The Casino Lobby Just Told You Prediction Markets Will Out-Trade the Entire Legal NFL Market. It Is Doing Kalshi's Advertising For It.

Editor's note: TrueEdge builds odds tools and earns affiliate commissions from licensed sportsbooks. The trade association discussed below represents our commercial partners. Weigh it accordingly.

The American Gaming Association's NFL forecast landed on September 4 and it contained one genuinely new fact: legal handle has stopped growing. The AGA projects $29.5 billion wagered with US sportsbooks this season against $29.4 billion last year. That is 0.3%, which after eight straight years of double-digit expansion is a flat line with a rounding error attached.

Bill Miller, the association's president, named the culprit in the first paragraph. "Since the widespread launch of backdoor sports betting on so-called 'prediction markets,' the growth of legal handle has stalled."

The same day, ESPN's David Purdum reported the number that has been travelling with it ever since: Eilers and Krejcik Gaming expects prediction markets to do $36.8 billion in NFL trading volume this season, roughly double last year's, with August 2026 NFL volume already running 4.6 times August 2025.

Set those side by side and you get the headline everyone wrote. The exchanges will trade more on the NFL than every licensed sportsbook in America will take in bets.

It is not true, it is not close to true, and the trade association whose members are being harmed by prediction markets is the entity working hardest to make you believe it.

Volume is not handle, and the people producing the volume say so

Handle is stake. When you put $100 on the Seahawks at a sportsbook, handle goes up by $100 and stops.

Notional volume on an exchange counts the face value of contracts changing hands, and the same dollar can be counted repeatedly. A market maker quoting both sides of a Week 1 total generates volume all afternoon without ever taking a directional position. A trader who buys at 40 and sells at 55 before kickoff has produced two transactions and risked the spread. Kalshi's parlay-equivalent combination contracts, which now dominate its sports book, settle a chain of legs into a single notional figure that bears no relationship to what a bettor put up.

You do not have to take our word for this. Eric Ramsey and Dustin Gouker, writing on August 26 in the most granular public analysis of Kalshi's data anyone is producing, note in passing that exchange volume "is not directly comparable to traditional sportsbook handle" because of how parlays are structured and traded — and then decline to quantify the gap, because nobody outside the exchange can. Their own numbers show why it matters: of Kalshi's incremental year-over-year volume, sports parlays accounted for roughly 80%, at $15.2 billion against $13.5 billion in straight sports markets.

Gouker had put it more bluntly a year ago, in September 2025, when Kalshi's football weekend was $303 million rather than $2 billion: strip out market-maker activity and the residual handle looks like a decent-sized state, not like the American sports betting industry. That analysis is twelve months old and we are flagging the date because it is exactly the kind of piece that gets recirculated as current. The reasoning has not aged. The numbers have.

The AGA's own numbers are the weakest part of the AGA's case

Two figures in the September 4 release are doing heavy rhetorical work and neither arrives with a method.

The first: prediction markets have cost states more than $1.3 billion in gaming tax revenue since 2025. That is a counterfactual — it assumes every dollar traded on an exchange would otherwise have been staked at a taxed sportsbook in a state that taxes it. Some of it would have been. Some of it is the same dollar counted four times, and some of it belongs to people in the eleven states where there is no legal sportsbook to lose the business.

The second is stranger: $5.1 billion in Kalshi sports volume from traders aged 18 to 20, as of September 1. Kalshi takes customers at 18; Michigan and most states require 21 to bet. The age gap is real, it is the single most defensible thing in the AGA's brief, and Judge Aquilina cited it in her injunction. But the AGA does not hold Kalshi's account-level age data, the release does not say where the figure came from, and we could not find a published methodology. We are reporting it as an AGA claim because that is what it is.

This is a lobbying document that would have been stronger with less in it.

Where the substitution actually shows up

The honest evidence is smaller, duller and considerably more alarming for the AGA's members than the headline number.

New York, week ending August 30: $420.3 million in handle, up 10.1% on the week — and down 24.7% year over year. Gross gaming revenue $39.3 million at a 9.5% hold. DraftKings took $149.5 million of it, FanDuel $146.8 million, Fanatics $52.8 million. One week is one week, and Labor Day weekend falls differently across years, so treat the year-over-year figure as a signal rather than a measurement. But a quarter is a quarter, and it is in the largest, most mature, most heavily policed market in the country, where the explanation cannot be that customers had nowhere legal to go.

Meanwhile the exchange side is not on a straight line either. Combined August volume at Kalshi and Polymarket was $45.33 billion, down 14.5% month over month — the first decline in more than a year. Kalshi fell 7.3%; Polymarket fell 36.7%, to $8.16 billion against Kalshi's $37.17 billion. Part of that is the World Cup receding out of the comparison. Part of it is that Polymarket is being beaten badly in US sports.

The strongest case against us

Every discount above can be applied and the direction still does not change. Football volume on Kalshi grew roughly thirtyfold across the first 33 preseason games. The 2026 Hall of Fame Game traded $21.2 million against under $500,000 in 2025. The least-traded preseason game this year did double the most-traded game last year. Daily volume went from about $25 million in August 2025 to over a billion on ordinary weekdays. Whatever fraction of notional volume is real money at risk, that fraction has been roughly constant while the absolute number went up fifty times. Cut $36.8 billion by three quarters and you still have a competitor that did not exist two seasons ago taking a bite the size of a top-five state.

That is correct, and it is why the AGA is right about the trend and wrong about how to prove it.

Here is what the association should have used and did not. Kalshi's best single day of fee revenue on record was $17.2 million, in July. Fees are the number that cannot be inflated by churn, because somebody paid them. That figure is the real measure of the threat, and note what happened when it entered circulation: it was immediately annualised to $6.28 billion and compared to DraftKings' 2025 revenue — which is the identical error in the opposite direction, extrapolating a World Cup record day across a calendar year.

Both sides of this fight want the same inflated number for opposite reasons. The exchanges want to look inevitable to investors. The AGA wants to look besieged to legislators. The trouble with borrowing your opponent's marketing is that in January somebody will check it, and the AGA will be the one holding a forecast it cannot walk back.