Glossary term
False Positive
Subject: Risk Management, Risk Assessment and Quantification
Also known as: False alarm, Type I error, False hit
A false positive is an alert raised by a test, model or control when nothing is actually wrong, which costs reviewers time and can delay legitimate activity.
Explanation
Detection systems trade between two mistakes. A false positive is a wrong alarm: a spam filter binning a genuine email, a screening tool matching a customer to a sanctioned person with a similar name, or a fraud model blocking a real purchase. Each one needs an analyst to look and clear it. The false positive rate is the share of alerts, or of non-problem cases, that turn out to be wrong. In compliance monitoring it is common for the large majority of alerts to be false, which drains staff and can hide real cases. Tuning thresholds cuts the noise but risks missing true cases, the false negatives. Left unchecked, the volume swamps analysts and slows customers, and tired reviewers begin clearing alerts too quickly.
Example
Each week a wealth manager's name-screening tool produces 1,000 alerts. Analysts clear 980 as different people with similar names, leaving 20 needing escalation. The false positive share is 98%. The team adds date of birth and country as matching fields, and the next week's alerts fall to 400.
Not to be confused with
- False Negative
- A false negative is a missed real problem. A false positive is an alert on something harmless.