The mechanics of a modern prop firm are easy to describe and surprisingly hard to classify. A client in Lisbon pays a few hundred euro for an evaluation. They trade a simulated account, on a demo server, against synthetic balances. If they hit a profit target without breaking the drawdown rules they move to a second simulated account, and from then on the firm pays them a percentage of the profit that account shows. Real money leaves the firm. No client money ever went into a market.
Ask five European lawyers whether that requires a MiFID authorisation and you will get answers ranging from "obviously not" to "obviously yes, and the model as sold is unlawful". The honest position is that the perimeter has not been drawn yet, and firms are operating in the gap.
Where the MiFID perimeter actually sits
MiFID II works through two lists in Annex I. Section C names the financial instruments, and contracts for differences appear there explicitly. Section A names the investment services and activities: reception and transmission of orders, execution of orders on behalf of clients, dealing on own account, portfolio management, investment advice, underwriting and placing. Authorisation is required when a firm provides one of the Section A services in relation to one of the Section C instruments, as a regular occupation or business.
Both halves have to be present. That is the entire structure of the argument that follows, and it is why the discussion always returns to the same question: in a simulated account, is there an instrument at all? The rest of the MiFID framework, from client classification to best execution, only engages once you are inside the perimeter.
The case that nothing regulated is happening
The industry position is coherent. The client is not a client in the MiFID sense but a candidate in an assessment. The account is a demo environment. No order the candidate submits reaches a venue, a liquidity provider or any counterparty balance sheet. The number on the screen is a bookkeeping entry in the firm's own database, and the firm has no market exposure arising from it. There is therefore no financial instrument, no execution, no transmission and no portfolio managed on anyone's behalf.
On this reading the evaluation fee buys a service that sits outside financial services regulation entirely: a skills assessment, comparable to any paid examination, with a subsequent commercial arrangement in which the firm chooses to pay a bonus calculated from performance in an environment it controls. Firms that structure carefully lean on the fact that the trader has no claim on any asset, no account balance to withdraw, and no counterparty risk to a market.
The case that something is
The supervisory counter-argument is substance over form. Strip the labels and the transaction looks like this: a person pays a sum of money to a firm, takes a directional view on gold or an index, and receives a cash amount determined by how that underlying moved. That is the economic shape of a contract for differences with an upfront premium and a capped downside. The word "simulated" describes the plumbing, not the payoff.
Two features sharpen that argument. The first is instant funding, where there is no evaluation at all. Without an assessment there is no service to point at, only a fee exchanged for a contingent payout, which removes the strongest part of the industry's defence. The second is any arrangement where trades from the "funded" stage are copied, hedged or mirrored into a live account. Once decisions taken by a paying client cause real orders to be sent to a real venue, the firm has to explain why it is not receiving and transmitting orders, and why the client is not being provided with a service.
Nothing here is legal advice, and the analysis differs by Member State. Anyone building a prop firm aimed at EU residents needs a written opinion from a lawyer qualified in each country they intend to accept clients from, refreshed as national positions move. Treat this article as a map of the arguments, not a conclusion about your own model.
What regulators have actually done
There is no dedicated EU framework for prop trading firms. What exists is a set of national responses that have arrived at different speeds. Supervisors in several Member States have published consumer warnings describing the model, flagging that the firms behind it are typically not authorised and that the accounts involved are simulated. Some have gone further and questioned publicly whether particular offerings fall inside their national implementation of MiFID. Others have said nothing at all, which is not the same as approval.
Two other bodies of law apply regardless of how the MiFID question resolves. Consumer protection rules bite on how the product is described, so a landing page that implies a trader is receiving real capital when the account is simulated is exposed even if no licence is required. And in a small number of Member States there is a live question about whether a paid contest with a monetary prize touches gambling legislation. We track the wider picture in prop firm regulation, and the parallel debate in the United States runs through the CFTC angle on futures prop firms.
What a firm can control while the question is open
Ambiguity is not an excuse for sloppiness, and the firms that get into trouble first are usually the ones whose marketing describes a different product from their terms.
Say what the account is. If it is simulated, the word appears in the hero section, not in clause 14. Avoid "we fund you" language where no capital is transferred. Keep the payout formula contractual and specific, because a discretionary payout that the firm can decline for undefined reasons is both a consumer protection problem and a commercial one. Publish the rule set that can breach an account, and apply it identically to every trader, which is the entire argument for writing consistency rules as measurable code rather than as a paragraph of prose.
On the money side, do not hold client balances. Fees should be revenue on receipt, not a float the trader could argue is theirs, because a firm sitting on redeemable client balances starts to resemble a payment or investment business in a way that invites the exact question it wants to avoid. Geofence the countries whose regulators have taken a position you cannot live with, and geofence properly: IP, KYC country and payment instrument country all agreeing, not a checkbox.
The last control is corporate. Keep the operating entity, the payment flows and the technology cleanly separated, so a change in one country's position becomes a distribution decision rather than an existential one. Firms that started as a single company with one bank account and one domain find that a national ruling forces them to rebuild everything at once. Those that planned for divergence, as most experienced operators do when they read how the model is put together, tend to have options.
"The honest answer is that nobody knows yet. Firms building as though the answer will be 'no licence needed' are placing a bet, and the least they can do is admit to themselves that they are placing one."
— Roman Onta, Executive Director, SINGUARD
Key Takeaways
- MiFID authorisation needs both a Section A service and a Section C instrument, so the whole debate turns on whether a simulated account involves a financial instrument at all.
- Instant funding is the most exposed variant, because removing the evaluation removes the service the fee is supposed to buy.
- Copying or hedging client-driven trades into live accounts creates real order flow, which is the hardest fact for the unregulated reading to explain away.
- Consumer protection and advertising law apply whatever happens with MiFID, so marketing that implies real capital where the account is simulated is a problem today.
Frequently Asked Questions
Do prop firms need a MiFID licence in the EU?
There is no settled EU-wide answer. Firms argue that a simulated account involves no financial instrument and therefore no MiFID investment service. Supervisors in several Member States have questioned whether an evaluation fee paid for a payout linked to price movement is a derivative in substance. The position differs by country and is still developing.
Does copying client trades to a live account change the analysis?
It makes the argument harder to run. Once decisions taken by a paying client cause real orders to be sent to a market or a liquidity provider, the firm has to explain why that is not reception and transmission of orders or a form of portfolio management, and why the client is not receiving an investment service.
Is instant funding more exposed than a two-step evaluation?
Generally yes. A two-step model can at least point to an assessment service delivered in exchange for the fee. Instant funding removes that service, leaving a payment made in exchange for a contingent payout linked to price movement, which is closer to the shape of a derivative contract.