A contract that pays a dollar if an event happens and nothing if it does not is a derivative. It is also a bet. For most of financial history those were different regulators, different venues, different customers. In 2026 they are the same screen, and that screen has order books.
Prediction markets stopped being an election-night novelty somewhere in the last eighteen months and became a venue class with exchange-grade plumbing. The numbers force the point. Global prediction-market notional reached roughly $25.7 billion in March 2026, and Bernstein projects close to $240 billion for the full year, around 3.7 times the 2025 total. Polymarket alone cleared $10.6 billion in notional in March, six times its level six months earlier on Dune Analytics figures.
$240B · projected 2026 prediction-market volume (Bernstein) $22B · Kalshi valuation, on roughly $1.5B annual revenue 12B · event contracts traded on Robinhood in 2025
Capital is pricing it as durable, not faddish. Polymarket is raising near a $15 billion valuation. Kalshi already sits at $22 billion on roughly $1.5 billion of annual revenue. Intercontinental Exchange, the owner of the New York Stock Exchange, has committed close to $2 billion to Polymarket and now distributes its probability data through the same Consolidated Feed that carries NYSE equity prices. Robinhood traded more than 12 billion event contracts in 2025, its fastest-growing line by revenue, then stood up a CFTC-licensed exchange and clearinghouse through a joint venture with Susquehanna. When a quant prop firm of that size buys an exchange, it is not planning to click Yes on the Super Bowl.
The category line dissolved
One caveat frames everything that follows. The regulatory detail here is almost entirely American, because the US is the only major market building a federal home for these instruments instead of banning them. Polymarket won full designated-contract-market status in November 2025 and reopened to US users, and the CFTC has event-contract rulemaking underway after taking comments in April 2026. Everywhere else leans the other way. The EU has no dedicated prediction-market framework and falls back on ESMA's 2018 retail binary-options ban, with France, Germany, the Netherlands and Portugal treating the platforms as unlicensed gambling, and Portugal gave Polymarket 48 hours to close in January 2026. The UK's FCA banned retail binary options in 2019, so Polymarket geoblocks British traffic. Malta and Gibraltar are the only jurisdictions drafting a purpose-built regime. The deepest liquidity and the institutional capital both sit in one country, which is also where the law is being fought in real time, so a serious read of this asset class is mostly a read of US regulation.
The legal system cannot agree on what these instruments are, and the disagreement is the story. Kalshi argues its sports contracts are swaps under the Commodity Exchange Act, federally regulated, and therefore beyond the reach of state gambling law. On 6 April 2026 a divided Third Circuit agreed two to one in KalshiEX v Flaherty, the first federal appellate court to hold that the CEA preempts state gaming statutes for sports event contracts traded on CFTC-registered exchanges. State regulators in Massachusetts, Nevada and Tennessee read it the opposite way. The CFTC and the Department of Justice sued Arizona, Connecticut and Illinois on 2 April to assert federal primacy, while senators introduced the Prediction Markets Are Gambling Act on 23 March. Prediction markets themselves price a roughly 64 percent chance the Supreme Court takes a case before year-end.
That ambiguity is not academic. It decides whether a football result is a security-adjacent derivative or a wager, and the answer governs who may list it, who may clear it, and who may touch it. While the courts argue the taxonomy, the rails already merged. DraftKings runs a CFTC-licensed exchange through its Railbird acquisition. FanDuel partnered with CME to list exchange-traded sports contracts. Robinhood places event contracts beside stocks, options and crypto inside one app. The same wallet that holds tokenized exposure and USDC now holds a position on a rate decision and a quarterback's passing yards.
A football result, a Fed decision and a Bitcoin price bracket now clear through the same kind of binary contract, on the same kind of book, sometimes funded from the same wallet.
