Jeremy Maletz

Susquehanna’s prediction-markets business began as a side project of its macro team, which used election-style contracts on Betfair and PredictIt for information. Sports was not part of the original plan but now makes up more than half and less than 90% of the business, depending on the season. Non-sports markets are growing faster from a smaller base. Jeremy Maletz sees institutional block trading as a distinct next phase: bespoke trades that combine financial-market pricing with insurance-like risk transfer, including for exposures conventional insurers may not cover.

A California goat herder’s labor-cost exposure illustrated the model. The business paid $50,000 for $500,000 of protection against California failing to change a law that could require around-the-clock minimum-wage payments. The bilateral trade was printed on Kalshi, whose contract listed 12 developments that could relieve the exposure. Because there was no market-implied probability for the risk, Susquehanna estimated a fair price using input from goat herders, lobbyists, other informed people, and its own researchers. Kalshi drafted the contract with input and approval from the firm and client. The law changed, so the goat herder’s position paid nothing and Kalshi received the $500,000 back. Maletz noted that the client could potentially have sold its position after buying at 10 and seeing it trade at 15.

Maletz said Susquehanna seeks a fair price from available information and is willing to take risk, rather than simply balancing order flow. The firm retains roughly 90% of its OTC risk and lays off about 10%, generally working large positions out gradually to limit exit costs. When evaluating a trade, it first asks whether the client’s position contains information that should change its probability estimate, then considers portfolio risk. If it already has substantial exposure, it demands more compensation to add risk; it may also reduce risk at fair value without a spread. The institutional hedging book has dozens of trades, not hundreds, and OTC trades have not so far faced significant adverse selection.

The firm chooses among exchange contracts, swaps, and regulated insurance according to client needs. Exchanges offer transparency and clearing, while swaps can have different capital treatment and insurance may have regulatory advantages. Exchange trades currently require full cash funding because portfolio margin is unavailable, making positions—especially long-dated ones—more expensive. Maletz said funding 95 cents per dollar of exposure would be theoretically inefficient across a hypothetical portfolio of 100 uncorrelated trades, each selling 5%–10% exposure to an unlikely event. Larger trades are more likely to be swaps, with margin negotiated individually. Swaps can use binary triggers and payouts linked to assets such as chips, compute, or power if a reasonable settlement index exists.

Data-center operators have shown substantial interest in hedging local-moratorium risk, and tenant bankruptcy is another recurring concern. A governor’s race in Texas might serve as a proxy for a moratorium risk if the two are highly correlated, but that creates basis risk: unlike a contract triggered directly by a moratorium, an election contract pays based on an event that is only correlated with the exposure. Maletz acknowledged the difficulty of modeling how election outcomes affect businesses and asset prices. Hedge funds may want direct election exposure, while businesses often seek narrower hedges; financial-market firms may prefer broader positions. Short market histories, second-order effects, and the rapid turnover of markets complicate correlation estimates, though more history and liquidity could help.

Developers and financiers may hold concentrated OpenAI-related lease and credit exposures that they would rather transfer. Maletz said shifting data-center risk could lower financing rates and help banks fund more projects, and some counterparties believe a transfer could save money at the right price. Susquehanna is willing to take substantial exposure but would need partners for capacity; it would not warehouse hundreds of billions of dollars of OpenAI bankruptcy risk and has not yet worked out how to package and syndicate that risk. The discussion raised a hypothetical $10 trillion in frontier-lab and hyperscaler credit risk and a potential $10 billion data-center trade as examples of scale, while Maletz said the firm would not take all such exposure. He expects some trades in major products, potentially worth hundreds of millions of dollars, within the next few months. The speakers characterized institutional hedging as important to prediction markets’ long-term success, even though it is not currently most overall order flow.