FanDuel Made the 3% Disappear Eight Days After Charging It. The Refund Is the One Product an Exchange Cannot Copy.

Editor's note: TrueEdge builds odds tools and earns affiliate commissions from licensed sportsbooks. This piece argues that a sportsbook product feature is structurally unavailable to prediction markets, and our commercial interest runs toward the sportsbooks. Weigh it accordingly.

On September 1, FanDuel announced Bet Protect+ as a paid add-on: pay 3% of your stake and if the player in your prop leaves the game injured and does not come back, you get the money returned in cash, or the leg pulled out of your parlay and the price recalculated on what is left. Eight days later, on the morning of the season opener, the 3% went away for the whole of September — pre-game and live, player props, same-game parlays and SGP+, overtime included, in the United States and Canada. Customers who already paid the fee are getting that back too.

Take the promotional framing at face value and it is unremarkable. Every book buys the first weekend. FanDuel is one of three official NFL online sportsbook partners alongside DraftKings and Fanatics, and September 9 is the day the whole industry lights money on fire.

The reason to spend a thousand words on it is that the feature underneath the promotion is not copyable by the competitor everyone in this business is currently worried about. A refund requires someone to be holding money they are willing to hand back. A sportsbook is that someone. A prediction-market exchange, by construction, is not.

What a refund actually costs, and who pays it

At a sportsbook, the house takes the other side. When FanDuel voids a losing receiving-yards prop because the receiver walked off in the second quarter and never came back, the money it returns comes out of the roughly 10% it holds across the book. That is a marketing expense with a line item, absorbed by a balance sheet that priced for it. DraftKings has run its Early Exit programme long enough to pass one million refunds, free and automatic, across the four major leagues plus the WNBA, golf and selected soccer, and its own framing is explicitly about trust: "With features like Early Exit, DraftKings aims to continue to establish itself as an industry leader as an operator that consumers trust." ESPN BET has the same product now. Fanatics has kept a version of it through periods when FanDuel dropped its own.

On an exchange there is no house holding the float. Every dollar that would fund your refund is currently sitting in the account of the person who took the other side of your contract — a person who was right, or at least who was on the correct side of the settlement rules. To refund the loser, the venue has to either claw the money back from the winner or pay it out of its own revenue. And its own revenue is thin: H2 Gambling Capital's April analysis put prediction-market revenue at about 1.1% of volume, against roughly 10% hold at a sportsbook. A book has ten points of margin to be generous with. An exchange has about one.

We watched the first version of this problem four days ago, and it was ugly. On September 5, Kalshi settled the Western Michigan–Michigan contract for the wrong side on $18.6 million of volume, then reversed the settlement, clawed the payouts back from the people it had wrongly paid, made the correct winners whole and issued a $50 credit to smooth it over. Coinbase hit the identical error and chose the opposite remedy, letting the wrongly-paid keep their money and paying the correct winners out of pocket. Polymarket, whose regulated app went down during the same slate, refunded losses and let wins stand. Three venues, one football game, three different answers about whose money it was.

That was an error, not a product. But it is the same mechanical problem in a more dramatic costume: discretionary generosity on a two-sided venue requires taking from a counterparty, and there is no elegant way to do it. A sportsbook can decide on a Tuesday to void a category of bets and nobody's account gets smaller. An exchange cannot make that decision without someone specific paying for it.

The obvious objection, which is correct

"Free" is a marketing word, not an economic one. If a book is refunding injury-affected props, it will price the props to cover it — Deadspin's version of this is that an over which would be -110 in an unprotected market can drift to -125 while the under gets the friendlier number. That is the whole trick. FanDuel has not eliminated the 3%; it has moved it from a visible fee that customers decline into a spread that customers cannot see. From a consumer-transparency standpoint that is a step backwards, and anyone reading this as a gift should stop.

We are not arguing Bet Protect+ is good value. We have no way to compute whether it is, because that would require knowing FanDuel's injury-void rate per market and the vig differential on protected props, and neither number is public. The honest position is that we do not know, and the one-million-refund milestone at DraftKings tells you about volume, not about whether customers came out ahead.

The claim here is narrower and, we think, harder to argue with: the ability to convert margin into discretionary customer protection is a moat, and the exchanges do not have the margin to dig one. The books spent two years competing on parlay breadth and same-game pricing, which the exchanges could and did answer. They are now competing on something that requires a balance sheet.

The counterargument worth taking seriously

An exchange operator would say this is a category error. Kalshi and Polymarket are not trying to sell a friendlier version of a sportsbook; they are trying to sell a market, and the reason someone routes a football position through an exchange is precisely that the venue does not have discretion over the outcome. A house that can void your bet can also void it when voiding is convenient. The trader who held an $888,234 Ben Shelton position through a 21% dip at four in the morning on Wednesday did not want a refund. He wanted a book that could not touch him.

That is a real argument and it is the exchanges' best one. It also has a boundary, and the boundary is the product most Americans actually buy. Player props and same-game parlays are where sportsbook handle lives, and they are the market type most exposed to a hamstring. Someone constructing a six-leg SGP is not looking for counterparty neutrality. They are looking for a lottery ticket and a reason to keep watching, and "we will give it back if your guy limps off" is worth more to them than any argument about federal preemption.

So the moat is narrow but it is in the right place. If the next twelve months of this fight are decided on handle — and H2's forecast of a 0.8% NFL decline says handle is exactly where it will be fought — then the books' most durable advantage is not their licences or their lobbying. It is that they can afford to say sorry.