FanDuel Took $12 Billion in Bets and Kept $3 Million. Prediction Markets Haven't Taken Its Customers. They've Raised the Price of Finding New Ones.

Editor's note: TrueEdge builds odds tools that include prediction-market prices and earns affiliate commissions from licensed sportsbooks. Our standing position is that states should not criminalise federally registered exchanges. Weigh the conflict.

FanDuel took $12.02 billion in bets in the third quarter, up 13%, and turned $1.455 billion of US revenue into $3 million of adjusted EBITDA. A year earlier that line was $51 million. Flutter blamed spending on its prediction-market product and its Alberta launch, cut its full-year guidance, and paused its buyback "for the foreseeable future."

The easy headline says prediction markets are eating FanDuel. Flutter's own numbers do not say that yet. They say something less dramatic and possibly harder to fix: the exchanges have not pulled many bettors off the sportsbook, but they have made the entire category more expensive to advertise in, and FanDuel now pays that bill twice, once as a sportsbook competing for attention and again as the owner of a small exchange it says it must keep funding.

Where the cost actually landed

Dan Taylor's shareholder letter draws the line himself. Flutter continues "to see a limited impact from prediction markets on the betting activity of our existing customer base," he wrote, but is "beginning to see an impact on our overall media costs as prediction market operators increase marketing spend."

On the call he gave the size. Category media spend was up more than 60% year on year in September and more than 50% in October, and "the increase has predominantly been driven by the predict operators." That is not a FanDuel figure. It is what everybody buying a Sunday NFL slot is now competing against. Taylor said how long it lasts "will depend in part on how the regulatory landscape develops," which is a chief executive telling investors that his marketing line depends partly on a stack of court cases he does not control.

CFO Rob Coldrake put a date on the rest. Flutter expects promotional spending to hold at similar levels through 2027, and Taylor said about 6% of handle is the right aggregate number for promotions. Prediction-market investment continues next year "at similar or higher levels." The company did not give a dollar figure for that investment in anything we read, so we cannot tell you how much of the $48 million year-over-year EBITDA drop it explains. Investors should want that number, and they did not get it on Wednesday.

How small the product still is

FanDuel Predicts averaged 66,000 monthly players in the quarter, up 39% from the second quarter, and handled $119 million in volume, up 102%. Set that beside the sportsbook: $119 million is about 1% of $12.02 billion. Set it beside the competition and it nearly disappears. TickerTracker data cited by Dustin Gouker of The Event Horizon has Kalshi at $25.4 billion in the single week ending November 2, Polymarket's US exchange at $4 billion and DraftKings' DKeX at $1.27 billion. A full summer quarter against one NFL-and-election week is an unfair comparison, and exchange notional is not sportsbook handle. The gap is still too wide for that to explain away.

Taylor's case is about direction. He said volumes were "almost trebling in October when compared with July," cited about 140,000 monthly actives in October, and said retention and frequency are running at 80% to 85% of sportsbook levels. "We are acquiring at materially lower CPAs than our sportsbook," he told an analyst, while conceding the economics are uncertain. RG's account of the letter gives October volume as 271% higher than July, which is a larger multiple than "almost trebling"; we could not read the release itself, which returned a 403, so we do not know which description is the precise one.

There is also a hedge inside the hedge. Flutter makes markets on these venues, with 2026 guidance of about $50 million in revenue and about $40 million in adjusted EBITDA, which Coldrake declined to raise. So the same growth that inflates FanDuel's media bill also pays part of it back through a trading desk. Not most of it: against a US EBITDA guide of $715 million, $40 million is a modest offset.

The case that $3 million means nothing

The strongest objection is that this quarter is a poor place to measure anything, and it is a serious objection. The third quarter carries only a sliver of the NFL season. Taylor said flatly that "sports results went against us in the quarter" and that early fourth-quarter results have also been unfavourable. The World Cup was about 6% of handle and brought in 2.3 million customers, about a third of them reactivated, whose value shows up later if it shows up at all. Alberta is a launch cost. Free cash flow rose 22%. And a 94% decline off a $51 million base is a big percentage on a small seasonal number. Anybody extrapolating $3 million into a year is making a mistake.

We accept all of that, and it still does not rescue the bull case, because the better evidence is the full-year guide. Flutter now expects US revenue of about $7.34 billion, up 5%, and US adjusted EBITDA of about $715 million, down 22%. Bad luck in a single quarter does not produce a forward statement that marketing intensity continues into 2027 at similar levels, promotions stay at similar levels, and prediction-market investment stays at similar or higher levels. Those are decisions, and management announced them while cutting its buyback with leverage at 4.6 times against a 2.0-to-2.5 target.

The other gap to watch is handle up 13% against sportsbook revenue up 4%. Sports results explain some of that and Flutter did not quantify how much. Coldrake expects US net revenue margin to stay broadly flat next year even with parlay mix growing, because promotional cost rises alongside it. Volume is arriving; the company is paying to keep it.

What would prove this wrong

This thesis has a near-term test. DraftKings reports on November 5, with its call on November 6, according to its own October 21 release. If DraftKings describes a similar media inflation with handle holding up, the story is the one we are telling: an industry-wide tax on customer acquisition, set by exchange marketing budgets, which ends only if those budgets shrink or courts shrink them. If DraftKings reports flat media costs, then Flutter's problem is more Flutter's own, and the prediction-market explanation becomes a convenient place to put it.

The second test is Taylor's own claim. "Limited impact" on existing customers is a statement about today, and Coldrake described sportsbook market growth at mid-single digits through 2027. If handle growth at FanDuel falls toward that number while exchange volume keeps climbing, "limited" will need a new adjective. Asked directly about handle, Taylor said: "I'm not going to sort of try and predict handle trends." That is a reasonable answer from an executive, and it also means nobody at the company has promised anything.

Flutter also took an $873 million non-cash Brazil impairment, posted a $1.148 billion group net loss and is cutting about 1,400 UK and Ireland roles, so the US line is one of several pressures on the group. Premarket quotes on the stock conflicted in the coverage we read, showing moves in both directions, so we are not reporting a move.

One clean sentence covers the quarter for bettors and investors alike. The biggest sportsbook in America grew its handle by 13% and paid for it with almost all of the quarter's profit, and it told shareholders the spending continues.

Gambling involves risk. If you or someone you know has a gambling problem, call 1-800-GAMBLER.