The European Securities and Markets Authority (ESMA) has zeroed in on the use of artificial intelligence and machine learning (AI and ML) in financial markets regulation, in its latest high-level report on trends, risks, and vulnerabilities in the sector.

As jittery markets consider a short- to medium-term future in which weakening growth, rising debt, Brexit uncertainty, and the fallout from the Sino-US trade war threaten instability, how might AI and ML help – or hinder – the work of financial watchdogs?

Authorities such as the European Central Bank and the US Federal Reserve are using natural language processing (NLP), a subsection of AI, to help them identify risks to financial stability, says ESMA.

Another potential application is to help detect trade syndicates in the securities market, using AI’s ability to spot patterns and correlations in data that may not be obvious to human observers.

AI and ML systems are able to analyse years of data, which not only helps to deal with historic volumes much more quickly, but may also uncover complex, networked relationships that have existed over long periods, often involving numerous participants.

Collusive behaviour and price manipulation can be especially hard to detect using traditional methods, explains ESMA. Rules-based systems, such as transaction-monitoring, have high false-positive rates, creating extra – often costly – work for exchanges and regulators.

By contrast, ML-based surveillance systems have, through mathematical optimisation techniques, been able to reduce false positive rates, suggests the report.

ESMA believes that AI/ML tools could also be applied when potential misconduct cases are detected. At present, external human experts are required to verify that cases warrant further investigation. Such experts are costly to employ and are limited in number, so regulators “would benefit from any potential extension of AI/ML technologies into this context”, says the report.

However, professionals should perhaps beware of rushing to automate processes that demand deep expertise and human insight, especially if the impetus is cost-cuts rather than making the business smarter.

That’s the conclusion of a number of other recent reports into Industry 4.0 applications. The rush to deploy new technologies is often tactical and superficial, rather than strategic and considered, they suggest. Indeed, some frauds or illegal behaviours may be driven deeper underground by automation, while ‘bad actors’ could use the same technologies to hide their activities from regulators.

Nevertheless, recent attempts to use ML to detect potential cases of market abuse show promise, claims ESMA.

Some regulators, such as the UK’s Financial Conduct Authority (FCA), have been exploring how best to analyse large datasets to study suspicious trading behaviour. In this context, AI/ML tools may help to identify collusion to manipulate share prices, or circular trading to create false impressions of market interest, explains the report.

Be part of a discussion and connect with like-minded leaders in your sector at our exclusive event series on banking and RegTech.