New York Just Took Its Fourth-Biggest Week of Bets Ever and Finished 59th in Revenue

New York's online sportsbooks took $619,392,725 in the week ending October 4. That is the fourth-largest handle in the state's history — behind only the weeks ending November 2, 2025, October 26, 2025 and November 30, 2025 — and it is the biggest week of the 2026 football season so far.

They kept $49,154,704 of it. Which, against the same 248 weeks of the Gaming Commission's own file, ranks 59th.

That is the entire story of the 2026 New York market in two sentences, and the gap between those two rankings is where the money in this industry is actually made or lost. Handle is a demand number. It goes up when there is football and down when there is not, and it is the number operators put in press releases. Hold — gross gaming revenue divided by handle — is a pricing number, and in New York right now it is going the wrong way for everyone except one company.

The number that should be the headline

We downloaded the Commission's statewide weekly file and the weekly file for every operator, converted them, and added up the five weeks of the 2026 NFL season against the matching five weeks of 2025.

Autumn 2025, the five weeks ending September 7 through October 5: $2,759,319,638 handled, $224,781,793 kept, a hold of 8.15 percent.

Autumn 2026, the five weeks ending September 6 through October 4: $2,881,733,458 handled, $198,062,235 kept, a hold of 6.87 percent.

More money in, less money out. Handle up 4.4 percent year over year; revenue down 11.9 percent. At New York's 51 percent tax on mobile GGR, that swing cost Albany roughly $13.6 million across five weeks on a business that grew.

The weeks inside those totals are worse than the totals. On the week ending September 13 — the first full Sunday of the NFL season — New York's online books handled $594,595,203 and kept $16,963,508. That is a hold of 2.85 percent — the eighth-lowest of the 248 weeks in the Commission's file, and the lowest since the week ending March 30, 2025. A sportsbook running a 2.85 percent margin on a half-billion dollars is not running a business; it is running a clearing house and paying for the privilege.

FanDuel is playing a different game, and has been for 28 weeks

Here is the week ending October 4, operator by operator, straight out of the Commission's files:

• FanDuel — $228,193,108 handled, $23,600,001 kept, 10.34 percent
• DraftKings — $213,980,181 handled, $13,756,092 kept, 6.43 percent
• BetMGM — $38,130,090 handled, $2,671,551 kept, 7.01 percent
• Fanatics — $70,025,735 handled, $4,894,277 kept, 6.99 percent
• Caesars — $44,565,729 handled, $2,177,048 kept, 4.89 percent

FanDuel and DraftKings took 71.4 percent of the state's handle that week. They took 76.0 percent of the revenue. Inside that, FanDuel handled 36.8 percent of the money and booked 48.0 percent of the margin; DraftKings handled 34.5 percent and booked 28.0 percent. On near-identical volume, over the same slate of NFL games, with the same injuries and the same closing lines available to the same public, one book kept $9.84 million more than the other.

The obvious objection is variance, and it is the right objection. One week of hold on parlays is noise. So we ran all 28 weeks of fiscal 2026/27, from the week ending March 29 to the week ending October 4, for both books.

FanDuel's hold exceeded DraftKings' in 23 of 28 weeks. Mean weekly gap: 1.58 points. Cumulatively: FanDuel took $4,871,222,115 and kept $486,904,275, a hold of exactly 10.00 percent. DraftKings took $4,684,733,906 and kept $402,467,271, a hold of 8.59 percent. FanDuel earned $84.4 million more revenue on $186.5 million more handle. Twenty-three weeks out of 28 is not variance. It is a price.

The finding we did not expect

The gap is not constant. It is seasonal, and it closes in football.

Through the spring and summer — the weeks ending April 5 through July 19, which is baseball, the NBA playoffs and the World Cup — FanDuel beat DraftKings on hold by three to four points over and over: +3.79, +3.64, +3.60, +3.53, +4.37, +3.29, +3.28. Then football starts, and the gap mostly vanishes: +0.32 on the week ending September 6, +1.53 on September 13, minus 1.39 on September 20, minus 0.16 on September 27. Across the whole five-week NFL window the cumulative gap is +0.87 points, a quarter of what it was in June.

October 4's +3.91 is the exception, and it is the reason the week looks dramatic. It is also, on this evidence, more likely to be a good week for FanDuel than a new normal.

If that pattern holds up — and five weeks is not enough to be sure, which is the honest version of this claim — it says something specific about where the margin in American sports betting lives. The NFL is the commodity. Everybody has the same number, the public bets the same sides, the market is deep enough that nobody can get away with a bad price, and the hold compresses toward the vig. The money is in the long tail: a WNBA alternate line, a Tuesday-night baseball same-game parlay, a tennis correlated-leg price that four people in the building understand. FanDuel's structural edge is not that it prices the Chiefs better. It is everything that is not the Chiefs.

What else could be true

Three alternative explanations deserve stating, because each of them would change the conclusion.

Promotional accounting. We do not know, from the Commission's files, how bonus and free-bet credit is treated inside the GGR line — the reports carry two notes, one on cash-basis recognition and one on the 51 percent tax rate, and nothing on promotions. If promotional credit reduces reported revenue, then a book spending harder on bonuses posts a lower hold without pricing a single market worse, and DraftKings' September is a marketing decision rather than a trading failure. We cannot separate those two from these files, and we are not going to pretend otherwise.

Cash-basis distortion. The Commission's own note says New York recognises futures wagers as revenue when they are written and payouts when they are redeemed. That is not a rounding problem. It produced a week — the one ending June 14, 2026 — in which the state's books reported negative $48,515,979 of gross gaming revenue on $587,580,123 of handle, because the World Cup settled and the tickets came in. Any single week's hold in this market can be bent by a futures book settling. Five weeks is better. A full season is better still.

Customer mix. A book that has been in New York longer, or that attracts recreational parlay players rather than price-shoppers, holds more without pricing differently. FanDuel's 10.00 percent fiscal-year hold is consistent with a parlay-heavy book. That is a real alternative to "FanDuel prices better," and we cannot rule it out. It is also not very comforting, because it means the hold gap is a measure of which book's customers are worse at this.

What none of those three explains is the direction of the whole market. Promo accounting, cash-basis timing and customer mix do not add up to a statewide hold that fell from 8.15 percent to 6.87 percent year over year on rising volume. Something in New York is getting cheaper for the person placing the bet. Whether that is competition, promotional spend or a public that has finally learned what a 13-leg parlay costs, it is the first good news bettors have had out of this state in a while — and it is the number operators will be asked about on their next earnings call.

The week ending October 11 posts to the same file next Monday. The row is already there, blank.