Hartford keeps one of the cleanest datasets in American gaming, and almost nobody in venture or public markets reads it. Connecticut’s Department of Consumer Protection makes four online licensees file the same form every month, and the file carries a column most states never publish: total wagers. Not revenue. Wagers.
In July 2026 those licensees took $1.87 billion in online betting wagers and kept $65.35 million of it. That is a margin of 3.50%. Run the same arithmetic on the previous month and you get 3.46%. On the month before that, 3.58%.
Where does money that steady come from? Not one game, and not one fortunate month. Part of the answer is how the regulated market is structured. Real-money online betting play is live in only seven states, each one licenses its operators through its own gaming authority, and a player has to be physically inside the state to wager. Each state’s monthly figure is therefore what a closed set of licensed brands keeps. Which brands make up that closed set in each state is laid out at Legal Sports Report under popular online casinos.
The usual version of this story then makes a second claim: that steady revenue keeps the share prices quiet too. It does not. The revenue is the smooth thing, and those are not the same statement.
A Month of Wagers Is a Very Large Sample
A single slot session is close to a coin flip with a weighted coin. A player can win. Over a hundred spins the arithmetic is still noisy enough to feel like luck. Across nearly $1.9 billion of wagers in one state in one month, it stops being luck at all. The realized margin converges on the designed margin, and it converges tightly enough to budget against.
Connecticut’s record shows exactly that. Across the 24 months to July 2026 the state’s online betting hold averaged 3.48% of wagers, with a standard deviation of 0.17 of a percentage point. The lowest month was 3.17%. The highest was 3.84%.
Widen it to the full 58 months since the October 2021 launch and the band only stretches to 2.47% and 3.84%, and most of that spread is drift in the game mix rather than variance. The market’s first reported month already put $313 million through the system. The sample was large enough from day one.
That is the mechanic the phrase “the house always wins” fumbles. The house does not always win. It wins a predictable fraction of a very large number of small events, which is a much duller and much more bankable statement.
Where the Revenue Line Comes From, and Why It Barely Moves
Multiply a near-constant margin by a handle that grows steadily and you get a revenue line with no interesting shape. Connecticut’s online betting win ran $339.4 million, $473.1 million, $621.4 million and $758.7 million across its last four completed fiscal years. Four straight years of growth, no down year, no shock.
Month to month the same series moves by an average of 5.7% in absolute terms, and its worst single month-on-month swing over two years was 13.9%. For a consumer business with no contracts and no subscriptions, that is remarkably close to a subscription profile. The reason is who generates it. This is a repeat-deposit business, not a transaction business. DraftKings reported roughly 3.6 million monthly unique payers in the second quarter of 2026, up about 9% year on year, at an average of $132 of revenue each.
Those two numbers are the whole engine. A recurring payer base times a stable per-payer spend times a margin set by the game math, and the output is a line a CFO can forecast a year out. Whether a favorite covers has nothing to do with it.
The Sportsbook Next Door Posted a Negative Month
Connecticut runs a control group for this argument, and it is sitting in the next spreadsheet. The two commercial operators that dominate both files are the same, the customers are the same, the state is the same. The math is not. Over the identical 24 months, online sports wagering in Connecticut held 10.39% of wagers on average. The standard deviation was 2.40 percentage points, fourteen times the betting figure. The range ran from 3.03% to 13.20%.
Then there is June 2026. The Mashantucket Pequot licensee that carries DraftKings in the state, filing as MPI Master Wagering License CT, LLC, reported negative online sports wagering win of $4.67 million on $76.2 million of wagers. A licensed sportsbook can post a losing month. A licensed online betting operation, at that scale, effectively cannot.
Measured on revenue rather than margin, the gap is starker still. Connecticut’s monthly betting win moves an average of 5.7% between months. Its monthly sports win moves an average of 35.6%, and one month-on-month jump over the past two years was 157%.
Flutter Entertainment put the same contrast in a single set of results. In its third quarter of 2025, as the trade press recorded at the time, FanDuel’s US iGaming revenue grew 44% year on year to $530 million while sportsbook revenue fell 5% to $783 million.
Flutter cut full-year guidance by $570 million of revenue and $380 million of adjusted EBITDA. One half of the business was doing arithmetic. The other half was watching games.
Steady Revenue Has Not Bought a Steady Share Price
So the premise deserves the numbers. If smooth revenue produced smooth equities, the listed operators would be quiet stocks. They are not. DraftKings carries a five-year beta of 1.63 on a widely cited market-data profile of the stock, and its price fell 48.99% over the 52 weeks to late September 2026. That is not a low-movement security by any definition a portfolio manager would accept.
Flutter closed at $92.91 on 5 August 2026, down 11.5% in a day, after trimming its 2026 adjusted EBITDA guidance by $210 million to a $2.655 billion midpoint. Its US revenue fell 6% to $1.683 billion in the quarter while US sportsbook revenue dropped 15% to $1.039 billion, which tells you which half moved.
Rush Street Interactive is the closest thing the US market has to a betting platform-first listed operator, and it is no calmer. Its second-quarter report showed revenue up 46.3% to $393.8 million, adjusted EBITDA of $64.62 million, and raised full-year guidance.
The stock traded at $31.91 after those results and around $23.30 two months later, against a 50-day average of $27.76. Its beta is 1.51.
Two clocks run over this sector, and that is part of why. Regulators publish monthly, on a fixed date, without commentary. Listed companies report quarterly, with a great deal of it. The publication whose operator pages are linked above is on X.
The One Low-Beta Name Sells the Mechanics
There is an exception, and it proves the mechanism rather than the headline. Evolution AB, the Stockholm-listed supplier that builds and runs live betting tables for other companies, has the flattest revenue line in the sector.
Its own interim report gives five consecutive quarters of net revenue in euros: 524.3 million, 507.1 million, 514.2 million, 513.0 million, and 517.8 million. The whole series fits inside a 3.4% band. The EBITDA margin was 65.9% in the second quarter of 2026 and 65.9% in the same quarter a year earlier.
Evolution is also the one name here with a beta below the market, at 0.73, and it is the one furthest from the player. It takes commission from more than 870 operator customers rather than carrying any individual outcome, so it is two layers removed from the variance that the game itself has already flattened.
Even that has limits worth naming. The same report records a GBP 4.75 million settlement concluding a UK Gambling Commission license review, and a quarter-on-quarter decline in Asia that management put down to cybercrime activity it is still working through. The stock is up 8.20% over 52 weeks, and it now trades near 866 kronor against a 200-day average of 662. A gap of more than 30% between price and trend is a rally, not a calm. A flat revenue line does not buy a flat chart. It buys a slightly flatter one.
What Actually Gets Priced
Founders should take the operating lesson and leave the market one. Predictable unit economics are a real asset: they shorten the planning cycle and they make working capital cheap to model. That is most of why a gaming license is worth what it costs.
They are not a valuation multiple. Nothing in a stable hold percentage tells a buyer how fast the market grows, how many states open next, what a new tax rate does to contribution margin, or whether a rival product siphons off the same wallet. Those are the variables that actually move a price, and none of them is smoothed by the law of large numbers.
The pattern is not confined to gaming. Software platforms that embed payments into their own checkouts monetize by taking a small portion of payment volume, which is the same shape: a fixed cut of a very large number of transactions, dependable at the unit level and silent about what the company is worth.
So the answer splits in two. Regulated online betting revenue really is unusually smooth, and the reason is mechanical rather than managerial. The shares of the companies booking that revenue are not smooth at all, and anyone who expected the first to deliver the second was reading the wrong line of the accounts.
