Two of Eight State Reports Can Tell a Parlay From a Game

Every month the Maryland Lottery and Gaming Control Agency posts a spreadsheet with two sheets in it. The first sheet is the one the trade press quotes: handle, prizes paid, hold, the state's cut. The second sheet is the one worth reading. It is headed Sports Wagering Bet Type, and alongside rows for football, basketball and tennis it carries a row called Parlay / Combinations. In the agency's October 2026 file, that single row holds 42.50 percent of the money wagered in the state and 71.50 percent of the money the sportsbooks kept. The federal record for event contracts has no row like it, and the rule the Commodity Futures Trading Commission proposed in July 2026 to govern event contract reporting would not create one.

That gap is the subject here. The standard answer to questions about combination products has been that nobody can measure them. Maryland has been measuring them monthly, in public, in a spreadsheet anyone can download. The difference between cannot be measured and is not asked for turns out to be a line in a form.

What the Maryland file says about where the money actually ends up

The numbers below are our own arithmetic over the agency's own cells, taken from the October 2026 workbook linked above on 11 November 2026. Statewide handle for the month was 710,097,348 dollars. The Parlay / Combinations row accounts for 301,801,097 dollars of it. Statewide hold was 72,435,079 dollars, and 51,793,653 dollars of that sat on the same row.

Put the two halves against each other and the shape of the business appears. Combinations held 17.16 percent. Everything else in the state, added together, held 5.06 percent. Four months into Maryland's fiscal year 2027, which began in July 2026, the same sheet shows combinations at 38.20 percent of handle and 65.21 percent of hold, holding 17.01 percent against 5.61 percent for the rest.

None of this is an investigation. It is a licensee filing a monthly form, and a regulator printing the form. Operators can produce the number because they are asked for it, which makes the claim that compound products are too entangled to count separately a claim about forms rather than about systems.

Eight state reports, read one document at a time

Before building any argument on Maryland, we read other states' own documents rather than trackers that summarise them. Seven of the eight are below and all seven were opened on 11 November 2026; the eighth, Colorado, follows the list. Each finding is limited to the document named and to nothing else.

Colorado belongs on the list as a second yes. The Colorado Division of Gaming has carried a Parlays / Combinations line in its monthly sports betting proceeds reports for years, and when we read its June and July 2026 reports in September 2026 that line came to 36.5 percent of handle and 53.4 percent of gross revenue. We could not reopen those files on 11 November 2026, because the Division's site refused requests from our data centre, so every figure dated today in this piece comes from Maryland.

So the count is two states with a usable bet type line out of eight whose own documents we have read, and the two are not the two largest markets. This is not a national census and should not be read as one. It is enough to retire the idea that the number is unobtainable.

Nevada kept a bet type line and the product moved out from under it

Nevada's monthly report is the only one of the eight that names a bet structure as well as a sport. In the statewide table of the July 2026 edition, Sports Parlay Cards shows 35 reporting locations and 180,000 dollars of win at a 15.38 percent win rate.

Two lines above it, Sports Mobile shows 37,302,000 dollars of win. The report's own footnote explains why that line has no detail: Mobile Sports data is being displayed in total for all sports and represents wagering activity from personal devices. Every sport split in that table, and the parlay card row with it, describes the counter in the building. The app is a single number.

Paper parlay cards came to 0.48 percent of the mobile win line that month. The taxonomy is not wrong, it is simply pointed at a product that stopped being where the money is. Nevada's report also quotes win and win percentage without reporting handle at all, which is a second reminder that a state report answers the question its form was written to answer and nothing else. A category that inherits a reporting template without asking what the template was built to see ends up in exactly this position.

Connecticut shows that machine readable and informative are different achievements

Connecticut publishes its online sports wagering figures as a queryable dataset rather than a PDF, updated on 15 October 2026 at the time we read it, with licensee, fiscal year, month ending, wagers, patron winnings, cancelled wagers, resettlements, win, federal excise tax, two promotional columns, gross gaming revenue and the payment due. The notes field even explains that resettlements are wagers whose original result was later corrected.

It is, mechanically, the best of the seven. It is also the emptiest on the question asked here. Nothing in those fifteen columns says what the customer bought. For anyone designing event contract reporting, Connecticut is the useful warning: the format we should want is already in service somewhere, and format alone settles nothing.

The hold column does not transfer to an exchange, and the question under it does

It would be easy and wrong to move Maryland's 17.16 percent across to a prediction market. In a sportsbook that figure is the house margin. On a designated contract market the losing side of a contract pays the winning side, and the venue earns fees, so there is no equivalent line and importing one would misdescribe the product. We said the same thing when we first wrote about combination contracts in September 2026 and it has not changed.

What does transfer is the concentration. A product that carries 42 percent of the volume and 71 percent of the losses is the product that decides what a venue's revenue is made of, whoever ends up holding the losses. Directive 06 of our manifest commits signatories to revenue that comes from many informed participants over years, not a few ruined ones over months. That commitment cannot be audited by anyone, including the venue, while the single largest product category is invisible in the published data.

