal regulators over who controls what. Within that space, the prediction markets have managed to allow sports event contracts, as they might want to call them, to occur in some states where sports betting isn't allowed — predominantly California and Texas. That's given them a big foothold. Add publicity, growth, interest, excellent UX, excellent customer acquisition and topical markets, and that's allowed them to grow significantly over the last two or three years.
Who takes the other side of the retail bet?
Michael: So it's a betting exchange essentially, meaning there are people on either side. They've obviously done a great job of getting a flood of retail punters on there. Can you explain how they've been able to do that, and then explain who's on the other side taking their bets?
Bernard: Look, I'm not the marketing expert, and I understand they spend significant sums. I think they originally got footholds around political voting — mayoral elections in New York, where most of the exchanges are. There weren't a lot of places to do that in a reliable manner, so I think they got footholds there. Then they got into sporting contracts, and obviously the World Cup was just unbelievable growth for firms like Kalshi, Polymarket, ProphetX and Novig. They were going three, four, five, six times client numbers — not volumes, actual client numbers — in that period of time.
Michael: And so who's on the other side of all these retail punters?
Bernard: That's the distinction between a sportsbook and a prediction market. On a sportsbook, the house is almost exclusively the equity holders of the company. They are the house, versus the retail people they recruit. In a prediction market — and there are some subtleties around agreements and arm's length companies — in real terms the prediction market has to be the exchange, and they can't be a counterparty to the bet. So retail come in and bet, and market makers provide the service or the liquidity.
There are broadly two types of bets occurring. The first is a single: will Carlton beat Collingwood on the weekend, yes or no? Carlton's trading at 64% — I'm a Carlton fan — and you're buying and selling at 63 or 64%. Retail can take either side. Retail might even buy at 63 and sell back in five minutes at 67. So in the singles, the distinction between house and retail, and between retail and professional, becomes a bit more blurred.
Singles vs. parlays: where the retail layer disappears
Bernard: When you get to the other main element of a sportsbook — a parlay, or a multi as they call them in Australia, or a combo in the US — these are multi-leg bets. Carlton to win, Essendon to win, Bears to win, Western Sydney to win, in a four-fold. In that environment there's really no retail layer to take that risk.
That four-fold multi might be 10-to-1. You click, click, click, it updates, updates, updates — 10-to-1, do you want it? That 10-to-1 is presented by one of the market makers in the back end. The request comes in, Kalshi or ProphetX send that request to their market makers, all the market makers fire back their answers, the exchange presents the best answer in the UX — the highest price — and the retail person can take that or walk away.
Why parlays carry more margin
Michael: These parlays are obviously a prized product for the US sportsbooks. I think circa 25% of their turnover is parlays, but it's a significantly larger share of their margin, because parlays are a higher-margin product. What's different about a parlay offered by a sportsbook versus one offered by a market maker, in terms of margin and other characteristics?
Bernard: If you look at the American Gaming Association data across the various states, parlays are in the mid-30s in terms of turnover, but mid-50s and approaching 60% of P&L, of GGR, because they're a higher-margin product.
The margin generally comes out of two things: the number of legs you add, because each leg is multiplying up with some margin attached; and the price. If the price is larger, the retail person becomes less price sensitive, and the house or the market maker probably wants a larger margin to reduce their own volatility. Package those together and you tend to get an escalation in margin as you get an escalation in price and in the number of legs.
The 1,000-to-1 ceiling: tick size and collateral
Bernard: There's not a lot of other difference until you get into the technical structure of the exchange. The exchange has a minimum contract price of $0.001, which means you can't offer a price bigger than 1,000-to-1. A lot of parlays on DraftKings, FanDuel and bet365 go through at a much bigger price than that. Retail people are looking for that lottery-type ticket — parlaying up 10 or 12 games at 40,000-to-1 for a tenner, looking to get $400k and change their life. So there's not much difference if you look at the 2-to-1 to 20-to-1 range, but there are components that give the sportsbook parlay a longer duration and a longer price than the exchange parlay.
Michael: What share of parlays do you think are those lottery tickets versus the rest? How much of the market is cut off from market makers on prediction markets, versus the sportsbooks?
