Meet the invisible sportsbook powering the prediction market boom

By jili tutorial · September 3, 2026

The Hidden Architecture of Prediction Markets

Strip away the jargon of contracts, order books, and event trading, and Bernard Marantelli has a blunt way of describing what’s really happening inside U.S. sports prediction markets. “Kalshi is a sportsbook that’s just not allowed to have an in-house risk team,” says the founder of White Swan Data, one of the specialist firms now making markets on prediction exchanges. The difference comes down to how risk is managed. A traditional sportsbook employs traders to set odds and control its exposure. An exchange, by contrast, provides an API through which competing firms submit prices and supply liquidity. “Here’s an API. Bernard and 88 other people can market-make all these request-for-quotes (RFQs),” Marantelli explains. “Some people might come in and just do esports because they’re esports experts. Others do everything. Some focus on same-game parlays. But it’s a sportsbook.”

This institutional layer is largely invisible to customers, who see a peer-to-peer product. Retail users may technically trade against one another, but the depth required for a mass-market product can’t come from occasional customers alone. Professional firms must be ready to quote continuously and commit significant capital.

How Market Makers Operate Behind the Scenes

White Swan accounts for 40% of activity on some secondary exchanges. Marantelli says the London-based firm is a major market maker on several platforms, with a particular focus on the RFQ (request-for-quote) parlay market. “Just better margins,” he says of that decision. “I think it’s more defendable. It’s the area that fewer people can do well. So I think it’s more defendable margin, more ability to get long-term contracts and beneficial positions.” Singles can be profitable, but parlay pricing requires the market maker to calculate correlations between multiple outcomes and respond dynamically to individual requests. It’s a skill set built over years in the sharper corners of the existing sports betting ecosystem.

Marantelli names White Swan and Susquehanna as two firms operating at industrial scale in parlays, with Jump Trading, Mojo, and DL Trading among the likely leading group. Below them are numerous smaller syndicates, some managing between $5 million and $10 million, alongside sports-specific specialists.

The Rapid Move Into U.S. Prediction Markets

Enda Kendrick, chief executive of service provider Veltium, says the largest UK and European sharp-betting groups have moved quickly into U.S. prediction markets. He believes there are also more than 100 smaller operations — ranging from individual traders to teams of about 10 — interested in entering the regulated U.S. market. Yet the presence of professional counterparties complicates the customer-facing idea that prediction markets simply allow users to trade opinions with one another. As Kendrick puts it, two ordinary customers are not going to place $10 million or $20 million behind the Philadelphia Eagles. Markets at that scale require institutions.

Why Customers May Lose Money Faster

For Marantelli, the exchange format could also cause some customers to lose money faster than they would with a conventional sportsbook. The ability to enter and exit positions creates a perception of flexibility, but that optionality can encourage users to commit more of their bankroll. A customer might buy a team at 55 or 56 cents expecting the price to rise to 58 or 59 cents, he explains. If it falls to 45 cents instead, the trader may refuse to accept the loss and continue holding the position. “Faster, faster, kill, kill,” he says.

“People will lose money faster on exchanges for lots of reasons,” Marantelli adds. “It inherently increases spend, volatility, lots of things. And you’re playing against a sharper audience than you’re playing against at the DraftKings sportsbook.” He compares the effect to sportsbook cash-out features, which gave customers more apparent control over their bets but may have also encouraged greater spending. The crucial difference is that an exchange customer can be facing a specialist whose entire business is identifying inaccurately priced contracts.

A Warning From the History of Betting Exchanges

Kendrick sees a warning in the history of betting exchanges. During their early growth phase, there was enough retail liquidity for numerous market makers to profit. As that retail pool weakened, the sharper firms increasingly found themselves trading against one another. His analogy is a poker table at which the weaker participants sustain the game. If those players disappear, the fourth-best professional at the table can suddenly become a loser because only the three strongest remain.

Sustainability Concerns and the Future

The U.S. addressable market is vastly larger, and customer recruitment remains strong. Marantelli says Kalshi increased its number of clients fivefold during the World Cup, while White Swan predicts that NFL prediction markets could generate between $5 billion and $7 billion of liability in a single week. But he acknowledges the possibility that faster customer losses could eventually test the sustainability of the model. “They lose quicker, dry up quicker, recruitment or re-recruitment,” he says. “If the recruitment of players dries up, then what are you going to do? Definitely there can be components like that.”

For now, the growth provides room for multiple market makers. Marantelli expects margins to “stay good during the growth period” before contracting as competition intensifies. The more complex RFQ and parlay markets may offer the best protection against that compression. The result is an emerging ecosystem that looks less like millions of customers casually trading predictions with one another and more like an outsourced sportsbook trading room. Exchanges own the platform and recruit the customers; specialist firms price the risk and provide the money needed to make those markets function. As Marantelli says: “Let’s call a spade a spade.”