Prediction markets have outgrown their origins. What started as a niche crypto experiment has matured into a sector that institutional finance now takes seriously: Polymarket returned to the United States in 2026 as a fully regulated exchange, Intercontinental Exchange has invested billions into the space, and some analysts project that trading volumes across the sector could reach a trillion dollars by the end of the decade. Wherever people disagree about the future, someone is now building a market on it.

Sooner or later, every founder building in this space arrives at the same question, and usually later than they should have: if you bring the Polymarket model to Europe, which regulatory regime are you actually operating in? The instinctive answer, that these are not financial products but simply "event contracts", is precisely the answer European regulators have now rejected. In this first instalment of our Prediction Market Series, we set out the analytical framework we apply when assessing prediction markets under MiFID II, and we run the archetypal Polymarket-style model through it.

The American detour, and why it matters for Europe

The natural starting point is the United States, because the American experience explains why financial services regulation, rather than gambling law alone, has become the primary lens through which prediction markets are analysed.

Polymarket's US story is genuinely instructive. In January 2022, the CFTC ordered the platform to pay a USD 1.4 million penalty for offering off-exchange event-based binary options without being registered as a designated contract market or a swap execution facility. The language of that order deserves attention. The regulator did not treat a market such as "Will ETH be above $2,500 on July 22?" as some novel and uncategorised product. It treated it as a binary option: a familiar, and heavily regulated, derivative wearing an unfamiliar wrapper.

What followed is equally telling. Polymarket's route back into the American market ran directly through traditional derivatives regulation. The company acquired a CFTC-licensed derivatives exchange in 2025, and in November of that year it obtained an Amended Order of Designation permitting it to operate as an intermediated designated contract market, subject to the full set of requirements that apply to federally regulated US exchanges. That is the same regulatory tier occupied by the CME. In other words, the largest prediction market in the world did not escape financial regulation; it embraced it, because embracing it was the only durable route to scale.

The lesson here is not US-specific. It is that regulators around the world analyse prediction markets by reference to economic substance rather than branding, and the European Union has now said the same thing in terms.

ESMA takes a position

On 3 July 2026, ESMA published a public statement on event contracts, its first direct supervisory intervention in the prediction market space. For anyone operating or planning to operate in Europe, the statement repays careful reading, but its core messages can be summarised as follows.

First, an event contract with a binary payoff, meaning a fixed amount if the event resolves one way and nothing if it resolves the other, may qualify as a financial instrument where the underlying event relates to one of the categories listed in Section C(4) to C(10) of Annex I to MiFID II. Those categories include commodity prices, interest rates, currencies, financial indices and measures, climatic variables and official economic statistics. Where a contract falls within them, it is a derivative.

Second, and this is where the statement bites hardest, a contract of that kind is a binary option for the purposes of the product intervention measures which ESMA originally adopted in 2018 and which have since been embedded in permanent national measures across the Union. The consequence is stark: the marketing, distribution or sale of such contracts to EU retail clients is prohibited outright.

Third, labels are irrelevant. ESMA was explicit that describing a product as an "event contract" or a "prediction market position" does not remove it from MiFID II if the contract functions as a derivative in substance. Any compliance strategy that rests on terminology rather than product design has been closed off.

Fourth, even distribution confined to non-retail clients does not remove the need for authorisation. Where investment services are being provided in connection with these contracts, MiFID II authorisation is required regardless of the client base.

The statement did not arrive in isolation. The European Commission is consulting, until 30 September 2026, on how prediction contracts should be treated within the broader MiCA review, and in June 2026 nine European gambling regulators pledged coordinated action against unlicensed prediction market platforms. The perimeter is closing from three directions at once: securities regulation, gambling law and crypto-asset regulation. Against that backdrop, the question of whether the Polymarket model is regulated under MiFID II deserves a properly reasoned answer rather than a hopeful assumption.

That answer, in truth, is "it depends". But it depends on precisely identifiable design features, and two questions decide almost everything.

Question 1: Is the contract a financial instrument?

MiFID II's derivative categories are anchored to a single organising concept: settlement determined by reference to an underlying price, level, index or measure. In our regulatory perimeter work, we have found it useful to sort prediction market contracts into three archetypes, because nearly every structure we encounter in practice falls into one of them.

Archetype 1: pure event resolution. These are contracts that resolve solely on whether a predefined factual condition occurs. Will candidate X win the election? Will this film win Best Picture? Will a particular piece of legislation pass before year end? There is no notional exposure, no payoff calculated from price movements, no strike price and no mark-to-market mechanism. Where the resolving event has no connection to an underlying listed in Annex I, Section C, it becomes very difficult to characterise the contract as a MiFID II derivative at all. That does not mean the contract is unregulated. It most likely pushes the analysis into national gambling law, which is a subject we will take up properly in a later instalment of this series. But it does sit outside the financial instruments perimeter.

Archetype 2: threshold-based price exposure. These are contracts that pay a fixed amount if a price is above, below or within a specified level at a specified time. Will Bitcoin close above $100,000 on Friday? Will German day-ahead power exceed EUR 120 per MWh at noon? The framing is that of an event, but the "event" is nothing more than a price observation, and the threshold functions economically as a strike price. The payoff profile is that of a digital option on the underlying, and the user is acquiring exposure to a commodity or financial price level in everything but name. These contracts are highly likely to be characterised as derivatives, and this is exactly the structure that ESMA's July 2026 statement targets. It is worth noting that it is also the same structure the CFTC pointed to in its 2022 order against Polymarket. On both sides of the Atlantic, the threshold market is where regulators draw the line.

