ESMA Briefing on Algorithmic Trading

On 26 February 2026, ESMA published a Supervisory Briefing on Algorithmic Trading in the EU (the “Briefing”). The Briefing provides guidance to investment firms and national competent authorities (“NCAs”) on the supervision of algorithmic trading under MiFID II (and RTS 6). In direct response, the AFM published a news item on 1 April 2026 announcing that it will integrate the Briefing into its supervision and sharing its supervisory priorities for the sector.

Why did ESMA publish the Briefing?

The Briefing was issued by ESMA as a non-binding convergence tool to harmonise national algorithmic trading supervision between EU member states. Since the last ESMA Q&A on the scoping of algorithmic trading in 2022, algorithmic trading has further developed to become the norm in the trading landscape.

Moreover, the Briefing draws on the findings of ESMA’s 2024 Common Supervisory Action (“CSA”) on pre-trade controls (“PTCs”), the outcome of which ESMA wants to transpose into the day-to-day supervision of the NCAs. According to the CSA, while most investment firms have integrated PTCs into their trading activity and risk management framework, practices related to implementation and governance remain divergent and not always robust.

The Briefing complements existing guidance and focuses specifically on areas where divergence in supervisory practice has been identified: governance, testing, outsourcing, and the interpretation of key concepts.

What does the Briefing cover?

Scope of algorithmic trading and the definition of an algorithm

The Briefing clarifies that algorithmic trading under art. 4(1)(39) MiFID II covers any trading in which a computer algorithm determines any parameter of an order, regardless of whether a human also intervenes in the process. For example, instances where a human takes the ultimate decision to execute an order that was prepared by an algorithm fall under the definition of ‘algorithmic trading’.

Key Takeaways

  • The Briefing clarifies the scope of algorithmic trading, the definition of an algorithm and algorithmic trading strategy, and provides guidance on governance, testing, outsourcing and pre-trade controls;
  • The AFM will request the annual RTS 6 self-assessments from a select group of investment firms and banks in Q3 2026;
  • The AFM’s supervisory priorities focus on artificial intelligence and machine learning risks, definitions and retesting of algorithms, and pre- and post-trade controls for DEA providers;
  • Administrative burden is reduced: DORA-covered topics (art. 14 and 18 RTS 6) are excluded from the annual self-assessment, and interim notifications under art. 17(2) and (5) MiFID II are no longer required.

Contact

Jan Jans

Maarten Mol-Huging

Furthermore, ESMA gives the following list of activities that aids NCAs in ascertaining the presence of algorithmic trading:

  • Order Generation Logic – E.g., signal-based trading, quantitative models, or ML-driven strategies including where the activities are not involved in direct execution but influence algorithmic trade decision-making and parameter determination;
  • Execution Strategy Selection – Choosing between execution strategies (e.g., how and when orders are placed);
  • Market Condition Analysis – Analysing market data (e.g. volatility, spread, depth) to determine whether and how to trade. E.g., real-time decision-making based on external inputs;
  • Portfolio Rebalancing Decisions – Deciding to rebalance portfolios or adjust exposures based on pre-defined rules or market conditions;
  • Risk Management Adjustments – Adjusting trading behaviour based on risk metrics (e.g. VaR);
  • Liquidity Detection and Response – Detecting hidden liquidity or fragmented markets and adjust order placement accordingly;
  • Cross-Asset or Cross-Venue Optimisation – Determining which asset or venue to trade based on cost, latency, or execution quality – beyond simple routing.

Interestingly, some of these activities encroach on the previous guidance of ESMA that algorithms that only advise (human) traders of an investment opportunity are out of scope of ‘algorithmic trading’.

ESMA clarifies that an algorithm is defined as a computerised set of instructions or rules that autonomously determines one or more parameters of a trading order. An algorithmic trading strategy is a set of decision logic, implemented through one or more algorithms, that autonomously pursues a defined trading objective (e.g., market making in a specific instrument or on a given venue). Each strategy must be testable, distinguishable, and subject to supervisory scrutiny.

Testing requirements

Firms must test their algorithms, trading systems and strategies before deployment and following any material change or substantial update. A material change is defined as any modification that may alter the behaviour, risk profile, or compliance posture of an algorithm. The Briefing warns against the accumulation of minor recalibrations that, taken together, amount to a material change without triggering a retesting obligation.

This can be particularly relevant for self-learning artificial inteligence (“AI”) or machine learning (“ML”) trading strategies, which are generally updated (near) continually in small steps. ESMA notes that firms are required to timestamp, approve, and record all material changes to a trading strategy, thus requiring firms monitor whether the changes to a self-learning model are material or not.

In any case, ESMA recommends retesting in case of changes to: (i) logic or decision rules; (ii) execution behaviour; (iii) scope; (iv) risk controls; (v) external dependencies; and (vi) adaptive capabilities of the trading algorithm.

In respect of stress testing, the Briefing notes that firms must obtain a reasonable level of assurance that their systems are capable of processing twice the volume of the highest volume of trading (i.e. the full cycle of order generation to post trade processing) over the previous 6 months.
ESMA does note that the expected testing scrutiny is proportional to the complexity of the algorithm concerned. Furthermore, the Briefing stresses the importance of thorough documentation of the testing process.

