Reviewing surveillance alerts – is it time for a paradigm shift?
30 April 2019
We recently launched the results of our 2019 Market Abuse Surveillance Survey. This follows up on its 2016 predecessor, and focuses on how banks are continuing to evolve their surveillance capabilities to adhere to market abuse regulatory expectations.
The 2019 Surveillance Survey results bring to life just how onerous the alert review process has become and how challenging it is to conduct an exhaustive analysis of the data to eliminate the possible in search of the probable. Over the last 12 months the 21 banks participating in our survey, have raised a combined total of over 40m trade and e-comms alerts.
These alerts are raised by surveillance systems tuned to trigger based on certain characteristics of market abuse potentially being present. Every single alert is currently reviewed!
At the end of this process, our survey highlighted that less than 0.01% of these alerts are submitted as a suspicious transaction and order report (STOR).
Irrespective of whether one believes the number of STORs raised to be either too high or low, a false positive population of 99.99% is an extraordinary statistic. A huge amount of effort is being expended reviewing benign trading behaviour.
Is it time for the alerts generation and review process to undergo a paradigm shift?
The use of predictive modelling and machine learning techniques could help surveillance functions identify high and higher risk alerts and by focussing on a smaller population of higher risk alerts, surveillance functions could spend more time undertaking detailed investigations of this “more possible if not probable” population of high risk alerts.
The question for Surveillance practitioners and regulators is whether they are ready for a shift away from the perceived imperative of reviewing all alerts, and whether they can get comfortable that this will result in better outcomes.
To read the full results from our 2019 Market Abuse Surveillance Survey please click here.