Bernard: Firstly, let's discuss why it's cut off. It's cut off for two reasons. One is the minimum cent increment in the contract size — if you wanted to allow a contract to trade at a ten-millionth of the payout, someone could have a 10 million-to-1 parlay. The second is that when that 10 million-to-1 parlay transacts, a market maker has to supply $10 million of capital, because these are exchange contracts and they need to be fully collateralised before the transaction occurs.
So there's a technical reason — the contract size — and a collateral reason. There could be examples in the future where they net the contracts and allow those to be offset. At the moment, when someone wants a million-to-1 on a parlay on DraftKings or bet365, nobody stops to ask whether they have the capital to pay that parlay. It's based on balance sheet. Everyone assumes they've got enough balance sheet and pretty good risk management, and that it's not going to happen.
Having said that, there's a lot of volatility in pricing and more so in results. In Q4 2025, both DraftKings and FanDuel issued profit warnings around the $450–500 million mark, and that came specifically from adverse NFL results at the beginning of last season.
White Swan's route in: singles first, then RFQs
Michael: White Swan has been market making on Kalshi and other platforms for some time now. What do you need to do it — what do you need to have?
Bernard: Just to be clear on timelines: the prediction markets have been growing or in their infancy since about 2017–18, and became substantial around late 2024, early 2025. That's for singles. The RFQs — request for quote, which is the parlay or the multi — I think they only entered the market around last September, and only really got out of full beta into much more widespread use in about January of this year. Then, as we mentioned, they completely exploded during the World Cup.
We market make some of the singles and underlyings. These are complex places with lots of traction and lots of people doing it, and to be perfectly clear: less margin, more sharks, more problematic technology requirements and problems.
The RFQs are more complicated on one level because they're parlays. A lot of the risk is correlated; a lot of it can be negatively correlated. Your liabilities can accelerate rapidly if you don't have good risk management. We focus on the parlays because they've got a higher margin and a higher complexity, and we think we've got a greater edge the more complex something is.
The three pillars of edge
Michael: And what contributes to that edge?
Bernard: In a world where there are no market makers and there's an avalanche of retail, some elements are less important. Edge becomes, in some senses, technology — because really you just need to be live. If you're live and on the screen, you're going to take a fat margin.
As margins compress, or as smart people enter the market, you need fundamental pricing. Not just good technology, but making incremental gains over the other market makers through better pricing. They might be making 12% on turnover and you might be making 13%. That allows you to grow market share, compete, and handle more volatility. There are lots of reasons why incremental basis points between you and the market are super advantageous, particularly as margins go down.
And then of course you need capital — immense capital, because you have to collateralise everything. Let me explain how extreme that requirement is.
The collateral problem, illustrated
Bernard: Imagine head-to-head football, or point-spread football, which is really a coin toss. Someone comes in and says: on the next three football matches, I want home, home, home. I say great — fair value, 7-to-1, I'll give you $7 million to $1 million. Then someone comes in and says, I want home, home, away, and I offer them $7 million to $1 million as well. I have to put up $14 million of capital, despite the fact that I can only lose $6 million between those two transactions.
Add a third transaction — home, away, home. Again, completely mutually exclusive. I'm now at $21 million of capital, and in a worst-case scenario I can only lose $5 million.
So the capital requirement becomes a major moat for those with capital. There are lots of things you need, but I break it into three pillars: you need significant capital, you need outstanding tech, and you need very good underlying fundamental pricing — particularly correlated pricing.
How many firms can actually do this?
Michael: As someone a little on the outside of this part of the industry, it seems opaque who's in there competing. Do you have a sense of how many firms are out there with those pillars, and can do these things properly, like White Swan?
Bernard: I have some sense, because you see the participants in the market, and when you're in dialogue with prediction markets about coming on as a market maker there are indications of how many people they have. I think you're talking, in a lot of places, in the region of 20 to 80 firms quoting on an RFQ, from a smaller exchange to a large exchange.
Obviously there's a massive range in there, from very big firms through to what they call garage bands — a couple of kids sitting at home who might have some good pricing, who might have used a bit of Claude coding and vibe-coded something up, trading half a million bucks. Some of them will go broke and some will grow and become successful. A lot of that delta between success and failure will be volatility and risk management, for sure, and some of it will be pricing and technology.
Does latency matter in an RFQ?
