Iren Valova, Natacha Gueorguieva, Thakkar Aayushi , Pulluri Nikitha , Hassan Mohamed

Abstract: Research on deep learning time series regression for stock market predictions and forecasting has grown exponentially in recent years. Considering the complexity of financial time series, combining deep learning with financial market prediction is regarded as a very important topic of research. The experiments in this study are divided into three settings which offer different topologies. For our first experimental setting we create hybrid sequential nonlinear models. For our second experimental setting we propose a new deep learning topology based on Attention CNN_BiLSTM for pretraining and light gradient boosting machine (LGBM) as a regressor (Attn_CNN_BiLSTM_LGBM). With our third experimental setting we propose a new model based on weighted average ensemble and the following four topologies: Attn_CNN_BiLSTM_LGBM, and attention_ CNN_ BiLSTM with various regressors. The ensemble model (EM) allows the contribution of each ensemble member to the prediction to be weighted proportionally to the performance of the EM. The latter achieves significantly better performance in predicting stock market.

Keywords: stock market prediction, convolutional neural networks, long-short term memory, gated recurrent unit

Date Published: October 10, 2023
DOI: 10.11159/jmids.2023.002

View Article