We congratulate Rayan Ayari for his paper published in Quantitative Finance
We congratulate Rayan Ayari for his paper published in Quantitative Finance
Title: Feature Configuration Effects in DRL Portfolio Management: A Risk-Focused Evaluation under Market Stress
This paper investigates whether a DRL-based portfolio manager can leverage BARRA model factor returns while incorporating transaction costs and liquidity constraints. We integrate the TD3 algorithm with BARRA factors and the Differential Expected Shortfall (DES) reward function to create a framework that adapts to market conditions, prioritizing returns while minimizing downside risk. Using Monte Carlo sampling, we construct 1,000 portfolios of 30 S&P 500 stocks across four quarters of 2022 to examine the impact of different feature sets—from BARRA factors and technical indicators to no additional features—on portfolio performance, where the no feature configuration provides the benchmark. Statistical significance is assessed using a paired permutation test with false discovery rate control. Results show that incorporating BARRA features enhances downside protection and tail risk, improving Maximum Drawdown (MDD) by 0.71% (p = 0.02) and Conditional Value at Risk at the 95% level (CVaR) by 0.08% (p = 0.09). These findings suggest that feature-informed DRL strategies can provide actionable insights in volatile markets.
Read more at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5595570
Foto: viarami/Pixabay.com