Michael: Just on the tech — how quickly do the platforms require market makers to respond with a quote, and is there the ability to get significant edge there, or is it more of a table stakes thing?
Bernard: There's ability to get some edge. Think about the UX on a sportsbook in Australia: you click Carlton, Collingwood, Essendon. As you click Carlton it tells you 3.1; as you add Essendon it tells you 6.1. That generally doesn't change — it might change if the underlying sportsbook changed the price.
What happens on most prediction markets is that you click Carlton and Collingwood, and you add Essendon, and it says 6.0, or maybe 6.2, because that was the best offer they had. Then another market maker who was slightly slower from a technology perspective adds 6.3, and your price bumps up. You transact at 6.3 because you were happy. But if you left that UX open, it might go 6.3, 6.4, 6.3, 6.2 and jump around a little as market makers change their offer, cancel it, or let it time out.
The general process is seconds, not milliseconds, as it would be trading a single underlying. If you're just trading Collingwood, milliseconds completely matter. If you're trading an RFQ, it's still milliseconds, but you're talking about 500 or 800 — half a second, one second, even two seconds — that don't really impact your ability to be competitive.
The hardest parts of scaling
Michael: You said February was when you really ramped up the RFQ market making.
Bernard: February is when we got into it. The technology build, even for people like us who already have parlay pricing, outbound pricing, risk management engines and staff, was not instantaneous. You've got to go and get the APIs — I think we're live on seven RFQ platforms now. You've got to understand the feeds, the answers, the market mapping, the overs and unders. Some markets refer to over, some to "under, no, yes," which really means over. So there are mistakes to be made there.
And you've got to realise that the supplier of that technology — the exchange — is building out at the same time. They're in a market share race to be live with new products, so things are coming out that are, to be perfectly frank, advanced beta on some exchanges. You're working with growing technology on both sides. So yes, I think we were first live on the RFQs around February, and on singles last August or September.
Michael: What have you found the most challenging aspects of scaling that strategy?
Bernard: It's just the amount of things you have to do at once. Our tech team is excellent — we've built to seven platforms in the last four months, and we have a time span of getting new platforms onto our technology in less than two weeks. But the scale, the growth, the amount of data that's produced — how do you digest that data, use that data? How do you suck everyone in your business into this part of the business?
They're your typical scaling challenges. Where do you repurpose people from in other parts of the business? You want to go out and hire someone new to help you scale, and they're on a three-month or six-month non-compete, so they won't join you until next year. A lot of the manpower has to come internally, particularly at the skilled level, and that creates challenges in other parts of your business.
No palpable error: the $100k lesson
Bernard: The technology also has to work 24/7. There's one thing with prediction markets. If DraftKings accidentally offer you 20-to-1 on a parlay that should be 2-to-1, they're likely — and not just DraftKings, bet365, everybody — to declare what they call a palpable error, and say the price was so wrong it couldn't have been believable, therefore the bet is void.
You don't have that protection on an exchange. It's a financial contract. If you write that contract wrong, you pay that contract. We did that one day early on. There's still some confusion about where it originated, but an overs market was mapped to an unders market, and you're giving 4-to-1 on an alternative over that should have been 1-to-4. Of course, you write the contract, they got the overs, the result came the same way, and we dusted off $100k or something like that. It happens.
That's one element that means your technology has to be incredibly good, because there's no fat finger, no trades unwound. You ride it, or you're dead.
Audience question: Can you see the market an RFQ creates?
Michael: We've got a question in from David, who asks: do you have visibility of the market being dynamically created by an RFQ — for example, the prices and liability offered by other market makers?
Bernard: Broadly, no — just your own. You have some visibility about where they had transacted. Say a retail person comes in and asks for $10 on a treble, and then 40 seconds later someone asks for $100 on a treble. You might assume that's the same person increasing his stake, but it might be a new entrant. You don't know. You know that you were unsuccessful, but you don't really know whether you were unsuccessful by a penny or by a lot, and a lot of the time you don't know that live, in a way that would let you escalate your bid. Some of that information becomes available after the trade, so you can be more alert next time around.
