Forecasting of Bahrain stock market with deep learning: Methodology and case study

Deep learning methodology may shape the future of technology and data science. Recent trends in machine learning reveal a growing interest in deep learning to a wide range of complicated problems that were infeasible using traditional techniques. In this paper, we present a new deep learning predictive model for stock market forecasting. The historical data of Bahrain Bourse all share index is employed to perform the experiments. The proposed forecasting model is developed using a combination of LSTM autoencoder and stacked LSTM network. We also augment the input features with two of the most common technical indicators in order to improve forecasting performance. The developed model is compared with traditional LSTM and shallow MLP networks. Experiments show superiority of the proposed method over the other two evaluated models in terms of the error measures employed in this study.