Abstract—Accurately predicting stock prices is a long-standing challenge due to the volatility and nonlinear structure of financial markets. To address this, the present study conducts a comprehensive evaluation of Deep Learning (DL) architectures for univariate stock price forecasting within the Saudi telecommunications sector. Four DL models—Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), and a hybrid CNN-LSTM—are benchmarked against the Naïve, SMA-5, and Autoregressive Integrated Moving Average (ARIMA) models using five years of daily closing prices for Saudi Telecom Company (STC), Mobily, and Zain. All models performances were assessed using Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) on inverse-scaled predictions. The findings clearly show that DL models capture complex temporal dependencies more effectively than traditional baselines. LSTM consistently achieved the lowest test RMSE across all datasets (Mobily: 1.169705; STC: 0.708495; Zain: 0.271470), confirming its superior predictive capability. CNN and CNN-LSTM delivered competitive accuracy, while RNN exhibited the weakest performance. Overall, the results demonstrate the strong potential of deep learning—particularly LSTM—for improving short-term stock price prediction and advancing data-driven financial analytics in the Saudi stock market.
 
Keywords—Deep Learning (DL), stock market, prediction, models, regression, time series


Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).