Almost every argument about prediction markets in 2026 has been an argument about who decides. Which regulator, which court, which sovereign. Great Britain never had that argument, because it never needed one. Two regulators looked at the same product within six weeks of each other and asked a narrower question: what does this contract reference? The answer sorted the book into two piles with two different futures, and the pile that looks safest in the United States turned out to be the harder one in London.
That is worth sitting with, because a venue planning international expansion tends to carry one story about what its product is. The British answer is that the product is not one thing.
The Commission answered before anyone had launched
On 4 February 2026 Brad Enright, Director of Strategy at the Gambling Commission, published the regulator's position on prediction markets in Great Britain. The post starts from the ordinary rule rather than from anything special about the category: a commercial product that meets the statutory definition of gambling has to be licensed and regulated by the Commission, with spread betting the standing exception that sits with the financial regulator instead.
Applied to prediction markets, Enright's conclusion is that the mechanics are familiar. Whatever the presentation, the core of the product is "akin to what in the UK would be described as a 'Betting Exchange'", and so a prediction market operator would appear to fall within the definition of a betting intermediary. The post makes the point that betting exchanges have operated in Great Britain since 2000. The Commission was not being asked to invent a regime. It was being asked whether an existing one already reached.
The government said the same thing twelve days later. Answering written question HL14169 on 16 February 2026, Baroness Twycross told Baroness Bennett of Manor Castle that a prediction market operating in Great Britain would require a Gambling Commission licence and would be classified as a betting intermediary under the Gambling Act 2005, with the department monitoring the position and considering further action if needed.
A betting intermediary is a category that already had a shelf
The definition being invoked is short. Section 13 of the Gambling Act 2005 makes a betting intermediary a person who provides a service designed to facilitate the making or acceptance of bets between others, and says that acting as one counts as providing facilities for betting. An order book where users take opposing sides of a yes or no question fits that sentence without any strain. The drafting predates the product by two decades and still lands on it.
Providing facilities for gambling without the relevant licence is an offence under section 33 of the same Act, carrying imprisonment, a fine, or both on summary conviction. Enright's post is explicit about that exposure, warning unlicensed operators to make sure they are not targeting or transacting with consumers in Great Britain and noting the criminal offences attached to operating without a licence.
What tends to get lost in the summaries is the other half of that sentence. Britain did not close the door on sports and political event contracts. It pointed at a door and named it. A licence exists, applications are possible, and the Commission set out what holding one costs: obligations covering consumer protection, fairness, the integrity of betting markets and the prevention of crime. Those four headings are not a foreign language to anyone who has read our six directives. They are Directive 01 and Directive 03 written by a licensing authority instead of by operators.
The financial half went the other way
Six weeks later the other regulator drew its side of the line. The Financial Conduct Authority published its perimeter report on 26 March 2026, the annual document in which it sets out issues sitting at the edge of its remit. Prediction market products appear there as an entry in their own right, described as speculative products that let consumers take binary yes or no positions on the outcome of future events.
The routing rule in that entry is the cleanest statement of the British position anywhere. Products linked to non financial events such as sporting or political outcomes fall to the Gambling Commission. Products referencing financial or certain climatic events fall inside the FCA perimeter. Nothing in that sentence turns on what the platform calls itself, how it is marketed, or where it is incorporated. It turns entirely on the underlying.
Then comes the part that changes the commercial picture. The FCA records its view that the financial prediction market products it has seen are binary options, which means they sit under the authority's permanent ban on selling binary options to retail consumers. The FCA justifies keeping them there by reference to the "speculative, gambling-like nature of these contracts and the high risk of consumer harm".
That ban is not new and not narrow. The FCA confirmed it in Policy Statement PS19/11, with Handbook rules in force from 2 April 2019, and it reaches further than the European measure it replaced by capturing securitised binary options too. The operative prohibition lives in COBS 22.6 of the FCA Handbook. The perimeter report adds that, following the authority's discussion paper on consumer access to investments, it will consider whether to do further work on access to these products or to clarify the perimeter, so the position is stated as of March 2026 rather than settled forever.
Spread betting shows the border is twenty years old
The instinct is to read this as a new frontier being mapped. It is closer to an old map being consulted. Enright's carve out for spread betting is the tell. Britain has been splitting a wager shaped product between a gambling regulator and a financial regulator since long before anyone used the phrase event contract. The two authorities publish a memorandum of understanding describing how they work alongside each other, which is the sort of document that only exists where a boundary is expected to be crossed regularly.
For an operator the practical consequence is uncomfortable but clarifying. In Britain, prediction market is not a legal category at all. It is a distribution shape that two separate legal categories can wear. A single venue listing sports contracts and inflation contracts is running two regulated businesses under one interface, and the interface is the only part that is unified.
Brussels reached the same test from the other side
On 3 July 2026 the European Securities and Markets Authority issued a public statement on how binary options measures apply to event contracts (reference ESMA35-243228190-8148), prompted by what it and national competent authorities describe as the increased offering of event contracts and the growth of prediction markets. ESMA defines an event contract as an agreement whose financial outcome is binary and depends on a yes or no answer to a question about a future event.
Its classification rule is the FCA rule expressed in continental drafting. Not all event contracts are financial instruments, ESMA says, and only those whose event question relates to an underlying named in Section C(4) to (10) of Annex I of MiFID II qualify. Those that qualify are derivatives, which places them inside the permanent national product intervention measures on binary options that mirror ESMA Decision (EU) 2018/795 and are in force in every EU member state. Marketing, distribution or sale of those contracts to retail clients is prohibited.
