Algorithmic trading governance under RTS 6: the controls for systematic strategies

The threshold question is whether the fund is doing algorithmic trading, and the definition is broad. Under the onshored MiFID framework, algorithmic trading means trading in financial instruments where a computer algorithm automatically determines individual order parameters, such as whether to initiate the order, its timing, price or quantity, or how to manage it after submission, with limited or no human intervention. A systematic or quantitative strategy that generates and routes orders from a model is caught by that definition. So is a good deal of automated execution and smart order routing.

The regime gets its conduct expression in the FCA Handbook at MAR 7A, which sets standards for algorithmic trading, direct electronic access and market-making, and which can apply to asset managers using algorithms, not only to sell-side firms and venues. The obligations there are concrete. A firm needs effective systems and controls, with trading systems that are resilient, capacity-tested, subject to thresholds and limits, and incapable of being used contrary to UK MAR or venue rules (MAR 7A.3.2R). It needs business continuity arrangements and full testing (MAR 7A.3.3R). A market-making strategy needs a binding written agreement with the venue and continuous quoting (MAR 7A.3.4R and 7A.3.5R). The firm must notify the FCA immediately once it starts algorithmic trading (MAR 7A.3.6R) and provide information on request within 14 days (MAR 7A.3.7R). For direct electronic access, it needs a suitability assessment and pre-set limits under a binding agreement that makes the firm responsible for the client’s compliance (MAR 7A.4). MAR 7A is the Handbook layer over the onshored RTS 6, and a systematic fund is generally subject to both.

Three common misreadings lead funds to treat themselves as out of scope when they are not. Using brokers’ execution algorithms does not take the fund out of algorithmic trading: where the fund sets order parameters through its own systems or reaches markets through direct electronic access, the obligations attach to the fund, not just the broker. A discretionary fund is not automatically outside the regime either, since automated execution and order-management logic can pull it into scope for those activities. And RTS 6 has no size exemption. Application is proportionate to the nature, scale and complexity of the firm, but the obligations bind a small systematic fund the same way they bind a large one.

Two adjacent activities drag in more funds than expect to be there. Accessing a trading venue through direct electronic access, sponsored access included, brings RTS 6 obligations for that access, with the conformance-testing and control expectations applying to how the fund reaches the market rather than to the strategy alone. Posting two-way quotes on a regular basis can count as a market-making strategy on the venue, which brings the market-making obligations with it, including a market-making agreement with the venue and the associated continuous-presence requirements. Neither is intuitive.

The August 2025 review confirmed where the FCA’s attention sits. It examined how firms comply with the algorithmic-trading framework, read each firm’s most recent RTS 6 self-assessment and validation report, and, while it made no new rules, told firms it expects them to act on the findings. For a systematic hedge fund the safe working assumption is that it is in scope and will have to demonstrate compliance, not that it has been overlooked.

RTS 6 is detailed, but for a systematic fund it resolves into eight controls, and the fund has to be able to evidence each one rather than assert it.

Governance comes first. The firm needs a clear governance framework for its algorithmic trading, with defined lines of responsibility, senior-management ownership, and compliance and risk in the loop when an algorithm is signed off. Accountability runs to the individual. The FCA lists algorithmic trading as a significant-harm function at SYSC 27.7.3R, so the people who design and operate the algorithms have to be certified fit and proper at least annually. A firm running systematic strategies that has not certified those staff has missed a named requirement.

Testing must precede deployment. Before an algorithm goes live, and again after any material change, it has to be tested: conformance testing with the venues and brokers the firm trades through, testing in a non-live environment that cannot contribute to disorderly trading, and simulation and stress testing of how the algorithm behaves under extreme but plausible conditions. Untested deployment is the disorderly-trading risk the standard is aimed at.

Pre-trade controls are hard limits that stop erroneous or disorderly orders. The firm needs hard limits that stop an algorithm sending erroneous or disorderly orders: price collars that block orders outside a sensible band, maximum order values and volumes, and caps on how many messages an algorithm can send, throttles on repeated automated execution included. They have to be set at real values, not left at theoretical maximums.

Live trading has to be monitored in real time by staff who can step in, with alerts for limit breaches and abnormal behaviour, backed by post-trade controls that reconcile and review activity after the fact.

Kill functionality means the firm can cancel any or all of its unexecuted orders at once, and it has to be clear who is authorised to pull it, how, and when. A kill switch has to be tested and operable by more than one named person to meet RTS 6.

Changes to algorithms have to be controlled, version-managed, tested before they go live, and recorded, so the firm can show what was running when and that each change passed the testing gate.

The firm also needs staff with the skills to manage its algorithmic trading, business continuity for the systems, information security matched to the risk, and record-keeping good enough to reconstruct its algorithmic trading for the regulator.

Last, RTS 6 requires an annual self-assessment and validation of compliance with the standard, written up in a report, validated, and approved by senior management. This is the document the FCA asks for first.

Market conduct is built into all eight controls, not a separate workstream. The same automation that makes a strategy efficient can, mishandled, throw off abusive patterns, spoofing, layering, quote stuffing or momentum ignition, that UK MAR prohibits and that MAR 7A and RTS 6 expect the firm to design against. The obligation has two sides. The controls have to be built so the algorithms do not generate disorderly or manipulative order flow in the first place, and the firm’s surveillance has to be configured to catch it if they do. Algorithms must be assessed for their capacity to manipulate, and surveillance must include a scenario for algorithmic manipulation.

Article 9 of RTS 6 requires the firm to run a self-assessment and validation at least once a year and to produce a report. The report has to cover compliance with the whole of RTS 6: the algorithmic-trading systems, the controls, the governance, the testing, the kill functionality and the rest. It has to be validated, either internally by a function independent of the one that built the systems or with external support, approved by senior management, and kept so the FCA can call for it.

This is where the August 2025 review concentrated, and its expectations are specific. The self-assessment should cover every element of RTS 6, outsourcing and training included. It should link to the firm’s live policy repositories rather than describe a framework that no longer matches what the desk does. It should identify gaps with named remediation owners and timelines instead of asserting comfort, and the review pushed firms toward external validation and action-tracked findings.

Funds skip it because it is annual, demanding, needs cross-functional input from quant, technology, risk and compliance, and carries no external deadline until the FCA asks. A fund can run profitable systematic strategies for years without ever writing the report. When a supervisor, an allocator’s due-diligence team, or an exchange membership review asks for it and there is nothing to hand over, that gap is direct evidence the firm has not assessed its compliance with the regime governing its core activity.

A small fund can meet RTS 6, but only by design. A small systematic fund meets RTS 6 proportionately: a lean governance framework that records who signs off an algorithm, a testing checklist applied consistently before deployment and after material change, pre-trade limits set to real considered values, a kill switch with at least two named authorised operators, change control through proper version management, and an annual self-assessment that gains from external validation. Proportionate means a smaller firm builds the same controls at a scale it can actually run, and evidences them just as clearly.

Each is a requirement, and each has to be evidenced.

This insight is provided for general informational purposes only and doesn’t constitute legal, investment, or regulatory advice.