GAB Forum: Machine Learning for Risk-Based Enforcement
Thursday, February 22, 2018 at 11:05AM
Jen Bachner

This month's Government Analytics Breakfast Forum welcomed Jennifer Diamantis, Associate Director and Chief, Office of Market Intelligence and Austin Gerig, Assistant Director, Office of Research and Data Services from the Securities and Exchange Commission (SEC) to present their talk, Using Data Analysis and Machine Learning for Risk-Based Enforcement.

The presenters discussed how the SEC is using analytics to detect and gather evidence of insider trading and other market abuses.  Analytics allow the SEC to identify suspicious activity without receiving specific tips or complaints.  Further, the SEC is now better able to identify other individuals who may be involved with known suspicious activity. 

The SEC relies on a variety of statistical methods to make sense of the data they continuosly receive from the entities they regulate.  Using mostly open-source software such as R and Python, the SEC uses machine learning, text analytics, cluster analysis and network analysis to detect anomalies and abnormal behavior that may warrant further investigation.

The speakers were careful to note that these methods may, on occasion, produce false negative or false positives, so the results are carefully interpreted by those who have significant experience in the financial industry. 

The increased reliance on analytics has helped the SEC save resources by empowering them to accomplish more with less human capital.

A full recording of the presentation can be viewed here

Article originally appeared on Governmental Studies at Johns Hopkins University (
See website for complete article licensing information.