Prediction Markets Beat Sportsbooks. Design Explains Why.

Where the volume went

Kalshi's World Cup markets accounted for $7.4 billion in June. Polymarket's international exchange set a monthly record at $10.8 billion, and its US platform did $3.5 billion, nearly double May. Rothera, the Robinhood and Susquehanna joint venture, cleared $2 billion in its debut month and took about 7% of the US prediction market.

July looked the same. Sportico reported US prediction markets passing $20 billion in notional volume in about twelve days, matching all of May. England against Mexico drew $223 million on its own, the largest single sports result ever traded on a prediction market. Kalshi held 83% of US notional volume. DKeX, the exchange DraftKings built for exactly this moment, held roughly 0.01%.

The app data is the part operators should read twice. Between June 15 and June 30, Kalshi's daily active users rose 36%. DraftKings fell 36%, FanDuel fell 41%, BetMGM and Caesars each fell 32%. Kalshi and Polymarket together took 78.5% of installs across six major betting apps. A year earlier the two of them were at about 6%.

The audience nobody was reaching

Kalshi's female user count grew 106% during the tournament, against 54% growth for male users. By late June, women were 33.3% of its base. DraftKings and FanDuel sit around 22 to 23% and have for years. Those users were never blocked from downloading a sportsbook. They had the option the whole time and passed on it.

I think the interface explains more of that than anyone in the industry wants to say out loud.

Open a sportsbook and the first screen is an odds board. Minus 110, plus 240, a dense grid that assumes you already know what those numbers mean. American odds work as a literacy test at the door: you either learned them from someone or you close the app. Open Kalshi and you get a question, a yes, a no, and a price in cents. Sixty-three cents means a 63% chance. Nobody has to explain it.

The second difference is what you hold after you act. A bet is a ticket you keep until settlement. A contract is a position with a live price, and you can sell it in the 70th minute because the game turned. Sportsbooks have had cash out for a decade. It usually sits two taps into a bet history screen and reads like an escape hatch you are mildly discouraged from using. On an exchange, selling is the same size button as buying, on the same screen, all match long.

What sportsbook teams can copy

Not the legal structure. That fight is being decided in court and it sits above design pay grade. But most of what pulled those users in is interface work, and interface work ports.

Start with probability. A percentage toggle next to decimal and fractional odds costs a sprint and lets a new account read the board on day one. Several European books have offered odds format switching for years. Percentage is the obvious next option, and nobody has made it the default for first-time users.

Then move the position to the front. Put cash out on the bet card as a primary control with a live price, not a link inside history. Draw the line moving. A running price chart on an in-play selection is cheap to render and turns 40 minutes of dead waiting into a reason to keep the app open. That one change accounts for most of why an exchange feels alive next to a static slip.

The bet slip itself needs rewriting for people who have never used one. Accumulators, system bets, each way: all internally consistent, all opaque from outside. When we rebuild these flows for operators, the version that works states the position in one plain sentence before it shows any math. "You win $240 if all four of these happen" beats a table of stakes and returns for anyone in their first week.

What the exchanges are still bad at

Worth saying, because the copying should run both ways. Prediction market apps are poor at browsing. Kalshi's sports section is a flat list of questions with weak event hierarchy, and pulling up every market attached to one match takes more work than it should. Polymarket's US app still leaks trading vocabulary into places where a casual user needs none of it. Neither has anything close to the pre-match navigation a mature sportsbook has spent ten years refining. The exchanges won on entry and on the in-play moment. They have not won on depth.

The unstable part

None of this is settled law. On June 29, Ingham County Circuit Court Judge Rosemarie E. Aquilina issued a 14-day restraining order barring Kalshi from offering sports event contracts in Michigan, with a $120,000 daily fine for failing to geofence the state. On July 15 the CFTC rejected Kalshi's emergency rule filing and ordered the exchange to fulfill open trades with Michigan residents, saying it would not allow states or state courts to bully registered entities into violating the Commodity Exchange Act. The Michigan court wants certain contracts unwound by August 12. Kalshi's enforcement lead called the position impossible, which sounds about right.

There is also money underneath the numbers. Kalshi has spent an estimated $32 million on national TV this year and ran 36 spots during live World Cup matches on Fox and Telemundo. Polymarket ran 45. The rest of the US betting industry ran 22 combined. A World Cup also brings in people who watch four matches and vanish. Some of that install share is gone by September.

So no, do not rebuild a licensed sportsbook around an exchange model that state courts may litigate out of existence. But the conversion gap showed up inside a single month, and it showed up strongest among users nobody had reached with a decade of odds boards. The NFL season starts in about six weeks. That is enough time to ship an odds format toggle and a real cash out button, and not much else.