SupTech

The name is short for supervisory technology. Regulators use it to receive data directly from firms' systems in standard formats, instead of through spreadsheets and emails, and to run analysis on what arrives. Typical tools include dashboards that flag firms with unusual ratios, text analysis of complaints and filings, network analysis of connected parties, and machine learning that spots outliers in transaction data. The aim is earlier warning of problems and less reporting effort for firms. Limits include data quality, privacy, the need for staff who can interpret model outputs, and the difficulty of explaining automated findings in an enforcement case. Firms feel SupTech through new data requests, automated validation checks on submissions and more frequent supervisory queries. A firm that sees its own numbers the way the supervisor does can fix an outlier before the query arrives.

Every morning, supervisors at a securities regulator review alerts from a system that takes in daily trade data from 150 brokers. It runs pattern detection and flags accounts with repeated order cancellations just before large trades. Supervisors review the alerts each morning. In one case the alerts lead to an inspection that finds weak surveillance at a broker, a gap that quarterly reports had not revealed.