For most of 2025, prediction markets were a story about Washington rather than about bettors. Kalshi’s court wins against the states, a Commodity Futures Trading Commission that is borrowing FIFA’s playbook (yes, farmers hedging against the weather is obviously the same thing as someone wagering on an NFL game), and volumes that were real but small.
We estimate that sports event contracts across the US-facing prediction market venues carried a handle equivalent of c.$7bn in 2025, against $184bn of regulated online sportsbook handle, equating to just under 4% of the combined market, and most of it in states with no legal alternative.
That is no longer the picture. The Super Bowl was the first sign that prediction markets were becoming a material product in the US sports wagering landscape, and the World Cup has been confirmation of this. Kalshi’s sports notional volume ran at $29bn in June, $38bn in July, and August has come in at $33bn.
DraftKings and FanDuel now sell sports contracts of their own, while Robinhood and PrizePicks route their customers straight into Kalshi’s order book. None of them have waited for the regulatory question to be settled, rather they have priced the uncertainty and moved.
Pump up the volume
The temptation is to read those notional figures as handle. They are not. Notional volume counts every contract at its $1 settlement value and counts both sides of every trade. When a customer buys a contract at 30 cents from a market maker, the customer has 30 cents at risk and the market maker holds the other 70 as liability, exactly as a sportsbook holds the other side of a bet.
When that customer sells the contract at 70 cents with their team ahead, the exchange records another dollar of volume, but no new money has been staked; it is effectively a cashout. Parlay contracts, now roughly half of Kalshi’s sports volume, compound the effect: a four-leg combination of even-money legs is priced at around six cents a contract, so $2 staked on it is recorded as $16 of notional volume, against $2 of notional volume for a single even-money bet.
H2 has built a dataset that makes the correction – turning the much-publicised notional volumes (everyone loves quoting big figures) into a handle equivalent (a figure that can actually be used for market analysis). We start by converting the notional volume (collected by venue and by contract type) into taker volume (a proxy for the amount consumers actually paid – rather than market maker volume).
We then convert this to a sportsbook handle equivalent by removing trades that closed a position rather than opened one. This is effectively the amount that customers have ‘at risk’, which is comparable with the handle figures state regulators publish. This will be tracked on a weekly and monthly basis, alongside our existing detailed US sports wagering state data.
This year has seen this relatively niche segment become mainstream. Across the first seven months of the year, we estimate the handle equivalent at c.$33bn (from $130bn of notional volume), compared to $102bn of regulated sportsbook handle. That is roughly a quarter of combined sports wagering, up from under 4% last year.
In July, the month of the World Cup final, it was c.40% of combined sports wagering, and while we are yet to see any regulated state handle data for August, prediction market volumes have held up strongly ahead of the new NFL season. H2’s analysis shows that notional volume currently converts to handle at between 20 to 25 cents on the dollar, and this ratio has been falling as the parlay share rises.
Where the money comes from is the question that matters most to the regulated industry, and it is the one no venue reports. Our working assumption is that a quarter to a third of prediction market handle originates in states with legal online sports betting, since the product is strongest where it has no competition. On that basis, prediction markets were around 1% of the legal-state market last year and around 8% so far this year, reaching 15% in July. These are estimates we will continue to refine, but the direction is not in doubt.
Line of scrimmage
The NFL season has long been the battleground for US sports wagering operators, and the momentum that prediction markets have gathered during the World Cup will make this one arguably the most important yet. H2 forecasts regulated sportsbook NFL handle will be flat this season year on year.
Underlying demand is still growing at c.4% a year, and it will be further boosted as Missouri adds its first full season, but prediction markets are taking an estimated 4% to 6% of sportsbook handle in the competitive states and considerably more in Florida, where the contracts compete against a single licensed operator. To put this into context, we estimate Florida’s NFL handle could be down 12% as a result.
However, a word of caution on the first data point that will be released. September’s state releases, published throughout October, will show NFL handle sharply lower, with a year-on-year decline of c.25% in the reporting states. No doubt that certain market commentators will attribute this to event contracts.
However, most of this is the calendar impact: the NFL season kick-off on 10 September leaves three NFL Sundays in the month against four last year; by contrast, Week 17 moves into January, which will lead to a substantial year on year uplift that month. That said, a September fall that is materially larger than the calendar explains would be the signal, and it’s important to read regulated state handle and prediction market handle equivalent as a pair, rather than individual data points.
The industry spent 2025 arguing about whether prediction markets should be allowed to exist – and that debate is far from settled. Whatever the answer, they exist, at a scale that now shows up in state handle. Measuring them properly is the first step in an evidence-based response for regulated operators and the states themselves.
Ed Birkin is managing director of H2 Gambling Capital, the leading market intelligence provider to the global gambling industry. He has 20 years’ experience across land-based and online betting, gaming and lottery markets, covering both market and company analysis, having previously led gaming research at Credit Suisse and Barclays Capital.
The post H2GC: Prediction markets riding the World Cup wave into the NFL season first appeared on EGR Intel.
Managing director Ed Birkin warns against equating notional volume to sportsbook handle, as accurately measuring the figures is a “first step in an evidence-based response” for regulated operators and states
The post H2GC: Prediction markets riding the World Cup wave into the NFL season first appeared on EGR Intel.