Prediction Markets Had Their First $11 Billion Weekend. On DraftKings' Exchange, the People Taking the Parlays Lost 25.8 Cents of Every Dollar

Editor's note: TrueEdge builds odds tools and earns affiliate commissions from licensed sportsbooks, including DraftKings. This piece argues a position on prediction-market market structure and on the regulation that governs it. Our commercial interest runs toward the sportsbooks. Weigh it accordingly.

The record arrived on schedule. Aldrin Research, whose count Covers published at 12:31 p.m. on Monday, has Saturday, October 17 as the biggest single day it has ever recorded for prediction markets: $6.03 billion of notional volume across eleven venues, with Kalshi alone clearing $4 billion for the first time. Sunday added $5.74 billion from the nine venues that had reported by Monday morning. Covers called it the first $11 billion weekend. Add the two daily figures and it is nearer $11.8 billion; Covers' own piece notes that Robinhood, Crypto.com and ProphetX had filed only partial data, so the final number will move.

The same morning, Daniel O'Boyle at InGame published a different number from the same weekend, and it is the one worth keeping. On DKeX, the exchange DraftKings launched on August 26, parlay takers have now staked $239.4 million and lost $60.8 million of it after fees. That is 25.8 cents on the dollar. Kalshi's parlay takers, over the same autumn, have lost 14.4 cents. DraftKings' own sportsbook customers in Illinois, betting parlays against the house in January, lost 19.5.

The volume record and the loss rate are the same story told from opposite ends of the trade. The first is the number the exchanges publish. The second is what the people on the retail side of it are paying.

What the record is made of

Start with what $11 billion counts. Aldrin reports notional volume, which on an exchange means the face value of every contract matched. A parlay that pays $100 and trades at 4 cents is a $100 contract, not a $4 one. We walked through this on October 5, when Kalshi's previous records were set, and the ratio has not improved: Kalshi's weekend notional was $8.51 billion; its taker volume, the money customers actually put down, was $1.89 billion. The previous weekend those figures were $7.87 billion and $1.76 billion. The headline grew 8 percent. The cash grew 7.

Then look at what was traded. On Sunday, by Covers' breakdown, single-game sports contracts across the reporting venues did $1.44 billion. Combo contracts, which is the industry's name for parlays, did $3.7 billion. Kalshi's NFL volume on Sunday was $497.2 million, down from $575 million the Sunday before, which Covers attributes to four teams being on byes. Dallas at Green Bay, a 23-13 Packers win, was the single biggest market at $134.5 million. Bills-Raiders was $59 million. The Texans' walk-off field goal against Jacksonville at Wembley drew $57.8 million.

So the record is a parlay record, set on a weekend when the straight-bet football volume fell. That has been true of every record since September, and it is the reason InGame's analysis matters more than Aldrin's: the growth is concentrated in the one product where the exchange model behaves least like an exchange.

Who is on the other side

A parlay on DKeX, Kalshi or Polymarket US is not matched between two customers who disagree about the Cowboys. The app user can only take the contract. The other side is quoted by market makers through a request-for-quote process that, as O'Boyle notes, exists only in the exchange's API, which is why the makers are professional or institutional. The taker's loss is the maker's gain, less the venue's fee.

On DKeX, InGame does not know who the makers are, but it quotes Jason Robins telling investors in August that DraftKings was market making on three exchanges, and a line from his shareholder letter: "As DKeX grows, it will create more opportunities for our market maker." O'Boyle's figures show the growth happening. DKeX has done $2.7 billion of parlay volume since launch, topped $200 million on each day of this weekend, and appears to have got there because DraftKings Predictions began routing its parlay orders to its own exchange instead of Crypto.com's as the NFL season began.

Put the pieces together. DraftKings sells a parlay to a sportsbook customer in Illinois and keeps 19.5 percent of the handle, per the state's January figures. DraftKings sells a parlay to a prediction-market customer through DKeX, and that customer loses 25.8 percent of stakes, 19.5 percent of it before the exchange's fee, to a market maker that is, in Robins' own phrase, "our market maker." The company has spent a year telling shareholders it would rather the product did not exist while building one. On September 22 Robins said the stock would rise if the Supreme Court killed prediction markets. On the parlay book, at least, the product he would rather not have is the one holding more.

Why is it fatter? O'Boyle offers a mechanical reason. DKeX prices have a floor of 0.1 cents, which caps any contract at 1,000-to-1, and customers are building parlays of up to 20 legs whose true probability is far below one in a thousand. Every such ticket is sold at a price the maker knows is too high, and the taker pays it. A sportsbook would simply refuse the 20-leg parlay or cap the payout; an exchange with a price floor sells it at the floor. The hold on a parlay compounds with every leg, and a product that lets a customer add a twentieth leg at a fixed minimum price has found the one design that compounds it faster than a book.

The case that eight weeks is not a verdict

The strongest objection is that this is early, lumpy data from three young products, and the Polymarket US series shows how lumpy. Its parlay takers were net winners for a month after the August 11 launch; on August 21 they made $9.4 million on $4.4 million of stakes, most of it in a two-hour window that O'Boyle reads as mispriced maker quotes. Takers were lifetime profitable until September 12. They have since lost $69.6 million, 15.4 percent of stakes, and 17.7 percent if that one August morning is excluded. Rates that swing by a third on a single day are not rates a regulator should write into a rule.

There is a second objection, and it is the exchanges' best one: nobody is forced to the taker side. Anyone with an API key can quote. Competitive making is supposed to compress the markup over time, as new makers chase the margin that InGame has now published for them. A 25.8 percent loss rate in a product's eighth week may be the price of makers protecting themselves against uncertainty they have not yet priced, and it should fall as the book matures.

Both of those are fair, and the second one is testable. When Kalshi started charging its makers a fee on August 20, the markup on the affected parlays quadrupled within days rather than compressing, which is the opposite of what competition between makers would predict. If the DKeX taker loss rate is below the Illinois DraftKings parlay hold by the end of December, the argument in this piece is wrong and we will say so. If it is still above 20 percent after a full season of volume and every maker who wants in has had four months to connect, the "no house" description of the product should be retired.

What changed this morning

The CFTC published two documents in the Federal Register on Monday. The first, an interim final rule effective on publication, writes into the definition of a swap an exclusion for wagers placed through state- and tribal-licensed sportsbooks and casinos: a product is not a swap if a licensed operator offers it under state gambling law and it is not traded on an exchange or any other multilateral market. The second, a proposal, would define event contracts "including those based on sports-, politics-, cultural-, and weather- related events" as swaps. Comments on both close November 18.

Read them against InGame's table. The same company now offers the same bet to the same customer under two legal regimes the Commission has just formalised. The sportsbook parlay is gambling, regulated by Illinois, holding 19.5 percent. The exchange parlay is a swap, regulated by Washington, holding 25.8. The Commission's documents say nothing about hold, because hold is a consumer-protection question and the swap definition is a jurisdictional one. That is the point. The jurisdictional argument has now been formalised by the agency that is winning it, and the product that argument protects is, on the parlay book, the more expensive one.

We do not know whether DraftKings is the dominant maker on DKeX or one of several, because the exchange does not publish it. We do not know whether Kalshi's rate would look like DKeX's if its own makers had a 0.1-cent floor to sell into. What we know is that the weekend's record was built on parlays, that the parlays on the newest exchange cost their buyers a quarter of their stake, and that the company collecting both sides of that spent today being formally exempted from federal jurisdiction on one side of its business while invoking it on the other.