Polymarket

Prediction markets stopped being a curiosity somewhere around the middle of 2025. Combined monthly volume across Polymarket and Kalshi went from under $5B in September 2025 to roughly $24B in April 2026 to $44.8B in June 2026, a ninefold move in nine months, through a period with no US presidential election in it. The 2024 cycle was supposed to be the peak. It was the on-ramp.

The regulatory question that governed the category for a decade is largely settled: the CFTC treats these as event contracts, not gambling, and withdrew its proposed restrictive rule amendment in February 2026. What replaced regulatory risk is competitive risk. Since the start of 2025, twenty new Designated Contract Market filings have appeared in the CFTC registry. The licence is no longer the moat.

The land grab is now for the stack: exchange, clearing, and distribution. Polymarket bought its way in, acquiring CFTC-licensed QCEX for $112M in July 2025 and relaunching to US traders that December. DraftKings bought Railbird and moved its flow onto its own exchange in June 2026. The players who own all three layers are the ones setting terms.

Monetization arrived abruptly. Polymarket charged nothing until January 2026, then rolled fees across crypto markets, sports in February, and a broader schedule in March, reaching roughly $1B annualized revenue by June. That is a real business built in six months, and it is also six months of pricing history to underwrite, which is not much.

Kalshi. Direct app plus broker integrations on a native CFTC-regulated DCM since inception. $6B 30-day volume, 58% of flow, roughly 80% sports. Winning on volume by leaning into sports. The direct comp, and the one that would set the price in any consolidation.

Robinhood. ~25M funded accounts already in the app on fully owned regulated infrastructure; 16B+ event contracts traded in 2026. The structural threat. Distribution is the scarce asset and Robinhood already has it.

DraftKings. Sportsbook base across 38 states, own exchange (DKeX) after acquiring Railbird; launched December 2025 and migrated off CME in June 2026. Proves the vertical integration thesis, and that an incumbent can buy the stack in a quarter.

CME / ICE. Institutional, not retail: the incumbent rails everyone else rents. ICE holds ~22% of Polymarket. ICE is the tell. The largest exchange operator in the world chose to buy in rather than build.

The bet is not that people will trade on events. $44.8B of monthly volume settles that. The bet is that one venue becomes the quoted number, the price newsrooms cite, terminals embed, and risk desks reference, and that the data business stapled to that position is worth more than the exchange underneath it.

ICE’s behaviour is the strongest evidence available. The largest exchange operator in the world put roughly $2B into Polymarket across two rounds for about 23% of outstanding shares, and simultaneously became the global distributor of its event data. ICE does not need to buy an exchange; it owns several. It bought a reference price and a distribution right. That is a read on where the margin sits, from the party best positioned to know.

Against Kalshi, Polymarket is losing the volume race and winning the citation race. Kalshi holds 58% of flow, but roughly 80% of it is sports: high-velocity, low-margin, and commoditized the moment DraftKings and Robinhood point their funnels at it. Polymarket’s mix skews toward the political, geopolitical and macro contracts that get quoted, which are lower volume and higher signal value. Those are different businesses wearing the same interface.

At a 20% discount the entry implies roughly $16.8B against a $21B last primary, about 17x the ~$1B annualized revenue run rate reached in June 2026. That is not cheap on its face. It is defensible only if the data and distribution layer compounds off the exchange rather than scaling with it, which is precisely the question the diligence list below is built to answer.

Distribution beats brand, and Robinhood has distribution. 16B+ event contracts traded in 2026 through an app that already holds funded accounts and a payments relationship. Polymarket has to acquire every user it gets.

The licence stopped being a moat. Twenty new DCM filings since the start of 2025, and both QCX and Aristotle carried DCO approvals too, meaning a buyer can acquire exchange and clearing together, as DraftKings did with Railbird.

Six months of pricing history. Fees only began rolling out in January 2026, and the US taker fee is 1bp against $1.00–$1.75 per 100 shares internationally. The blended take rate is not yet a stable input, and the mix is shifting toward the most fee-sensitive traders.

The volume base is reflexive. Nine months of vertical growth flatters every trailing metric, and 2026 had a World Cup in it. The comparison this sets up for 2027 is unforgiving.

State-level taxation is arriving piecemeal. Illinois’s operator tax took effect 1 July 2026 at 1.75% rising to 3.5% above five million contracts. Fifty separate versions of that is a margin story, not a compliance story.

A $21B mark set weeks ago in the hottest corner of the market is not a conservative anchor. A 20% discount to a euphoric print may still be above where the next round clears.

Secondary positions carry no information rights. Reporting arrives on the company’s schedule, not ours.