Archetype 3: price-difference exposure. These are contracts under which settlement is calculated from price differences, averages, curves or formulas applied to a notional amount, potentially with margining, leverage or close-out mechanics layered on top. Structures of this kind replicate the economics of forwards, swaps or contracts for differences, and they constitute derivative exposure in substance. There is little analytical ambiguity here; the debate, if any, concerns which derivative category applies rather than whether one does.

The uncomfortable truth for anyone operating a Polymarket-style, list-everything marketplace is that this classification exercise is carried out market by market, not platform by platform. The same order book routinely hosts Archetype 1 contracts, such as elections, awards and sporting outcomes, alongside Archetype 2 contracts covering crypto prices, central bank rate decisions expressed as threshold questions, inflation prints and commodity levels. The elections book raises a gambling law question. The "Bitcoin above $X" book is, on ESMA's stated view, a binary option whose retail distribution in the EU is prohibited. A single product category can therefore drag part of a platform across the MiFID II perimeter while the remainder stays outside it, and a compliance posture built on the assumption that the platform has one uniform regulatory character will not survive contact with a supervisor.

Question 2: Is the platform a trading venue?

Even where individual contracts qualify as financial instruments, a second and independent analysis applies to the platform itself, and this is where the architecture of the Polymarket model becomes decisive.

A multilateral trading facility under MiFID II is, in essence, a multilateral system that brings together multiple third-party buying and selling interests in financial instruments, under non-discretionary rules, in a way that results in contracts. Stripped of the statutory language, the analysis turns on three design features. The first is multilateralism: do users trade with each other, or only bilaterally against the operator? The second is transferability: can positions be sold on to other users before the market resolves? The third is matching: does the platform operate an order book or matching engine whose function is to bring third-party interests together?

The Polymarket model answers yes to all three. It operates a central limit order book, positions are freely tradeable between users on a peer-to-peer basis until resolution, and the platform's entire value proposition is the continuous matching of third-party buying and selling interests at market-discovered prices. If the contracts traded on such a system are financial instruments, that architecture sits squarely within trading venue territory, with the authorisation requirements, capital requirements, transparency obligations and organisational rules that follow. And even below the venue threshold, the facilitation of third-party transactions in financial instruments can engage other investment services, most obviously reception and transmission of orders, which carries its own authorisation requirement.

The contrast is a deliberately constrained design in which users contract bilaterally with the operator, positions are non-transferable, there is no secondary market and no user interest is ever matched against another. A structure of that kind does not operate a multilateral system, whatever else it may be. The point founders should absorb is that the regulatory cost of the Polymarket model's greatest commercial strength, namely continuous, liquid, tradeable markets with real price discovery, is that transferability and multilateralism are themselves perimeter triggers. The features that make the product good are the features that make it regulated.

The cross-border trap: reverse solicitation is not a strategy

There is a recurring assumption among founders in this space that deserves to be addressed directly: the idea that a platform can operate from outside the EEA, refrain from formally "offering" into the Union, and simply serve whichever European users happen to arrive.

Under Article 42 of MiFID II, a third-country firm may provide an investment service to an EU client without authorisation only where the client initiates the service at its own exclusive initiative, and ESMA interprets this exemption narrowly and restrictively. Any form of direct or indirect marketing into the EU, localised interfaces, EU-targeted content, affiliate or influencer campaigns aimed at European audiences, or promotional activity through digital channels can defeat reliance on the exemption. Even where it applies, it covers only the specific service the client initiated and the relationship strictly connected to it. Reverse solicitation is, properly understood, a defence available in cases of genuinely incidental contact. It is not a market-access model, and it becomes progressively less defensible the more successful a platform's growth marketing becomes, which is precisely when the exposure matters most. Where European exposure is commercially material, the robust structures involve an authorised EU entity, whether directly or through a joint venture, and we will return to structuring options later in this series.

What this means if you are building one

Distilled for founders and operators, the framework produces a set of design-stage questions that are far cheaper to answer before launch than after a supervisory enquiry.

Start by auditing the market catalogue against the three archetypes. Anything that resolves on a price, rate, index or economic statistic is presumptively an Archetype 2 problem, and under ESMA's July 2026 statement it is likely to constitute a binary option whose retail distribution in the EU is prohibited. Treat transferability as a regulatory decision rather than a purely product one, because tradeable positions combined with third-party matching is the anatomy of a trading venue. Do not build a European go-to-market strategy on reverse solicitation, for the reasons set out above. Map the exposure that exists outside MiFID II as well, since contracts that fall outside the financial instruments perimeter do not fall into a vacuum; gambling law and, prospectively, the outcome of the MiCA review pick up where MiFID II stops. And watch the Commission's consultation closely, because the treatment of prediction contracts within the MiCA review has the potential to reshape this entire analysis. Responses close on 30 September 2026.

The conclusion we would offer is this: prediction markets are not unregulated in Europe. They are, for the moment, uncategorised, and as ESMA has just reminded the market, uncategorised products get analysed under the oldest tool in the regulatory toolbox, which is economic substance. The platforms that thrive in Europe will be the ones that treated classification as a design input rather than a post-launch discovery.

Speak to us

Axis Group advises operators, founders and investors on the regulatory perimeter for prediction markets and event contracts across the EU and beyond, from initial product classification and platform structuring through to authorisation strategy and cross-border market access. If you are building, funding or assessing a prediction market and want to understand where your model sits before a regulator answers the question for you, get in touch with our team. We would be glad to discuss your project.

In Part II of the Prediction Market Series, we cross to the other side of the perimeter and ask when a prediction market amounts to gambling, and why the skill-versus-chance question matters more in Europe than most founders expect.

---

This publication is provided for general information purposes only and does not constitute legal or regulatory advice. The classification of any specific product or platform requires an assessment of its particular contractual and operational features.