Metaheuristic-optimized ANFIS and ANN models for stock price forecasting: evidence from the Borsa Istanbul 100 index

As financial markets worldwide continue to evolve, the importance of predictive models for investors and analysts is growing. This paper presents a comparative analysis of metaheuristic-based Adaptive Neuro-Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN) models for predicting stock prices in the Borsa Istanbul 100 (BIST 100) index. Using a comprehensive dataset of weekly data (2010–2023), researchers tested different model configurations, and the ANFIS model was improved with Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO). Model performance was evaluated using performance criteria such as Mean Squared Error (MSE), Mean Absolute Error (MAE), and R-squared (R2). The results showed that, in particular, the ANFIS-PSO and ANN-based approaches exhibited higher accuracy compared to traditional regression methods. The study also analyzed the impact of various macroeconomic and financial indicators, including the gold/TL exchange rate, VIX index, and government bond interest rates, on the BIST 100. The findings reveal that artificial intelligence-based modeling significantly improves stock price forecasts thanks to its adaptability to dynamic market conditions. In this respect, the study makes theoretical contributions and highlights practical implications for investment strategies and risk management. Future research can further enrich the financial modeling literature by testing the validity of similar approaches in other emerging or developed markets, as well as by incorporating additional macroeconomic factors and different meta-heuristic algorithms into the model.