@article{article_1813752, title={The Role of Financial Markets in Predicting BIST Sustainability Index Performance: New Evidence from Hybrid Machine Learning Models}, journal={Ekonomi Politika ve Finans Araştırmaları Dergisi}, volume={10}, pages={383–402}, year={2025}, DOI={10.30784/epfad.1813752}, author={Çolak, Zeynep}, keywords={Sürdürülebilir Finans, BIST Sürdürülebilir Endeksi, Makine Öğrenimi, SHAP, Açıklanabilir YapayZeka}, abstract={The increasing importance of sustainable finance makes it critical to understand and accurately model the performance dynamics of investment instruments in this area. This study aims to forecast the return of the BIST Sustainability Index using financial market indicators and to explain the underlying dynamics of this forecasting process, thereby understanding the complex structures of financial markets, investor behavior, and information flow. In this study, eleven different machine learning models were compared with a validation strategy suitable for the time series structure, and the most successful candidates were subjected to hyperparameter optimization. In order to overcome the limitations of single models, a sequential hybrid model based on the Residual Fitting approach was developed. According to the results of the study, the two-stage hybrid model, which uses the Voting Regressor as the main predictor and Random Forest as the error corrector, provided the lowest error (RMSE) and the highest R² value. The findings indicate that the BIST_100 index is the most critical determinant, while risk aversion indicators such as Gold, USD, and VIX have a negative effect. This evidence has far-reaching implications for understanding the dynamic relationships between the Sustainability Index and macroeconomic variables.}, number={Özel Sayı}, publisher={Ekonomi ve Finansal Araştırmalar Derneği}