Six SQL patterns I use to catch transaction fraud

A practical guide to six SQL patterns for detecting transaction fraud: velocity checks, impossible travel detection, suspicious amount patterns, merchant anomaly clustering, off-hours behavior analysis, and composable window function primitives. Each pattern includes working SQL with explanations of thresholds, false positives, and platform-specific notes (Snowflake, BigQuery, Postgres). The post also covers important caveats around NULL handling, privacy, cost of window functions, and the need for human review in fraud workflows.