Stock Price Prediction Using Bi-LSTM and GRU-Based Hybrid Deep Learning Approach

Abstract

AbstractStock market price movement prediction is a critical task for the investors due to its non-stationary and fluctuating nature. So, the automatic price movements forecasting techniques are now the hottest and crucial area for the researcher. Classical statistical models show the poor performance because of the random nature of stock price. In this paper, we proposed a novel hybrid deep learning model employing the bidirectional long short-term memory (Bi-LSTM) and gated recurrent unit (GRU) network. Individually, the long short-term memory (LSTM), Bi-LSTM, GRU, and traditional neural network (NN) modules are implemented to forecast the stock price. Then, the comparison between the individual model’s performances as well as with the proposed hybrid model is done in this work. The proposed stock price prediction model is implemented using the NIFTY-50 stock market data. Our model can predict long time ahead predictions precisely. Experimental results show the proposed hybrid Bi-LSTM-GRU model achieved higher performance than the above-mentioned individual models.KeywordsStock price predictionBi-LSTMGRUHybrid model