With AI developing at a significant rate, with both big tech companies investing and global policymakers debating the positives and negatives of innovation and regulation, entities also need to take into account compliance concerns.
This is addressed by Simon Dix, CEO and Founder of DX Compliance, who writes for Payment Expert on the developments of regulatory guidelines surrounding AI, how this will affect buy-side companies, and how being non-compliant could be costly and dangerous.
The UK’s Market Abuse Regulations came into force on 31 December 2020 following the end of the Brexit transition period. While these are similar to regulations in other jurisdictions – including most of the EU and US – both the number and value of fines for failing to meet its requirements have risen significantly over the last two years.
This poses a real commercial risk for financial services businesses, providing an increasingly strong incentive for organisations to review their compliance processes.
However, this is easier said than done for buy-side companies and asset managers, many of which rely on software developed for stock markets and sell-side professionals to meet their legal obligations. All too often, this brings problems for buy-side firms more interested in orders than trades.
Clumsy workarounds may result in alerts being turned off as they generate too many irrelevant notifications, while existing software often fails to draw the ideally-required data across traditional departmental silos, limiting the quality of analysis and risk identification.
With a recent increase in communications from the Financial Conduct Authority (FCA) highlighting its renewed focus on this area and growing expectations from companies when it comes to demonstrating compliance, how can buy-side firms – of all shapes and sizes – make sure they have the systems and controls in place to protect their business?
The first step is to better understand what market abuse might look like in this setting: Insider trading – based on non-public material information; market manipulation – creating a false or misleading appearance of market activity, such as as pump-and-dump schemes, spoofing, or wash trading; and front running – executing orders on a security for one’s own account while taking advantage of advance knowledge of pending orders – are all activities that buy-side companies should proactively look for.
Secondly, we need to acknowledge why buy-side firms may struggle to identify market abuse. Through our own research, we have identified several consistent issues. These include problems monitoring every order correctly, false positives feeding inefficiencies, quality of data governance, issues navigating data ownership and completeness, and challenges around effective model testing systems and processes.
So what about the solutions?
In our view, AI and machine learning hold the key to changing the compliance landscape for buy-side professionals during 2025.
This is in line with a broader trend highlighted by the UK government’s most-recent pledge to accelerate the implementation of AI to realise efficiencies across both industry and the public sector. The FCA has already integrated AI into its own surveillance approach and its ability to process huge amounts of data in almost real time could help organisations spot suspicious activity more quickly and build compliance trails more easily than before.
Pattern recognition can also help asset managers identify useful trends and anomalies in trading data that may give cause for concern, while quicker, more in-depth behavioural analysis allows companies to monitor trader behaviour across a wider variety of channels, increasing their ability to detect deviations from the norm.
For example, breaking down traditional silos to investigate data points from funds, investment managers and asset managers alike helps AML professionals identify more suspicious activity and ensure audit trails comply with legislative requirements.
The ability to support cross product surveillance also makes it easier for AML professionals to detect specific market abuse typologies, such as that involving bonds and equity, spoofing patterns, pump and dump scenarios, and manipulation of stock price to profit from options.
The increased speed at which AI or machine learning-based surveillance systems can now be set up also facilitates enhanced testing and configuration prior to system go-live as well as real-time software evolution.
By monitoring all major asset classes at once, AML professionals can review a single source of truth, reducing unnecessary noise while allowing for a high-degree of parameter tailoring along different asset classes. This ensures systems can be tailored to individual or organisational requirements quickly and effectively, without generating ‘unwanted noise’ that reduces trust in alerts while supporting continuous performance improvement.
Crucially, AI’s ability to process huge quantities of data in almost-real time means that ‘red flags’, such as suspiciously large order volumes or unusual order frequencies, will be picked up more quickly, alongside more-complex activities such as cross-product or cross-market manipulation.
The FCA requires financial companies to not only review how effective their surveillance and controls are, but also to ensure that data is being shared and reviewed between often-disparate areas to create an accurate picture of behaviours and patterns.
Investing in AI or machine-learning led software offers a reliable, cost-effective route to ‘getting it right’ that can evolve with the needs of any financial business, large or small.
This software can ingest different data points – such as from funds and the activities of investment and asset managers; combine it; run it through proprietary risk engines to identify higher-risk behaviours; auto-generate workflows without breaking audit trails; and guide professionals through the decision-making process, finally empowering buy-side asset managers to navigate complex compliance challenges with greater confidence and precision.
Is 2025 the year AI enables the buy-side market – previously the Cinderella of the industry – to access the support it needs to truly mitigate risk?
