MiFID II

The EU framework in force since 2018. For HFT it adds algorithm registration and testing, mandatory pre-trade risk controls and kill switches, order-to-trade limits, market-maker agreements, and microsecond clock sync.

The idea

What is MiFID II / MiFIR?

MiFID II (the second Markets in Financial Instruments Directive) and its paired regulation MiFIR are the EU's framework for investment services and trading venues, in force from 3 January 2018. They replaced the 2007 MiFID I, hugely expanding the rules on electronic and algorithmic trading, transparency, and investor protection. The UK adopted them, then began diverging after Brexit.

The intuition first: where Reg NMS is a market-structure rulebook centred on price protection and the consolidated tape, MiFID II is broader and more conduct- and controls-focused. It governs who may trade algorithmically and how (registration, risk controls, testing), how time is measured (clock sync), how prices may be quoted (tick sizes), and how venues and internalisers must behave (transparency, best execution). It is the closest thing to a single, explicit HFT rulebook anywhere.

The structure: a Directive (MiFID II, transposed into each member state's law) plus a directly-applicable Regulation (MiFIR), fleshed out by detailed technical standards, the RTS (Regulatory Technical Standards), which carry the specifics traders care about (RTS 6 for algo controls, RTS 11 for tick sizes, RTS 25 for clock sync). (Source: Directive 2014/65/EU and Regulation (EU) No 600/2014, applying from 3 January 2018.)

Why it matters even to a non-EU reader: MiFID II is the world's most explicit attempt to regulate HFT as a named activity. Its definition and its control regime are the clearest available statement of "what counts as HFT and what a regulator demands of it", useful as a reference even for a crypto trader whose venue imposes none of it.

How MiFID II defines high-frequency trading

MiFID II is notable for writing an explicit, operational definition of HFT into law, which most regimes do not. A firm is doing "high-frequency algorithmic trading" if it uses low-latency infrastructure (co-location, proximity hosting, high-speed direct access), an algorithm that determines order parameters with no human intervention for individual orders, and generates high message rates (orders, quotes, cancellations).

The three-part test, in plain terms (from MiFID II Art. 4(1)(40) and the technical standards): infrastructure, the firm uses co-location, proximity hosting or high-speed direct electronic access, i.e. it has paid for speed; autonomy, the system determines order initiation, generation, routing or execution without human intervention for individual orders, so it is genuinely algorithmic, not a human clicking fast; and message intensity, it produces high intraday message rates (the standards give quantitative thresholds for "high").

Why an explicit definition matters: it makes "are you an HFT?" an answerable, auditable question, which in turn triggers specific obligations (registration, the algo controls below). It also draws a clean line a crypto or prediction-market reader can borrow: when this atlas says "HFT", this three-part test is a precise version of what it means.

The honest note: the definition is a regulatory trigger, not a value judgement. Meeting it is not wrongdoing: most legitimate market makers and arbitrageurs meet it; it simply pulls them into the controls regime. The conduct that is illegal (spoofing, layering) is governed by market-abuse law, separately.

Algorithmic-trading and direct-access controls (RTS 6 and DEA)

MiFID II requires any firm engaged in algorithmic trading to register, and to maintain systems and risk controls for its algos: pre-trade limits, testing and conformance before deployment, kill-switch functionality, business-continuity, and ongoing monitoring. Firms offering direct electronic access (DEA) to clients must apply the same controls to those clients' flow. These are the EU/UK parallel to the SEC Market Access Rule.

The core obligations (MiFID II Art. 17; RTS 6): pre-trade risk controls, order-size and price limits, position limits, message-rate limits, exactly the pre-trade gate this atlas describes, made a legal requirement; a kill switch, the ability to immediately cancel orders and halt trading, the kill-switch capability, mandated; testing and conformance, where algos must be tested (including against a venue's conformance suite) and not deployed live without it, the backtesting and simulation discipline made compulsory; and monitoring and records, meaning real-time monitoring, and records sufficient to reconstruct what an algo did.

Direct Electronic Access (DEA): where a member firm lets a client send orders through its membership, the firm must apply pre-trade controls to that client's flow and remain responsible for it, closing the "naked sponsored access" gap, just as Rule 15c3-5 does in the US.

The through-line: MiFID II makes the entire risk-management topic a legal floor, not a best practice. If you trade algorithmically in the EU/UK, the pre-trade limits and kill switch are not optional engineering; they are a registration condition.

Clock synchronisation (RTS 25)

RTS 25 requires firms and venues to synchronise their business clocks to UTC to a defined accuracy (for HFT activity, to within 100 microseconds of UTC, timestamped to the microsecond). The point is that, in a market where events happen in microseconds across many venues, you cannot reconstruct what happened in what order unless everyone's clocks agree.

One-line gloss: if you cannot trust the timestamps, you cannot sequence the events, so the law mandates a shared, accurate clock. Why it is a genuinely important rule, not a technicality: surveillance and abuse detection depend on ordering, because to prove or disprove spoofing or a manipulation cascade, a regulator must reconstruct the precise sequence of orders and cancels across venues, which is impossible if clocks disagree by milliseconds; best-execution and cross-venue analysis need aligned time to compare prices and fills across venues meaningfully; and it standardises something HFT firms already needed internally (accurate, traceable time from GPS- or PTP-disciplined clocks) and makes it auditable.

