Forecasting the XBANK Index in Türkiye Using Macroeconomic Indicators: A Model Comparison with Ensemble Learning Methods

The objective of this study is to predict the monthly closing prices of the BIST Bank Index (XBANK) utilising macroeconomic and financial indicators. The explanatory variables encompass the real exchange rate, inflation, the consumer confidence index, the policy rate of the Central Bank of the Republic of Türkiye (CBRT), the growth rate of M2 money supply, CBRT reserves, deposits, the industrial production index, the Türkiye CDS spread, and the VIX fear index. In the initial evaluation, three machine learning models – GradientBoosting, XGBoost, and RandomForest Regressor – with the highest predictive power were identified using the LazyRegressor method, and hyperparameter optimization was performed on these models. The performance of the models was evaluated using the R² and RMSE criteria. The most successful result was obtained with the GradientBoosting model, which had an R² score of 0.99i Pursuant to feature importance analysis, it was determined that inflation (37%), policy interest rate (29%), and Central Bank of the Republic of Türkiye (CBRT) reserves (13%) were the variables exerting the most influence on the movements of the banking index. The findings of this study suggest that monetary policy and macroeconomic stability exert a significant influence on the stock performance of the Turkish banking sector.