But in real terms, the RFQ is not like a visible, clean order book where you see 62, 63 and how much money is in each price bucket. And let's be clear — even on a single, when you see 62 and 63 with a thousand here and 500 there, that is a millisecond snapshot they publish to you. Even if they're updating every 30 milliseconds, there are 40 transactions going through in that time. Those things are really snapshots. If you're using an API call, or third-party trading software, you have the snapshot with full visibility, but those snapshots are updating so quickly that they're a representation of what was there at that time.
You have to trade that very quickly, and you have to infer some of that information — and some of it you only get after the trade, which may allow you to reconsider your price for that trade later. It's the same as a single. A single can be 62–63, you say you're a backer at 62, and as you send your order to the market you get matched at 64, because between your visualisation of 62 and your bet arriving, someone moved the market up to 64 cents. Obviously we know price priority there — the person coming into the market takes out the order book. But these things are ultra, ultra fast.
Audience question: Do you request quotes yourself?
Michael: Another one on the mechanics of the RFQ process. Sean asks: do you request quotes in the hope that other market makers get it wrong, or do you just concentrate on the lay side of the bet?
Bernard: I think a lot of people will request quotes. But you've got to contemplate the value in doing that versus the value in growing out your tech stack, integrating more markets, more people, more exchanges, better pricing, better ecosystem knowledge. Anyone who gets it wrong — properly wrong — is going to fix it quickly or go broke quickly. So I think that opportunity is more limited than focusing on the other side of the business.
Will the CFTC relax collateral requirements?
Michael: Going back to capital requirements, since it's obviously a big deal that you need to collateralise all of your liabilities. The CFTC regulates these platforms — do you think they'll look to change those strict collateral requirements, or is that how it's going to look for the foreseeable future?
Bernard: I think it's definitely going to look that way for a year or so. It's not ultra easy to net off. Remember, there could be 3,000 markets on an exchange on a Saturday, and you could parlay any of those 3,000. Most of them are head-to-head markets, but that's technically two-to-the-300 probability trees. A massive space, a ridiculous space.
Of course they're not all going to be used — nobody's going to take a 3,000-leg parlay. But there are five, 10, 15, 20-leg parlays, and they're all independent from each other, in that retail selection is independent. So you end up with a gigantic spiderweb of risk that you've got to digest to understand your own personal risk — and then you've got to digest it further again if you even wanted to imagine netting those from a capital requirement perspective.
I don't think that gets solved quickly. And I don't think there's enough upside for a regulator or an exchange to solve it, because if something went wrong with the netting process, who's left holding the baby? Bill and Ted from the garage band net wrongly, they write off tens or hundreds of millions of dollars of liability they don't have — who's carrying that?
So I think it'll be slow and cautious, and I think there are participants in the market with capital who won't even want it done, because that's a moat for them. I'm not in the dialogue between the exchanges and the CFTC, but I'd be extremely surprised if it happened this year.
Can the big hedge funds price you out?
Michael: That leads to my next question, around the big hedge funds — who I'm guessing are some of the people with a vested interest in keeping those capital requirements high. What's to stop them coming in and pricing out players from the traditional wagering world, like White Swan? They seem to have significant capability. Are they going to get there?
Bernard: We know some of them are there, some of them are definitely entering, and some of them are looking around buying syndicates out for their technology and their pricing technology.
In the big hedge fund you have infinite capital, or broadly infinite capital. Very good tech for trading single markets, because they're already market makers on crypto or financial exchanges. Diminished tech, or developing tech, for the RFQ market. And they've never really been involved in pricing the correlation of sports bets, and never really been involved in avoiding sharps in sports betting — they might be avoiding sharps or informed traders in the stock market, but they don't have fundamental pricing.
So against the three pillars: they're 11 out of 10 for capital, maybe eight out of 10 for RFQ technology and growing quickly, and maybe three out of 10 for fundamental pricing — which might not be essential now, but is going to become essential in time. They're probably going to have to buy or grow that third pillar, and growing it will be slow. It's a couple of years to get even with money, to get somewhere near where White Swan Data is. So then you say: do I want to grow that over two or three years, or go and buy it and provide the capital?
And if you look at someone like White Swan Data, we're nine out of 10 on pricing, nine out of 10 on RFQ technology, and four out of 10 on the capital these people have. So we've got to improve the capital end, and they've got to improve the pricing and technology end.
Exchange-affiliated market makers
Michael: Some of these market makers are affiliated with the exchanges. Have you seen any advantages afforded to them?
