Surveillance for the prevention of market abuse is being fundamentally reshaped by nascent techniques that apply artificial intelligence (AI), behavioral analytics, and holistic solutions across both trade and communications data to detect suspicious activity. Market participants have in the past eked by with reactive, threshold-based alerts designed to primarily satisfy regulatory requirements. This is no longer sufficient.
With the regulatory landscape quickly tightening around the Market Abuse Regulation (MAR), the upcoming revised Markets in Financial Instruments Directive (MiFID II), and the General Data Protection Regulation (GDPR), market participants should expect increased scrutiny. Outside of Europe, market participants are also still contending with regulations such as the U.S.’s Dodd-Frank, Hong Kong’s Securities and Futures Ordinance, and Singapore’s Market Misconduct Enforcement Regime. Total fines for market manipulation hit a high point in 2015, reaching nearly US$ 9 billion, but Opimas expects noncompliance with MiFID II to be cited for market misconduct fines as well, once the regulation takes effect in January 2018.
The use of AI in surveillance has the potential to produce more intelligent alerts, improve automation, and reduce the manpower needed to manage investigations. If applied judiciously, it has the potential to pre-emptively identify some of the suspicious behavior indicative of future misconduct, before any grave damage is incurred. But adoption of these approaches is still limited. About 80% of trade and communications alerts currently rely on static rules-based thresholds. By 2022, Opimas expects this figure to halve, and be supplanted by more intelligent alerting methodologies. Some of the solutions already being sold to the market include K-means clustering, support vector machines, neural networks, relationship mapping, behavioral analytics, as well as holistic approaches.
Much of the market still approaches surveillance in a siloed manner, with separate systems creating alerts of suspicious behaviors for each communication channel and asset class. This evolution has been inevitable, given the piecemeal expansion of regulatory requirements and the challenges unique to each data set. The result is that compliance teams have often tacked together ill-fitting solutions for speedy regulatory compliance, or expanded legacy systems to haphazardly analyze additional data sets ineffectively.
Distinct data channels require targeted approaches to handle the unique challenges they present. Applying the same algorithm to fixed income as one would to equities, for example, is useless. Communication channels present their own difficulties. The result is that compliance teams are overwhelmed with a flood of alerts, nearly negating the possibility for thoughtful investigation and identification of true positives.
This siloed nature of surveillance also makes MiFID II’s required trade reconstruction quite difficult. Reproducing all communications and documents related to a trade or order, chronologically, within 72 hours of a client or regulator’s request is too often still a manual, time-intensive process. Vendors like NICE Actimize, Nasdaq’s SMARTS, Bloomberg, IBM, and newcomer Behavox are positioning themselves to handle trade and communication monitoring in a holistic manner.
Figure 1. Surveillance Headcount 2017-2021
While visualization and a user-friendly case management system were the selling points of the past, the focus for market participants is now firmly around surveillance approaches that reduce costs by improving the quality of alerts and automate investigations. With false alert reduction as the primary driver, Opimas expects spending on technology to rise—up to US$1.4 billion by 2021.
Market participants will increase their spending on automated surveillance solutions that employ AI, machine learning, and behavioral analytics to enable preemptive discovery of suspicious behavior before damages are incurred. As many manual processes are automated, we will also see the headcount of surveillance teams start to decline in 2019 (see Figure 1).
While established solution providers like NICE Actimize, Nasdaq’s SMARTS, Bloomberg, and b-next have a solid client base, this space has been flooded with newcomers fighting for the surveillance slice of the US$19 billion spend across capital markets regulatory technology, commonly referred to as RegTech (see Opimas report FinTech Spending and Innovation in Capital Markets). With Ancoa picked up by Cinnober and Sybenetix by Nasdaq earlier this year, we expect to see a continued shakeout of this market over the next 12 months.
In this report, Opimas provides an overview of the evolution of trade and communication surveillance for the prevention and detection of market manipulation in the context of demanding regulations. Case studies of BNP Paribas, London Stock Exchange, and TP ICAP demonstrate approaches currently undertaken by market participants. This report details the challenges presented by each monitored data channel: trade/order, e-communications, voice, and mobile. Alongside highlighting the advanced surveillance techniques select vendors currently offer, Opimas estimates their future rate of adoption and impact on total IT spend and headcount in the industry. The report closes with a mapping of vendors active in surveillance for the detection and prevention of market abuse.