Evaluating AI Disclosure Practices and Regulatory Gaps in Quantitative Finance

Abstract

As artificial intelligence becomes increasingly prominent within financial markets, regulators lack the frameworks necessary to assess whether AI-driven investment strategies comply with fiduciary standards, manage systemic risk responsibly, or produce returns through legitimate means. This gap is particularly prevalent among leading quantitative trading and investment firms, which deploy sophisticated machine learning systems while simultaneously disclosing little about how those systems are built or monitored. This paper analyzes the publicly available AI disclosure practices and performance of five major quantitative trading and investment firms (Renaissance Technologies, Two Sigma, D. E. Shaw, Citadel, and Jane Street) from Q1 2023 to Q1 2026, comparing returns from their publicly disclosed top-20 equity holdings against the S&P 500 using data from the website HedgeFollow. The findings reveal that all five firms score below 40 out of 100 on a disclosure rubric created for this study, with the largest gaps appearing in model governance, retraining procedures, risk limits, stress-response behavior, and incident-reporting protocols. At the same time, three firms (Renaissance Technologies, D. E. Shaw, and Jane Street) display peak cumulative return gaps ranging from approximately 40 to over 100 percentage points above the S&P 500 during the study period, demonstrating that strong performance can coexist with limited public visibility into the systems and controls behind that performance. The paper concludes with four targeted regulatory proposals: mandatory AI disclosure standards, explainability requirements, systemic risk monitoring, and human oversight attribution rules, all of which must be implemented while preserving market innovation.

Read full paper (PDF) →