A $150 billion target
The biggest prize sitting in front of this is the sportsbook. US legal sportsbooks will clear something north of $150 billion in handle in 2025, and prediction markets are aimed straight at it. Citizens estimated that the spread of event-contract platforms across all fifty states has already pulled roughly five percent out of regulated sportsbook handle, on the order of $8 billion a year. Bank of America downgraded both DraftKings and Flutter, the parent of FanDuel, citing thinner margins and competition from prediction markets in the same note. The two incumbents, which together hold close to 80 percent of US betting handle, responded the only way they could, by launching their own event-contract products while PrizePicks and others crowded in behind them. That is not a moat being defended. It is a business model being conceded.
The disruption is structural rather than cosmetic. A sportsbook is a house. It sets the line and takes the other side of every wager, keeping a percentage of everything that runs through it as hold. DraftKings booked a sportsbook net revenue margin near 5 percent last quarter, which is the cut the house keeps. A prediction market is an exchange. Participants trade against each other at a market-clearing price, and the venue earns a thin fee or a spread instead of booking the bettor's expected loss. For anyone holding a view, the exchange is simply a better price, which is the same reason trading migrated from bucket shops to real order books a century ago.
The exchange also wears a federal coat. A sportsbook needs a license in every state it touches, and DraftKings is live in twenty-five of them, covering about half the US population. An event-contract platform registered with the CFTC claims the right to operate in all fifty at once, which is precisely the preemption fight now heading toward the Supreme Court. If that claim holds, a newcomer reaches a national market the incumbents spent a decade assembling state by state. The arithmetic for a market maker is direct. The hold a sportsbook keeps as the house is, on an exchange, the spread a market maker competes to capture, and $150 billion of handle is migrating toward the venue where that spread gets priced in an order book.
Strip the framing and it is an order book
Underneath the casino language sits a clean instrument. A prediction market is a continuous double auction in binary claims priced between $0.01 and $0.99, where price is implied probability and settlement is a dollar or zero. That payoff is simpler than most listed options. It comes with maker and taker flow, fees and rebates, latency, depth, and resolution rules that occasionally disagree across venues. Polymarket charges taker fees in the range of 0.06 to 1.56 percent against maker rebates. Kalshi quotes no explicit fee and carries the cost in the spread instead.
The venues do not share a population. Polymarket skews crypto-native, global and continuous. Kalshi skews US-regulated, institutional and biased toward US hours. Two capital pools that rarely hold funded accounts on both sides, which is precisely why the same outcome trades at two prices. Academic work on the 2024 cycle found Polymarket leading Kalshi on price discovery because of deeper liquidity, with the gap widest in the final hours before resolution. Identical events still diverge by more than five points something like 15 to 20 percent of the time, and the divergence is fattest on politically charged questions where the participant mix differs most.
Where the edge actually lives
The textbook trade is mechanical. When Yes on one venue and No on the same outcome on another sum to less than a dollar net of costs, the spread locks regardless of how the event resolves. During the 2024 election Polymarket and Kalshi diverged by three to eight cents for hours at a stretch. The trouble is that everyone can see it. Public arbitrage bots already watch ten thousand markets at once, the windows last seconds, and deposit friction plus transaction costs close most of what is left. This is the screenshot arbitrage, and like every screenshot arbitrage in crypto before it, it is compressing toward zero on a short half-life.
What survives compression is structural, and it is the same edge that has always won in fragmented digital-asset markets. Making continuous two-sided prices across venues that were never built to talk to each other. Financing the inventory that sits on both legs while a position is carried. Modeling resolution risk and settlement timing so that a visible gap does not turn into basis risk. Running a single risk view across platforms with different collateral, different settlement and different counterparties. None of that is a signal you can copy from a dashboard. It is capital, latency and a risk engine, which is to say it is infrastructure.
The screenshot arbitrage is a bot problem with a half-life. The durable edge is the one that always wins in crypto: capital, latency, and the risk architecture that lets you hold both legs at once.
previously on bequant insights tradfi is coming on-chain. here's what that pays. The convergence thesis: tokenized equities, the SEC innovation exemption, and the trades that open when securities and crypto settle on the same rail. Prediction markets are the next venue class to fold into that same fragmentation problem.