Market surveillance teams are facing unprecedented challenges to identify manipulation and abuse as European regulators increase their scrutiny of market operators and participants through initiatives like the Market Abuse Regulation (MAR) and MiFID II. But machine learning is now available to assist surveillance organizations in combating market abuse.
In 2016, Nasdaq started working toward a machine learning implementation of its SMARTS Market Surveillance product. SMARTS developers worked alongside Nasdaq Nordic surveillance analysts to identify and develop machine intelligence capabilities that could increase the efficiency of manipulation and abuse alerts as well as reduce the number of false positives, which in turn would enable analysts to use their time more effectively.
Once the solution was prototyped and tested against multiple years of historical surveillance analyst activity, Nasdaq Nordic implemented it, marking the first time this type of solution was deployed at an exchange and used as an integral part of the surveillance process.
“In the current environment, it’s important that our team can reduce the amount of noise they see from market activity and spend time where they can contribute their expertise to finding true positives and investigating them effectively,” said Joakim Strid, vice president and head of surveillance at Nasdaq Nordic.
In the future, Nasdaq sees an opportunity to change the alert design based on scoring, which will prioritize incoming alerts in situations where the workload is high.