Affiliations:
Finance Department, Faculty of Economics and Administration, King Abdulaziz University, Jeddah, Saudi Arabia
This study examines the predictive relationship between financial news sentiment and the performance of the Saudi stock market, measured by the Tadawul All Share Index (TASI). A news sentiment index is developed using financial headlines from the Saudi Gazette published between March 2017 and March 2025. The FinBERT model, a natural language processing tool designed for financial text, is used to calculate sentiment scores, which are then averaged on a monthly basis. These sentiment measures are combined with key macroeconomic and market variables, including crude oil prices, interest rates, inflation, exchange rates, and trading volume. For prediction, a hybrid modeling framework is applied, integrating ARIMAX, Random Forest, and XGBoost to capture both linear and nonlinear relationships between TASI returns and sentiment. Model performance is evaluated using root mean squared error (RMSE), mean absolute error (MAE), and the coefficient of determination (R²). The results show that news sentiment and oil price movements have a significant effect on market returns, with important implications for investors, analysts, and policymakers in sentiment-sensitive emerging markets such as Saudi Arabia.
Financial news sentiment, Saudi stock market, Tadawul index, Machine learning models, Market prediction
https://doi.org/10.21833/ijaas.2025.12.024
Alalmai, S. (2025). Prediction of TASI returns using sentiment analysis and hybrid modeling methods: ARIMAX, random forest, and XGBoost. International Journal of Advanced and Applied Sciences, 12(12), 266–279. https://doi.org/10.21833/ijaas.2025.12.024