State Street deployed continuous learning neural networks on its Alpha Data Platform to automatically detect anomalies in financial market data, flagging suspicious records for human review.

The Alpha Data Platform incorporates neural networks trained across multiple data domains — including reference data, pricing, and corporate actions — to detect anomalies. According to State Street, this approach reduced false positive data exception alerts dramatically compared to traditional rules-based validation: one cited comparison found that rules-based methods flagged over 31,000 exceptions in six months (of which only 250 were genuine), while the neural network identified 4,000 exceptions and still caught 100% of true exceptions. The system escalates suspicious records to human analysts for validation and remediation.