Abstract
<jats:p>This chapter explores the potential of advanced deep learning for financial forecasting in emerging economies. It develops a hybrid AI framework combining CNN, LSTM, and an attention mechanism to predict stock market volatility in the GCC, focusing on the Saudi Tadawul All Share Index (TASI). Using real-time data from TASI, Brent crude, VIX, S&P 500, and gold, the model forecasts short-term (1-day) and medium-term (5-day) volatility. CNN extracts features, LSTM captures temporal dependencies, and attention highlights critical time steps. Experiments show superior performance over ARIMA, SVR, and Persistence, with MAE of 0.0171 (1-day) and 0.0186 (5-day), improving 81–83%. The model is robust to shocks, adapts to GCC markets, and supports risk management and investment decisions. Ethical considerations include transparency, interpretability, and data reliability.</jats:p>