An Autonomous Control Framework For Data Quality Governance In Trading And Market Risk Monitoring
Abstract
Financial institutions increasingly depend on large, fast-moving volumes of reference, pricing, transaction, and valuation data to calculate market risk exposures, including Value-at-Risk (VaR), sensitivities, stress measures, and regulatory capital. Conventional data quality practice is often organized around reactive detection: controls identify exceptions after data have entered a processing chain, while remediation depends on manual investigation. This paper develops the Autonomous Control Framework (ACF), a conceptual maturity model describing progression from manual detection to rule-based monitoring, automated alerting, predictive anomaly detection, and bounded autonomous remediation at the level of Critical Data Elements (CDEs). The framework is grounded in BCBS 239 risk data aggregation principles, data quality and governance literature, and model risk management considerations. A key distinction is made between autonomous control execution and autonomous judgment: Stage 5 permits only pre-authorized, reversible, auditable actions such as quarantine or use of a documented fallback source, and does not delegate discretionary risk judgment to a machine. The paper is conceptual and illustrative rather than empirically validated. Its principal contribution is a structured bridge between CDE governance, detection latency, market-risk data propagation, and bounded automation, supported by explicit maturity dimensions, an autonomy-eligibility matrix, and testable research propositions. Future research should validate the framework using institutional data, controlled experiments, and expert-based assessment of maturity and autonomy eligibility.