Modeling Saudi stock index returns and volatility: a dual approach using GARCH and neural networks

The financial markets are the drivers of economic growth as they organize savings, bring in foreign investment, and they efficiently allocate resources. The Tadawul is the largest stock market in the GCC, which is highly impacted by prices of oil and gold. Correct forecasting of market returns and volatility is thus a key to investors and policymakers. The research analysis is performed on the Tadawul All Share Index (TASI) between January 1, 2000, and December 31, 2022, with the help of traditional GARCH-family models (GARCH, EGARCH, GJR-GARCH, and MGARCH) and a Long Short-Term Memory (LSTM) neural network. Explanatory variables include oil and gold returns, MSE, MAE, RMSE, R 2, and Diebold-Maasoony test are used to assess predictive performance. The results show that the LSTM model is the most effective model that captures nonlinear volatility patterns, whereas GARCH models, especially the GJR-GARCH model with GED distribution offers better returns projections. The findings also confirm that oil and gold returns have a significant influence on the performance of TASI, which proves their role in the oil-dependent economy. Altogether, the evidence demonstrates the synergistic advantages of both econometric and machine learning methods and provides useful implications of risk management and investment decision-making as well as policy guidance.