Martina Rejsjo, our VP of Product Management, shares insights into the latest industry trends, surveillance updates, and product developments. Her updates dive into new features, best practices, and strategic roadmaps designed to keep you informed and ahead of the curve.
The question on my mind this quarter, and I don’t think I am alone, is “What is the right calibration for a surveillance program”.
Calibration is a complex challenge, especially in dynamic markets. It’s tempting to adjust thresholds to handle manageable alert volumes: tightening them to reduce noise or loosening them to catch more activity. But this approach is risky, as it often shifts the focus from finding genuine threats to managing alert fatigue.
Reducing false positives by narrowing the data capture scope may seem like a straightforward solution, but it’s insufficient. Firms often miss critical risk factors or potential manipulation, leaving them vulnerable when regulators demand answers. The better approach is to capture more data, not less.
Surveillance providers have traditionally tried to define risk factors within narrow parameters, hoping to filter out noise. The logic is that fewer alerts will allow analysts to focus on the most essential risks. However, this method has inherent flaws. No system can predict every form of market manipulation. Over-calibration creates blind spots, where unusual but potentially problematic behaviors go undetected. Analysts spend too much time refining thresholds, focusing more on eliminating false positives than identifying potential issues. While fewer alerts might sound appealing, it’s meaningless if key risks are missed.
Effective calibration can be achieved by using two key elements: statistical analysis and automation.
Statistics revolve around the securities or products being monitored. Dynamic thresholds, based on historical data like average trade sizes and historical volatility, allow for more precise alerts across various asset classes. This ensures that the right amount of alerts is generated without overwhelming analysts.
Automation is equally crucial in improving both efficiency and consistency. By capturing a broad set of alerts, Validus enables firms to automate the resolution process according to their internal supervisory procedures. The system evaluates factors such as trade volume, prior alerts, or specific account types and can automatically close alerts when appropriate. Each step is documented, providing a complete audit trail.
Not only does this ensure consistency across the team but it also allows firms to identify patterns in near misses. Accounts frequently triggering alerts but falling just below thresholds can signal the need for recalibration, ensuring that the system remains effective. It also allows clients to retain a comprehensive set of alerts for trend analysis, providing valuable insights into potential risks in the “low value alerts”. If closed out once as not significant, maybe the 20th time the account is triggered, it becomes significant.
Focusing solely on minimizing false positives can create a dangerous illusion of security. Firms need to think differently, embracing solutions that capture more data and provide deeper insights. Rather than simply reducing alerts, cast a wide net and shine a light on your trading.
To learn more or book a demo of Validus click here.