Part 16 already carves one slice of volume out for separate display

The federal mechanism for this exists and is not exotic. 17 CFR 16.01 tells a reporting market what to record each business day and in which categories: futures by commodity and expiration date, options by underlying and put, call, expiration and strike, swaps by product type and term life. A contract on a compound condition falls into those categories without leaving a trace of its compoundness.

But the same regulation does something more interesting in paragraph (a)(2). After requiring total trading volume, it separately requires the volume of exchange for related position transactions and, in subparagraph (vi), the total volume of block trades included in the total volume of trading. Paragraph (e)(1) then requires all of it to be made available to the news media and the general public without charge no later than the next business day, and requires those three figures to be presented together.

So the Commission has already decided, for futures, that one structural slice of the day's volume deserves to be shown next to the total rather than folded into it. Adding a second slice is the same editorial act performed twice. That matters for how any proposal here should be argued: this is not a request for a new reporting regime, it is a request for one more subparagraph in a paragraph that already has the shape.

The event contract reporting proposal added four fields and not this one

On 1 July 2026 the Commission published Data Reporting Requirements for Certain Event Contracts, which we covered when it appeared and which gave the category its first reporting blueprint. We read the full text on govinfo on 11 November 2026: 36,224 words, in which parlay, multi-leg and combination as a product term appear zero times.

The proposal's own reasoning for using the futures template rather than the swaps template is worth quoting, because it is the hinge. The futures requirement of part 16, it says, allows reporting to the Commission, by product type and by expiry of the contract that captures the information inherent in Covered Event Contracts, with a footnote pointing at 16.01. The assertion is that the existing categories already capture what is inherent in these instruments.

And the proposal was perfectly willing to add fields when it wanted them. Proposed section 16.03(b)(2) applies 16.01 to a Covered Event Contract to the same extent as to a futures contract, except that in addition the exchange must record whether the event occurred, which event occurred, the time and date it occurred, and the source used to determine it. Four new data points, written for event contracts specifically, none of which is whether the contract resolved on one condition or on nine. As of 11 November 2026 this is still a proposal: the Commission's Federal Register output since 1 July 2026 contains no final rule under that title.

The exchange already treats a combination as its own object

The last place to look is the venue, and this is where the gap becomes hard to defend. Kalshi exposes combination products through a public endpoint for multivariate event collections. We paged through all of it on 11 November 2026 and counted 1,406 collections under 16 series tickers. Each one is a typed object with a minimum leg count, which was two in 1,405 of the 1,406, a maximum leg count, which was unstated in 1,368 of them and seven in 35, flags for whether the legs are ordered, the list of events the legs can be drawn from, and a plain language resolution rule: The resulting market will only resolve to YES if every associated market resolves to YES. Scalar outcomes are multiplied (rounded down).

Seventeen fields describe each collection. Not one of them is volume, and not one is open interest. The product taxonomy is machine readable and public. The traded quantity under that taxonomy is not published at all. Nobody needs to invent a definition of a combination contract in order to count one, because the exchange has already written the definition down in a form a script can read.

What a combination line would have to contain to be worth publishing

Directive 02 commits signatories to stating fees, spreads, settlement sources, custody arrangements and conflicts of interest in plain language on the surface where the trade happens. A product that carries most of the losses belongs in that same category of thing a participant should not have to infer. Five items would make a published line useful rather than decorative.

  1. Volume and open interest for contracts created from a collection, shown next to the single condition totals in the same daily file, the way 16.01 already shows block trades next to total volume.
  2. The distribution of leg counts, because a two leg contract and a nine leg contract are not one product. Kalshi's own objects start at two and most of them state no ceiling at all, so a single aggregate across that range says very little.
  3. The realised payout rate on resolved combination contracts over a trailing window, stated as a fraction of contracts that paid, not as a theoretical probability.
  4. The share of accounts whose combination volume exceeds their single condition volume. This is the Directive 06 number. A venue cannot claim broad informed participation while declining to say how concentrated its riskiest product is.
  5. The definition used, published alongside the figures, so that two venues' numbers can be compared and a change in the definition cannot quietly move the series.

The first two could be produced from data that already exists in the form the exchange already keeps it in. The third requires a settlement feed the proposed rule would, by its own text, start collecting anyway. The fourth is the only one that needs new work, and it is the one that would actually tell a venue something about itself.

A figure that survives a monthly tax filing can survive a daily volume file

Three documents settle the practical question between them. Maryland proves that licensees can report by bet type every month without the sky falling in. Nevada proves that a reporting taxonomy left alone keeps describing the product it was drafted around while the business moves to another line of the same table. Connecticut proves that modernising the pipe does nothing at all for what travels through it.

The category writing its own floor here is a short piece of work: pick the definition the exchange already publishes, count against it daily, and put the count next to the total. It is a worse outcome if the first person to publish a combination volume line for event contracts is a state regulator with subpoena power, because then the number arrives with an argument attached. It is a better outcome if it arrives as a column somebody chose to add.

Share X LinkedIn Email