Bernard: If there were advantages afforded to them, I probably wouldn't know what they were. I think it occurs, and it's the regulator's role to make sure everyone has free and open access. A number of them have their own in-house market makers — and when I say in-house, heavily affiliated companies market making. It may well be that at some stage the regulator says: hey, exchange 27, your in-house market maker is only allowed to market make on every other exchange, not your own.
They'll still have business opportunities for those individuals and groups. If you look at the traditional sportsbooks, they've got in-house risk teams that have run the parlays for the last eight or 10 years, since PASPA. Now they open a prediction market to get access to the states they can't reach with their online sportsbook — or because of the hubris of everyone saying, I've got to get on a prediction market rather than a parlay, and to the retail press there's not really a difference. So I think they're giving away some of their customer value by wanting the presence and competing in the prediction market space.
But maybe these companies end up running prediction markets in states where there's no sportsbook, and sportsbooks where there is, and they end up with these super apps — sweepstakes in some places, casinos where it's allowed, sportsbooks where it's allowed, prediction markets where it's allowed — but they still get some coverage in almost all 50 states.
Audience question: World Cup volumes and patterns
Michael: A question from Oscar about the World Cup. During peak periods, how many requests were you receiving? Was most of it pre-game or in-play? And what patterns did you see in the teams, players and outcomes punters wanted to back?
Bernard: Most of the parlay count, handle and risk is pre-off. The teams they wanted to back were everybody who won, and the players they wanted to back were everyone who scored.
The two major markets through the World Cup were the superstars to score — Haaland, Messi, Ronaldo, Kane, Mbappé. There were just avalanches of money for them to score every game, avalanches of money for those teams to progress, and avalanches of money for that key double. I'd say that was the single biggest thing through the World Cup: superstar player to score, followed by his team to win or progress — when we got to the progression part of the market.
Michael: A lot of those bets got up during the tournament. How did White Swan fare over the length of the World Cup?
Bernard: The length of the World Cup was good. But there was one day where, in game one, Mbappé scored and France won. Game two, Haaland scored and Norway won. Game three, Messi scored and Argentina won. Game four, Mexico's leading scorer scored and Mexico won. Those were the four games one night. I think the market did about 20% of its balance sheet — the global market doing RFQs, that day. Probably mid-teens of their balance sheet.
How the NFL differs from the World Cup
Michael: Looking ahead to the NFL, which is obviously the next major period for RFQs — how does that opportunity differ from the World Cup?
Bernard: I think it's going to be several-fold. It's very hard to imagine that $700–900 million of daily RFQ liability on Kalshi during the World Cup — where Americans don't really bet soccer, or until now have not bet soccer — won't translate to four, five, six billion dollars of RFQ volume, risk and liability on an NFL Sunday. So I think volume is going to be four or 5x again.
But the NFL is unique in that there are only about 17 regular weeks of about 17 games. It's widely known as the most volatile product in the American sportsbook ecosystem. The NBA has about 1,400 games a year, four or five most nights, five or six nights a week — I'm not an expert on the schedule. Whereas the NFL is 16 games on a Sunday, one on Thursday, one on Monday. You're spiking your volatility into a low number of games, and into a single day, which means you're spiking your capital requirement significantly.
Those are the unique elements: it's all condensed on a Sunday, it is the biggest betting medium per game in the US by several-fold, and it's all on the same date.
Capital lock-up and the weekly rhythm
Michael: Does that have an implication for the amount of profit you can make, given there are fewer game days and you lock your capital up on a day basis rather than a game basis?
Bernard: Yes. If you look at something like Wimbledon, nobody's trading Thursday's tennis on Tuesday, because they don't even know who's playing yet. So at Wimbledon you trade today, and almost nobody is taking a parlay from today into tomorrow — they take parlays today, then they take parlays tomorrow.
That's not really going to happen in the NFL. You might start laying parlays on Wednesday, Thursday, Friday, but they're not going to resolve until Sunday. So your capital requirement for the Sunday is going to spike, and that's going to be your maximum capital requirement for the whole year. NFL Sunday is the maximum requirement you will ever have on a daily basis — a quasi-weekly basis, because you grow into it.