ESMA adds a second obligation that is easy to skip and expensive to miss. Because qualifying event contracts are financial instruments, providing investment services in relation to them in the EU requires MiFID authorisation regardless of client category, so distributing them even exclusively to non retail clients needs that permission. The statement also reminds firms that participating in circumvention activity is prohibited, and that the classification work is theirs to do, honestly and in the client's best interests, product by product.
We wrote in August 2026 about nine European gambling authorities converging on one statement. This is a different set of regulators reaching a boundary from the financial side and arriving at a compatible answer. Two supervisory families, no coordination between them required, same organising question.
A contract can be a financial instrument and a bet at once
The most useful line in the ESMA statement sits in a footnote. Event contracts may also classify as a bet under national gambling legislation, and event contracts issued in token form that are not financial instruments may be crypto-assets regulated under MiCA. Read that slowly. The categories are not alternatives competing for one product. They can apply at the same time, in the same country, to the same contract.
Compliance teams building a jurisdiction matrix usually build it as a single choice per market, gambling here, derivatives there, unregulated in the third column. The European drafting says that model is wrong at the cell level. An inflation contract can be a MiFID derivative and, depending on the member state, a bet as well.
This is where the British split stops being a curiosity and becomes an operating constraint. The classification is done per contract, by somebody, against the law of each place a user can reach the book. If the venue does not do it, a regulator will, and the regulator will do it after the contract has already traded.
Washington and London cut the same book in opposite directions
The clearest way to see how badly a single global compliance story fails is to lay two of them side by side. On 13 August 2026 the Washington State Attorney General announced a King County Superior Court order requiring KalshiEX to stop offering wagers in the state on "sports, elections, politics, entertainment, culture, tech and science, or mentions", with an IP and residency geofence due by 19 August 2026 and a multi source solution by 2 September 2026. Reporting by Oregon Public Broadcasting on the scope of Judge John McHale's order records that contracts related to commodities, climate, economics and finance were left running.
Now overlay Britain. Commodities, climate, economics and finance are precisely the underlyings that the FCA perimeter report routes into the financial regime, where the authority's stated view puts them inside the retail binary options ban. Sport and politics, the categories a Washington court switched off, are the categories for which the Gambling Commission says a licence route exists.
The same book, cut by two authorities in the same month, in opposite directions. There is no ordering of the two rulings that produces a consistent global product. A venue that describes itself as a financial marketplace to reassure American regulators is describing the half of its book with no British retail route, and a venue that leans on sports volume is leaning on the half a Washington court removed. We argued in an August 2026 piece on contradictory court rulings that the only defensible posture while a map disagrees with itself is to operate to the strictest live reading. The transatlantic version of that is harsher: there is no reading strict enough to be right everywhere, so the classification has to be done per contract and per country.
What a venue should be able to answer about every line in its book
None of the documents above requires new technology or a new committee. They require an answer that most books do not hold in writing. Five questions, per contract, not per platform.
- What is the underlying, in the terms the law uses? Not the marketing category. Sporting outcome, political outcome, financial variable, climatic variable, official economic statistic. The FCA and ESMA tests both start here and nowhere else.
- Which regime does that answer select in each market you reach? For Great Britain the answer is a Gambling Commission licence or the FCA perimeter, and for the financial side the FCA position as of March 2026 is that retail access is closed.
- Can two regimes apply at once? ESMA says yes. The matrix needs a cell that holds more than one value, and someone has to have checked the gambling law of each member state alongside the MiFID analysis.
- Who signed the classification, and when? A named person and a date, revisited when the contract is amended. A determination nobody owns is a determination that will be revised under pressure.
- What happens to the contracts that fail the test? Delisting is the easy half. Open positions held by users in a market that just closed are the half that generates complaints, and the resolution terms should already say what happens before anybody needs to ask.
This is the same exercise we described when the CFTC published its definition of gaming: a regulator's test is something you can run on your own book first. The British and European versions are, if anything, easier to run, because they ask about the underlying rather than about twelve weighted factors. There is no interpretive skill required to know whether a contract references a football match or an inflation print.
Access is part of the product, not a setting around it
Directive 05 commits signatories to setting standards against the jurisdiction that regulates next, not only the one that has not yet. Britain and the European Union have made that unusually concrete. Both have told operators, in public and in advance, exactly which question they will ask. Neither is waiting for an enforcement action to explain itself.
For a user the consequence is the one we described in our piece on venue risk. Whether a contract is available, and whether it remains available next month, is determined by a classification the user cannot see and did not make. Directive 02 says the things that determine the economics of a trade belong on the surface where the trade happens. A contract's regulatory footing is one of those things, even though almost nobody displays it.
The classification you do not make is made for you
Britain did not decide whether prediction markets are good. It decided what they are, one contract at a time, and let the existing law do the rest. That is a low drama outcome and a demanding one. It leaves no room for the argument that the category is too new to classify, because the classification took a blog post in February and a paragraph in a March report.
The venues that will find this comfortable already know, in writing, what each of their contracts points at and which regime that answer selects. The ones that will find it painful are those for whom prediction market has been doing the work of a legal category. It never was one in Great Britain, and after 3 July 2026 it is not one in the European Union either.
If your organisation already runs this test on its own book, say so publicly. A standard that operators write and publish before a regulator asks is the only kind that travels.