Enhancing Transparency in Portfolio Optimization through Explainable Evolutionary Optimization Techniques

The growing complexity of financial markets and the increasing adoption of algorithmic decision-making highlight the need for optimization models that are not only accurate but also transparent and trustworthy. Traditional portfolio optimization approaches—such as mean–variance analysis or machine-learning-based allocation models—often operate as black boxes, limiting stakeholders’ ability to understand the rationale behind investment decisions. This study proposes an explainable evolutionary optimization framework designed to improve transparency in portfolio construction while maintaining competitive performance. Evolutionary algorithms (EAs) are utilized to explore diverse allocation strategies, while explainable artificial intelligence (XAI) techniques—such as SHAP-based feature contribution analysis and surrogate-rule extraction—are integrated to provide interpretable insights into model behavior. The proposed approach is evaluated using publicly available financial time-series data from Yahoo Finance and Google Finance, enabling reproducible experimentation across multiple market conditions. Preliminary analysis indicates that combining evolutionary search with explainability components has the potential to support more trustworthy financial decision-making by offering a clear understanding of risk–return trade-offs, influential asset drivers, and model stability across market regimes. This study provides a foundational step toward developing a fully transparent financial decision-support system based on explainable evolutionary optimization.