The engineering link: this is why a serious system runs a disciplined time source (GPS, PTP) and timestamps at the wire (how data is recorded, messaging protocols). RTS 25 turned good practice into a requirement, and made clock-sync-aligned, microsecond-timestamped data a compliance asset.

The tick-size regime (RTS 11)

MiFID II imposes a harmonised tick-size regime: the minimum price increment for a stock is set by a band table indexed on the instrument's price and its liquidity (trade frequency). Liquid, high-priced names get coarser ticks; less liquid or cheaper names get finer ones. It ended the "race to the bottom" where venues competed by offering ever-finer ticks.

Why a table rather than one tick (as in the US penny): MiFID II ties the tick to both price and liquidity, so the increment scales with the instrument. A €5 liquid stock and a €500 liquid stock sit in different bands.

The strategic consequence is the same logic as the Reg NMS tick debate: the tick sets the minimum spread and therefore the value of queue position and the economics of market making. A coarser tick means a wider minimum spread, a longer queue and more reward to being first in line; a finer tick compresses both. By harmonising ticks across EU venues, MiFID II removed tick competition as a venue-differentiation lever and standardised the queue-and-spread terrain market makers operate on.

The pre-MiFID-II problem it fixed: venues had competed by shaving ticks finer to attract aggressive flow, which fragmented liquidity across incompatible increments. The harmonised regime stopped that: regulation deliberately tuning the microstructure dial, exactly this section's theme.

Best execution, systematic internalisers, and the wider regime

MiFID II also imposed best-execution duties (take all sufficient steps to get the best result for clients, and, originally, publish execution-quality and venue reports), and a systematic internaliser (SI) regime governing firms that internalise client flow against their own book. Both shape where and how flow is executed, and both have been pruned in the UK's post-Brexit divergence.

Best execution: firms must take "all sufficient steps" to obtain the best result for clients across price, cost, speed and likelihood of execution. MiFID II originally required detailed public reports (RTS 27 for venue execution quality and RTS 28 for the top-five venues used) but the UK scrapped these reports as low-value, and the EU has also moved to suspend or revise them. The duty remains; the reporting is being unwound. (See why execution algorithms exist for the operational side.)

Systematic internalisers (SIs): a firm that, on an organised, frequent, systematic basis, deals on own account by executing client orders outside a venue is an SI, subject to quoting and transparency obligations. The SI regime governs a large slice of European flow that never touches a lit order book, relevant to fragmentation and to where liquidity actually sits. The wider regime (noted, not exhaustive): MiFIR transparency (pre- and post-trade), the share-trading obligation, the double-volume-cap on dark trading, all part of the same framework and all live subjects of the EU MiFIR review and UK divergence.

Post-Brexit UK divergence

The UK onshored MiFID II at Brexit, then began diverging via the Wholesale Markets Review and the FCA assuming rule-making. The UK has scrapped the RTS 27/28 best-execution reports, revisited the share-trading obligation and double-volume-cap, and is reshaping the regime, while the EU runs its own MiFIR review. "MiFID II" is therefore now two related but increasingly distinct rulebooks.

The mechanism: post-Brexit, EU regulation was copied into UK law ("onshored"), so the UK started from an identical MiFID II. The Wholesale Markets Review (HM Treasury/FCA) then began amending it for UK markets, and the FCA took over the detailed rules from ESMA. What has diverged (verify current positions against the FCA rulebook, as of 2026): the RTS 27/28 reports removed; the double-volume-cap and share-trading obligation revisited or relaxed; consultations on the tick regime, transparency and a UK consolidated tape.

The practical point for a trader: if you operate in both London and the EU, you increasingly face two rulebooks, not one, and the divergence is an ongoing process, so the right move is to read the FCA and ESMA sources directly rather than assume "MiFID II" is monolithic. The core HFT obligations (algo controls, clock sync, the HFT definition) remain broadly aligned; the transparency and reporting periphery is where the two are pulling apart.

Worked example

Concrete, dated reference points (as of 2026; verify against the FCA/ESMA rulebooks and the technical standards). The HFT definition test, applied: a firm co-locates at a venue (infrastructure ✓), runs an arbitrage algo that sends and cancels orders with no human touching individual orders (autonomy ✓), and generates well above the RTS message-rate threshold (intensity ✓). It meets all three limbs, so it is "high-frequency algorithmic trading" under MiFID II and must register and run the RTS 6 controls. A discretionary trader clicking fast from a desk fails the autonomy limb, so they are not HFT.

Clock-sync requirement: for HFT activity, business clocks must be within 100 µs of UTC and timestamped to 1 µs granularity (RTS 25). In practice that means a GPS- or PTP-disciplined clock and wire-level timestamping, the same infrastructure described on messaging protocols, now a compliance obligation rather than just good engineering. Tick-size band (illustrative): under the RTS 11 table, a liquid stock priced around €50 might sit in a band with a €0.005 tick, while a €5 liquid stock sits in a finer band, the increment scaling with price and liquidity, versus the US flat \$0.01 penny (Reg NMS). The exact bands are a table; consult RTS 11 for the specific instrument.

The divergence, concretely: a firm running the same strategy in London and Frankfurt in 2026 produces best-execution evidence under UK rules (no RTS 27/28 reports, FCA framing) for the London book and under EU rules for the Frankfurt book: two compliance footprints from one strategy, the direct cost of divergence. Figures and thresholds are specific, dated regulatory parameters, so reverify against the current technical standards and the FCA/ESMA rulebooks before relying on them (as of 2026). This is educational only and not investment advice.