Outsourcing and third-party algorithms

Where an investment firm uses third-party algorithms or outsources software used in trading, it remains fully and solely responsible for compliance with MiFID II and RTS 6. Outsourcing arrangements must clearly define roles and responsibilities, and must provide the firm with the unconditional ability to monitor, suspend, or terminate algorithmic trading without dependence on the third-party provider. Importantly, firms must continue to be able to understand their algorithmic trading systems.

AI Act and the use of AI in algorithmic trading

When an algorithmic trading system meets the definition of an AI system under the EU AI Act, it must comply with the requirements of that Act in addition to MiFID II and RTS 6. AI-based algorithmic trading is currently excluded from the scope of high-risk use cases under the AI Act, though this list of high-risk AI systems is subject to annual review.

The Briefing recommends that firms and NCAs recognise the use of AI in algorithmic trading, particularly in the annual self assessment under art. 9 RTS 6. Moreover, the Briefing stresses that firms must understand how their AI/ML algorithmic trading systems operate (thus requiring explainability) and that they are able to explain how these influence their algorithms’ decision-making (thus requiring interpretability).

Pre-Trade Controls (PTCs)

All firms must implement, review and evaluate controls – in line with their risk appetite – which are aimed at preventing the sending of erroneous orders to the market. ESMA considers it best practice to assess such controls along the PTC rules for algorithmic traders.

The Briefing indicates that firms engaging in algorithmic trading must implement PTCs for each financial instrument on which algorithmic trading is conducted, which must principally include all the following PTCs: price collars, maximum order values, maximum order volumes, maximum message limits, and repeated automated execution throttles. This includes firms engaging in market making and the quotes generated by those firms.

PTCs must be set collaboratively by the trading, risk management and compliance functions, and must be calibrated to risk management, legal requirements and market circumstances. In instances where a firm insources (a part of) its trading algorithm, it must actively contribute to the calibration of PTCs both practically and through contractual arrangements – as already established in existing ESMA Guidance.

The Briefing introduces a clear distinction between hard blocks (mandatory, non-overridable by traders) and soft blocks (strongly recommended, overridable alerts that trigger before hard limits are reached). These must apply on a holistic basis, e.g., to both parent and child orders.

Firms must test their PTCs at initial deployment, after major changes, and following risk events. Firms are expected to (i) document the calibration methodology; (ii) collect statistics on PTC triggers; and (iii) use these for periodic recalibration. One risk that must specifically be considered when calibrating PTCs, is that of trading signals generated by AI/ML systems of more advanced market participants.

PTCs must be monitored in real time, by two lines of defense: the trader and the independent risk function. The Briefing clarifies that this prevents the head of a trading desk or an IT function from being the risk function.

What does the AFM announcement mean in practice?

The AFM has announced it will integrate the Briefing into its supervision. In Q3 2026, the AFM will request the annual RTS 6 validation report from a selected group of investment firms and banks, where the AFM will pay particular attention to:

  • AI and ML – Firms must have a concrete understanding of the risks of AI/ML, and must ensure that trading algorithms and strategies are explainable, controllable and do not act unintentionally, with clear lines of accountability;
  • Definitions and (re)testing of algorithms – Firms must apply a consistent definition of what constitutes a ‘trading algorithm’ and of ‘material change’;
  • DEA – Pre- and post-trade controls are essential to limit market and credit risks for both DEA providers and their clients. The AFM will include DEA providers and credit institutions in its information requests.

As a simplification measure, firms do not need to include art. 14 and 18 RTS 6 in their self-assessments, as these topics are covered by DORA.

Furthermore, interim notifications under art. 17(2) and (5) MiFID II are no longer required. Instead, firms only need to notify the AFM at commencement and cessation of algorithmic trading on a Dutch trading platform and at commencement and cessation of offering DEA.

What should financial institutions do?

In light of the Briefing and the AFM announcement, firms engaging in algorithmic trading or offering DEA should consider the following actions:

  1. Review whether their activities fall within the scope of algorithmic trading as clarified in the Briefing;
  2. Assess whether their algorithm and trading strategy definitions are consistent and granular enough to support testing and supervision requirements;
  3. Evaluate change management procedures to ensure that accumulated minor recalibrations do not result in untested material changes;
  4. Review third-party arrangements to confirm that the firm retains adequate oversight, control, and the ability to suspend or terminate algorithms without reliance on the provider;
  5. Verify that PTCs are appropriately calibrated and cover all required control types for each financial instrument;
  6. Consider the impact of AI/ML on trading strategies and ensure that systems are explainable and that firms can explain the impact of AI/ML on algorithmic decision making; and
  7. Prepare for the AFM’s Q3 2026 information request by reviewing and bolstering the annual RTS 6 self-assessment, noting the exclusion of art. 14 and 18 RTS 6.

How can we help?

We regularly advise investment firms and banks on their MiFID II obligations, including governance frameworks for AI/ML and algorithmic trading, PTCs, outsourcing arrangements, and RTS 6. We can, for instance, assist with analysing whether your trading activities fall within the expanded scope of algorithmic trading or advising on the interaction between MiFID II, RTS 6 and the EU AI Act.