Abstract Accurate stock price prediction is pivotal for optimal financial decision-making, yet it remains challenging due to the non-linear and dynamic nature of financial markets. Traditional machine learning models often operate in silos, relying exclusively on either fundamental analysis (assessing a company’s financial health) or technical indicators (analyzing historical price patterns). This fragmented approach fails to capture the complete spectrum of factors driving market movements, limiting predictive accuracy. This study bridges this gap by proposing a hybrid machine learning framework that integrates fundamental and technical analysis to enhance forecasting accuracy in the Egyptian stock market using a Long Short-Term Memory (LSTM) network. The framework was applied to Telecom Egypt (ETEL.CA) data from January 2015 to December 2023, three configurations were evaluated: fundamental-only (e.g., financial ratios), technical-only (e.g., moving averages, RSI), and hybrid datasets. Framework performance was measured using Coefficient of Determination (R 2 ), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Results indicate that the fundamental-based LSTM model provides the most robust and generalizable performance. Technical indicators yield strong short-term results but exhibit overfitting, while the hybrid approach produces mixed outcomes with weaker generalization. These findings suggest that feature stability and relevance are more critical than simply integrating diverse data sources. The study demonstrates that combining deep learning with fundamental indicators offers a reliable approach for stock price prediction in emerging markets, providing valuable insights for investors and analysts.
Abd-Elwahab et al. (Tue,) studied this question.