On 16 March 2026 the Commodity Futures Trading Commission published an advance notice of proposed rulemaking on prediction markets at 91 FR 12516, under RIN 3038-AF65, and gave the public until 30 April to answer it. By the time the window closed, the docket held roughly three and a half thousand submissions. That figure has been repeated ever since as though it settled something.
It settles nothing on its own. A comment file is not a poll and the agency is not counting votes. What it does is decide which arguments sit in front of the people writing the rule, and on that measure the spring record is far smaller and far more specific than the headline number suggests. It is worth reading properly, because two more windows are open this autumn.
Forty numbered questions went out in March
The Commission announced the notice on 12 March 2026 in CFTC press release 9194-26, where Chairman Michael S. Selig called it "an important step in the Commission's continued effort to promote responsible innovation in our derivatives markets". The notice itself is short by rulemaking standards, nine pages, and almost all of it is questions.
Counting the numbered items in the published text returns forty questions, with twenty lettered sub-parts hanging off four of them. They are grouped into six areas: core principles and Commission regulations, the public interest, the five activities listed in the Special Rule at CEA section 5c(c)(5)(C), the procedure for making a determination under that section, inside information, and the classification of event contracts themselves.
The range is wider than the rulemaking that followed. Area A alone asks how a venue provides impartial access under Core Principle 2, what should happen when two parties disagree about whether the event in a contract occurred, how to decide that a contract is not readily susceptible to manipulation, how position limits should be aggregated across similar contracts, whether these markets should ever offer trading on margin and what disclosure retail customers would need if they did, and how any of it applies to a blockchain-based venue. Those are the questions an operator actually lives inside.
The notice also explains why it was issued when it was. In a footnote the Commission sets out the growth curve behind the whole exercise: from 2006 to 2020, designated contract markets listed an average of about five event contracts a year. In 2021 that number rose to 131, and in 2025 exchanges certified approximately 1,600 event contracts for listing. A regulator that watched a product line multiply by a factor of three hundred in five years was always going to ask what it was looking at.
The Commission graded its own record and put the number at three hundred
The arithmetic that matters is in the proposal that came out of the file. In Prediction Markets; Public Interest Determinations, published 12 June 2026 at 91 FR 35806 under the same RIN, the Commission writes that it received "approximately 3,500 comments" and then immediately narrows it: "approximately 300 submissions provided detailed comments and recommendations. The remaining submissions were either duplicative of points made in other submissions or non-substantive."
Two things are worth holding at once. This is the agency grading its own inbox, and no outside party audited the split. It is also almost certainly right in shape, because organised campaigns produce near-identical letters and every large federal docket carries them. The regulations.gov docket for the notice, CFTC-2026-0331, returned 3,535 public submissions when queried through the regulations.gov API on 30 August 2026. The Commission keeps its own copy on the CFTC public comments portal, named in footnote 155 of the proposal.
So the working number is not 3,500. It is somewhere near 300, and every one of those was written by somebody who decided the topic was worth a day of their own time.
Seventeen letters reached the footnotes, and four of them came from individuals
There is a harder measure available, and anyone can run it. Search the June proposal for the phrase "Letter from" and nineteen citations come back. Two of them point at older correspondence, a 2008 letter from Erik Sirri at the Securities and Exchange Commission and a 2020 letter from the Secretary of the Commission to ErisX. That leaves seventeen citations to comment letters filed on this notice.
The named commenters span most of the categories the Commission lists. CME Group and Kalshi are there, alongside Susquehanna International Group, the Coalition for Prediction Markets, Better Markets, the National Cattlemen's Beef Association, and the Commodity Markets Council filing on behalf of several agricultural trade organisations. Two tribal governments appear, the Pechenga Band of Indians and the Tohono O'odham Nation. A joint letter from the players associations of the NFL, MLB, NBA, NHL and MLS is cited for the proposition that contracts settling on a named player's injury should be prohibited, partly because they turn on personal medical information. A letter from eight United States Senators including Senator Jeffrey A. Merkley is cited on the enumerated activities, and former Senator Blanche Lincoln, writing through Lincoln Policy Group, is cited for the view that some contracts genuinely should be prohibited while others help users manage real exposure.
And four of the seventeen are individuals with no institution behind them at all. Harry Crane of Rutgers University is cited alongside CME Group for the same structural reading of the Special Rule. Amadeus Brandes, Daniel Cavero and Jeromee Johnson are each cited by name for a single specific argument. That is the part of this record worth sitting with. A one-person letter and a letter from the largest derivatives exchange group in the world are cited in the same sentence, on equal terms, because both of them answered a question that had been asked.
A staff meeting counts as a comment in this document
Footnote 156 of the June proposal contains a detail that changes how the file should be read. The Commission notes that information about meetings CFTC staff held with outside organisations regarding prediction markets is included in the same list as the comments, and states that the views expressed in the comments and at those meetings are "collectively referred to as the views of 'commenters'".
Read plainly, that means the written file is not the whole record. A firm that took a meeting sits in the same category as a firm that filed a letter, and both are visible in the same place. That is useful in both directions. It widens the channel, and it removes the excuse that the only way in was a legal brief.
The answers came back in two documents, not one
Here is where the record stops behaving the way most people assume. A Federal Register full text search for the notice's own citation, 91 FR 12516, returns two CFTC proposed rules that draw on it, as of 30 August 2026.
The first is the June proposal already described, which carries the same RIN and answers the middle of the questionnaire. It defines what it means for a contract to "involve" an enumerated activity, sets out the scope of gaming and the categories that fall outside it, adopts public interest factors, and rebuilds the review procedure in 17 CFR 40.11. We have written about the gaming definition as a test an exchange can run on its own book, about what falls outside that test and what falls out with it, and about the procedural half as a schedule rather than a protection. Its comment period ran 45 days and closed on 27 July 2026.
