Investigating inherently interpretable machine learning models for stock return forecasting and dynamic portfolio rebalancing, comparing predictive performance, risk-adjusted returns, and explainability against black-box approaches.