You'll also be able to do the college basketball, the college football and the EPL on a Saturday, because they're all running concurrently. So you'll have pretty good and consistent volumes during the week, that possibly grow and spike if the Americans keep betting European soccer — which they might not — then grow into college football, which I think is on a Friday night, then grow into Sunday NFL, which will be gigantic, and then settle down again on Monday through a bit of UEFA and NBA. And when I say a bit of NBA, the NBA is going to be hundreds of millions of dollars of liability, but it might not be billions. So you'll have this weekly process of maybe a billion to two billion a night during the week, and six billion on a Sunday.
How much of the market does White Swan want?
Michael: And of that six billion, how much does White Swan want?
Bernard: In a world where you want to beat the market, you can't be 80% of the market and beat the market, because you are the market. So we're probably looking to be 10-ish per cent of the market. There are going to be a couple of people who might be 20 or 30% of the market, but obviously there can't be many people at 20 or 30%.
And look — numbers are going like this. Market makers are growing, regulations are changing, clients are coming in, capital requirements may or may not change. There is an absolute ton of variables being resolved literally on a daily basis at the moment. But that's our current objective, to be there in around that size.
Life beyond the NFL season
Michael: My last question before we turn to the audience: the opportunity beyond the NFL season. A lot has been spoken about the NFL season and the opportunity there. Is it still an exciting opportunity for you outside of it?
Bernard: Yes, of course — and again, there's so much variance here that we don't know how this resolves. We're tracking about 125 prediction markets at the moment that are live, going live or in development. On top of that you're going to have the layers: routing layers, technology layers, pricing layers, trading layers. So there are probably a thousand companies trying to generate some sort of edge and business in this space.
Imagine the NFL is growing, and the NBA and the soccer and everything is growing underneath. When the NFL finishes, it definitely falls down, but it falls down to an escalating base as well.
We do a lot of risk management for all types of things. We've done risk management for weather, for airports, for snow, for cancelled flights, for promotion and relegation of football teams. We've done risk management for pubs that don't sell beer if England get knocked out of the Euros — they want an insurance product, for lack of a better word, so if England win the Euros or reach the semi-final or the quarter-final, in a tiered structure, they pay more. And if England go out early? They're smoothing their beer sales. Restaurants will hedge game seven of the Knicks versus the Spurs, because they've got to go and buy $3,000 of meat and vegetables to run their restaurant, and their restaurant is going to have 10 people or a thousand people depending on whether it's a Knicks night.
So I think these exchanges are definitely going to be very sports-centric to begin with, but I think they end up being: we hedge everything. Humans generally don't like volatility — except humans like me, we love volatility. We could be pricing more weather, footfall traffic, snow, promotions and relegations. I see worlds where you end up trading contract sizes of football players and soccer players, where you can own 1% of a synthetic — Wemby, or whatever his name is — where you can actually own a percentage of him in terms of his asset, his growth, his income. So I think we've not really seen past step one of what these exchanges are going to be allowed to trade and normalise as a trading product.
Audience questions: can you hedge an RFQ position?
Michael: Turning to questions submitted by the audience. I've got a couple of related questions around hedging. John asks: can you exit or manage RFQ positions before settlement? And Theo has asked something similar: what percentage of parlays do you hedge out?
Bernard: You can hedge, but your ability to hedge is complicated very much by whether the legs are concurrent or not — which means the NFL is the most unhedgeable parlay, because nearly all legs are concurrent. The slates are all overlapping and the games are long. Yes, if you write a treble across one Tuesday night game, one Wednesday night game and one Thursday night game, you could write it and hedge it out. But most parlays occur in some overlapping set of events, because people want a parlay that resolves in the next three or four hours, not the next three or four days.
So you could hedge, but I think the real requirement here is to make sure that before you write the risk, you're happy with the risk — rather than writing the risk and then going to hedge it somewhere.
Michael: That sounds like a big part of your strategy.
Bernard: It is. And there are significant transaction fees on hedging as well, because of the pricing for makers and takers in an RFQ world. Predominantly, at the moment, there's no cost for a maker, because they need the liquidity. But if I went and hedged it in a single I'd become a taker, and there would be some fees, probably to be negotiated. So in real terms the correct approach is to make sure you don't write it unless you want it.
Audience question: why choose Kalshi over a sportsbook?