The second is easy to miss. Data Reporting Requirements for Certain Event Contracts, RIN 3038-AF73, was published on 1 July 2026 at 91 FR 40102 and proposes an alternative reporting framework for fully collateralised event contracts under parts 15 through 18 rather than the swap reporting regime. It answers a piece of question 5.a from the March notice, the question about reporting event contract swaps to a swap data repository. Its comment period ran 30 days and closed on 31 July 2026.
Anyone who filed in April on reporting, then watched the docket they filed in, had a month to notice that the follow-up was in a different file with a shorter clock. The second window is not announced inside the first one.
One footnote in July closes the door on most of what was asked
Footnote 123 of the reporting proposal is the most instructive paragraph in this entire sequence. It records what commenters asked for on this topic: reporting designed to identify insider trading and fraud, scaled position reporting thresholds, and mandatory reporting by market participants deploying artificial intelligence driven trading strategies. It records that another commenter supported a regulator-mandated per-contract identifier assigned at listing, following the operating model behind CUSIP, LEI and UPI. Then it says the proposal offers an alternative reporting route with adjusted thresholds, and adds one sentence: "The commenters' requests are otherwise outside the scope of this Proposal."
That is not a rebuff and it is not bad faith. It is what scope means. But it is the clearest demonstration available that a comment does not fail only by being ignored. It can be read, recorded by name, and still land outside the boundary of the document written next. The remedy is not longer letters. It is answering the specific question in the proceeding that can act on it, which is what the notice asked for.
It matters for numbers as well as for procedure. We have argued that the figure a venue leads with says what it thinks matters, and a number nobody is required to file is a number nobody outside the venue will ever check.
Six areas were asked about and two of them produced a proposal
Set the questionnaire against the output and the gap is large, as of 30 August 2026.
- Public interest, enumerated activities and procedure, questions 7 through 28, became the June proposal.
- One sub-part of question 5, on reporting, became the July proposal.
- Everything else in area A, meaning impartial access, resolution disputes and dispute procedures, susceptibility to manipulation, surveillance capacity, position limits, margin for retail and institutional customers, operational risk and blockchain-based venues, has produced no proposal.
- The clearing questions, on how the DCO core principles apply and what would change if these contracts ever traded on margin, have produced no proposal.
- Inside information, questions 29 through 32, covering asymmetric information, events controlled by one person or a small group, cross-market manipulation and the statutory bar on federal employees trading on non-public information, has produced no proposal.
None of that means the questions were abandoned. Rulemakings run at their own pace. But an operator reading the March notice as a preview of the coming framework should notice which half of it has arrived and which half is still open. The half still open is the half about how a venue is run day to day.
Every one of these documents carries a single signature
The voting summary at the end of the March notice reads: "On this matter, Chairman Selig voted in the affirmative. No Commissioner voted in the negative." The identical sentence closes the June proposal and the July reporting proposal.
Take that as a fact about the Commission's composition through this period rather than a criticism of anyone in it. Its consequence for the file is practical. Where a multi-member commission produces dissents and concurrences that outsiders can read and build on, a single-member commission does not. In that setting the comment record is close to the whole of the public argument, and the people who filled it in were the only ones making that argument where anybody could read it.
What a comment that gets quoted actually looks like
The March notice told everyone how to be cited, in the paragraph that opens the questions. It asks commenters to "provide a detailed response, including the rationale for such response, cost and benefit considerations, and relevant supporting information, such as data, or studies when available", and it encourages them to include the assigned topic number of the specific request for comment they are answering, so that staff can sort the file.
Roughly three hundred submissions did that. Seventeen reached the footnotes. Read against the citations, four properties separate the letters that were quoted from the ones that were counted.
- They answer a numbered question rather than the subject in general, which is what makes a submission sortable at all.
- They carry a page number in the citation. Every one of the seventeen is cited to a specific page of the letter, which means the argument sat somewhere findable inside a structured document.
- They bring something the agency did not already have, whether that is trading data, an operational description, a legal reading, or the position of a constituency nobody else represents.
- They accept a boundary. The letters cited for narrow propositions were cited because the proposition was narrow.
Writing the standard down is cheaper in April than in October
This is the part that connects a procedural story to the thing we actually argue for. Directive 05 of our manifest says that standards are set against the jurisdiction that regulates next, not only the one that has not yet. A comment file is the cheapest moment in which that happens. Nobody grants permission, there is no fee, the deadline is published months ahead, and the document sits permanently in a record the drafters demonstrably read.
A venue that can write a serious comment on impartial access, or on how it resolves a disputed settlement, has already done most of the work of writing that standard down internally. A venue that cannot write the comment does not have the standard either, and that is the useful diagnostic. The letter is not the deliverable. The letter is the receipt for work that already exists.
Two windows are open. Comments on Conflicts and Affiliations, which reaches board composition and the exchange that owns its own market maker, close on 5 October 2026, and we looked at what that proposal would actually require when it was published. Comments on the request for comment on listing compute derivatives contracts close on 20 October 2026. Neither is the prediction markets notice, which is the point. The file that decides your category is not always filed under your category's name.
Three and a half thousand people had an opinion this spring. Around three hundred wrote it in a form the agency could use, and seventeen are quoted in the rule that came back. That ratio is not a complaint about the public. It is a clear price list for being heard, posted in advance. If the standard this category needs should come from inside it, the cheapest place to say so is a docket, and the commitment we ask firms to sign is the same argument without a deadline.