Market surveillance in FICC has undergone, and continues to undergo, significant change as a result of regulation, the evolution of market structure, and technological developments. This Spotlight Review considers these structural and technological changes, in particular the emergence of machine learning trading strategies, and sets out some of the challenges associated with these developments for surveillance teams in FICC markets. The review then examines the role of technology as a potential solution to these challenges, creating as it does opportunities to improve market surveillance through the application of machine learning.
Market surveillance has been an area of regulatory focus over recent years. In particular, the UK’s Financial Conduct Authority (FCA) has focused on the need for firms to continue to improve surveillance in FICC markets and to enhance both the quality and number of suspicious trade submissions relating to FICC markets activity. Furthermore, in the current remote working context, maintaining robust surveillance and suspicious transaction and order reporting has been flagged as a regulatory priority.
However, further enhancing FICC surveillance in increasingly fast moving, complex and data-driven markets is not a simple task. This paper draws on two key structural challenges surveillance teams face in FICC markets: namely data quality and availability, and the increasing sophistication of trading strategies and technologies deployed to support them. The amount of data available to surveillance functions has increased significantly in recent years, driven by regulatory reporting requirements and the proliferation of electronic trading. However, extrapolating signals from these data sets remains challenging given variances in the accuracy, robustness, timeliness and consistency of such data, in particular across different FICC asset classes. Furthermore, growth in algorithmic trading, systematic investment strategies and the nascent adoption of machine learning in trading is materially increasing the speed and complexity of FICC markets. This combination of increased data and trading complexity, and the possibility of new market abuse risks emerging as a result of these developments, may drive the adoption of new, or the improvement of existing, surveillance techniques.
The full report is available at “Monitoring FICC markets and the impact of machine learning“, FICC Market Standards Board, August 2020.