Securities regulators can better use big data and machine learning to help identify possible misconduct, and to decide which cases to prosecute, according to new report from Toronto-based C.D. Howe Institute.
The report, Regulatory Reform in Ontario: Machine Learning and Regulation, notes that businesses are already making extensive use of these technologies.
“There is probably not one large financial services firm in Canada that has not looked into using machine learning to predict which trades to make and which trades to avoid,” the report says.
Regulators should be using the same tools to improve oversight, streamline regulation, cut needless red tape, and better allocate scarce resources, the report says.
It suggests that analytics can help regulators predict which investigations to undertake. “With the right data and appropriate data analytics, predictions can be made about where to best place investigation limited resources,” the report says.
“One could easily imagine similar tools used to detect insider trading or fraudulent securities sales,” it says, noting that the B.C. Securities Commission “has done innovative work in this field, introducing a number of predictive risk models to improve targeting.”
Regulators can also use these tools to predict how a particular case will be adjudicated, the report suggests, “Regulators can then avoid wasting resources litigating cases they are likely to lose.”
Additionally these tools can reduce the administrative burden by enabling regulators to provide faster, more specific responses to industry queries, the report says, rather than relying on general guidance to guide their decisions.
“The effective use of big data in regulation is one of the most important steps regulators can take in the near future,” says Anthony Niblett, the report’s author, in a statement.