Michael: Olivia asks: why does a punter choose Kalshi over the traditional sportsbooks like FanDuel? Is Kalshi growing the total pie through accessing markets where sports betting is banned, or is it a better experience such that people are switching?
Bernard: It's funny — as a professional gambler, I'm not the right person to ask whether something is a better experience, because my interest in the front end of a UX is that I just want to get to the bet that is the wrong price.
They're definitely growing the pie in two ways. They're growing it by going to markets that don't allow sports betting. I'd say they're growing the global pie in those markets, and they're certainly cannibalising some of the black market in them. And I think they're growing the pie through access, because a lot of the financial platforms — Crypto.com, Robinhood and so on — are saying, well, we trade more or less everything except sport, let's trade sport as well. That's opening sport to some people who wouldn't be bothered logging out of their Robinhood account and going into a FanDuel or a BetMGM account.
These conversations are mostly in an American context for the moment. But yes, I think they're growing the pie in terms of geography, and in terms of access to client bases that weren't traditionally looking at that space.
Audience question: what happens when FanDuel and DraftKings become makers?
Michael: Luke asks: how does White Swan's edge hold up against FanDuel, Flutter and DraftKings entering as makers with their own decade of proprietary parlay correlation data?
Bernard: It's very hard to know, because when we offer 6.2 and somebody trades at 6.3, we don't really know if that's Bill and Ted in their garage or FanDuel. We don't have good visibility on that.
What we do have is some visibility through published data about what the ROI or margin is for FanDuel and DraftKings on a state-by-state or global basis, and some information about what our edge is on a prediction market. The edge for a market maker on a prediction market is generally a few per cent lower, because the prediction market is adding on a few per cent as well, and there will be some competition between a person getting $15 for a parlay on BetMGM, $15 for a parlay on Underdog Fantasy, and $15 for a parlay on Kalshi. The market maker will be a little bit cheaper on Kalshi, because they're adding a fee, than the in-house market maker on, say, Underdog — and Underdog just announced they're launching their own prediction market, so that dynamic will change a little.
I'd say we are their equals for sure, though it's a little hard to understand fully. They have only sat behind retail traffic for their decade, whereas we have generally been an outbound predator of those markets for the last decade. They sit behind a 15% retail edge, let's say, as a broad number, and we're generally betting into parlays against a 15% edge. Now, beating a 15% edge is not the same as being the book, but it goes a long way to being reasonably confident that our pricing is extremely strong. And if we look at our performance in the market so far, it's fairly clear that our performance is very strong.
Where do RFQ margins settle?
Michael: Where do you think the level of margin in RFQs settles longer term?
Bernard: It's hard to know. We haven't seen a lot of compression in parlay margins in traditional sportsbooks, so I don't think it settles quickly. But the ability for 50 or 100 people to market make obviously creates some pressure. I think the exchanges will want some pressure on price, because they'll want to migrate clients from a traditional sportsbook to an exchange through better UX, better pricing, or a better super-app experience.
I think it'll traditionally run a couple of per cent lower than a sportsbook for those reasons. I don't think there's anyone interested in deploying $10 billion to Kalshi and crushing the parlay market to 2%, because they're going to be sitting on a ton of volatility, which doesn't make that a responsible decision — even if you're Berkshire Hathaway or Elon Musk, unless you just want to do it for laughs.
So I think it will stay there, because the individual actors should be rational and shouldn't want parlays trading at those levels. And when I say parlays, there's a spectrum here: a money-line two-fold will go through at 2 or 3%; a 15-fold should never be going through at 2 or 3%. But in broad terms, if you chase the margin down to 4 or 5% on a six-leg parlay, now you get pros entering that market. Whereas at the moment, the margin protects you against not all pros, but some pros coming in.
Audience question: College football and the expanded slate
Michael: A question from Ben. With college football and the expansion of conferences like the SEC and Big Ten, and games played on Thursday, Friday and Saturday, how does that affect the market? Is that a significant part of the RFQs?
Bernard: I caveated before that I'm not an expert — I got in trouble with Brad Allen the other day for calling the Knicks versus the Spurs the World Series. So it's safe to say I'm not a sporting expert.
All of these things will grow. Nothing will replace the NFL in the mid-term future of several years. But if you've got limited capital, you'll have to say: hang on, I've got $4 million, I can't trade NFL until Sunday, because with $4 million I can do it all on college football, then all on European soccer, then all on NFL Sunday. That person with limited capital is going to have to say: I've got a request for a Sunday parlay on Thursday — I'm only going to write that if I get a 25% edge, or I'm going to write it on Sunday at a 12% edge. So there's this whole business of maximising the ROI of your bankroll, in an environment that is constantly evolving.
That's where you get back to the infinite requirement for data, the amount of data available, and literally being able to keep pace and digest all of it to make better and better decisions. It's a gigantic problem, and the most attentive, diligent people are going to win the game.
When you look at a lot of the smaller people, they've got their limits, because they're going to have to integrate to 10 counterparties to get access to the 10 markets. If you're a couple of kids at home, integrating to 10 people is going to cost you $300,000 — that's 30% ROI on your million-dollar bankroll. You can vibe-code one or two integrations and have bad reporting and bad data, but bad reporting, bad data and bad code are all going to turn into significant disadvantages over time.
What's the optimal bankroll?
Michael: What do you think that level is — not only the operational backing you need, but the bankroll level where you've got enough to be on all the platforms and lay all the bets you want to lay, but not be so big that you become the market? Is it 50 million, 100 million, 500 million?
Bernard: If someone said to me, in the NFL — let's say it gets to $5 billion — Bernard, you've got $50 million, I'd have a super nice ROI, because you could be extremely selective about what you were doing. But you're 1% of the market, and that would be suboptimal in terms of your GGR, or your EBIT, or whatever term you want to put on it. ROI when you're small is easy to keep high, because you can be very selective, but your cost of operating can be quite high in those environments.
So I think your sweet spot, where your cost of technology, development and staff is proportional to what you're doing, is $20 to $30 million upwards, and probably $200 to $300 million — in a world where it's a $5 billion Sunday market.
The White Swan trading competition
Bernard: And that's what — I'm sure you saw, Michael — we're running a competition at the moment, looking for some of these smaller groups and allowing them to come in under our umbrella, sit inside our technology, sit inside our risk management and reporting layers, and just trade and just price. The winner gets a $3 million trading balance on Kalshi, and the runners-up a $1 million trading balance on Kalshi.
I think there are hundreds of people coming up in that area, where they've got a bit of friends and family money and a bit of tech. Some of them will go broke, and some of them will become significant players in four or five years' time.
Michael: So you're giving them the bankroll and a bit of tech to couple with their pricing.
Bernard: Tech, risk management, consulting overviews — all the things that allow them to scale quickly and remove a lot of the cost base of reaching all of those markets. We've already consumed that cost base. There's no need for other people to write it again, particularly if they're our partners.
Audience question: Prediction markets in Australia
Michael: A couple more from the audience. Mark asks whether you have any views on prediction markets becoming a legal product within Australia, or other legalised markets around the world.
Bernard: I'm not really the regulatory expert. It's crazy — I see half of Europe saying prediction markets are banned, and I'm thinking, how are you really banning them, because they're already permitted in Malta, Gibraltar and the UK.
As I think I've said a few times, the scale at which these things are growing means I'm so tied up with technology, pricing and ecosystem that I'll let the regulators work out what's allowed and not allowed. Exchanges are already in Australia — Betfair's been there at least 10 years. I think people will chew at the edges of what's allowed around the world, and you'll have grey markets and you'll have white markets. But broadly, I'd say prediction markets will become popular everywhere.
Wrap-up
Michael: We've just hit the hour mark and I think we've got through most of the audience questions, so we might wrap it there. Bernard, thanks so much for your time — very appreciative of the masterclass in market making today. And thanks to everyone who tuned in. We'll share the recording soon and welcome any further enquiries about the topic; feel free to email either Bernard or myself.
Bernard: Thanks, Michael, and thanks very much for having me. It's always good to get on a podcast where you're talking not just to an informed individual, but someone coming out of an informed group, with the Waterhouse VC group and their experience, knowledge, posts and reports. It's good to be in an environment where you're getting interesting and tough questions, but very informed questions, from an informed group. So I've had fun — and as you say, anyone who's got questions, write them to you, flick them on to me, and hook me up with anyone who wants more information.