A Hybrid Convolutional Recurrent (CNN-GRU) Model for Stock Price Prediction

Abstract

Stock price forecasting systems are on-demand that used for prediction in the financial world. The deep learning models are used for handling large data and making predictions more accurately than the traditional methods. This paper proposed a hybrid convolutional recurrent model architecture for stock price prediction based on exploitation of two different deep learning models properties into a single model. Hybrid Convolutional Recurrent Model consists of 1D-CNN and GRU where 1D-CNN is leveraged as feature extraction task and GRU for temporal regression. The proposed CNN-GRU Model outperform for stock price prediction. The performance evaluation and comparison analysis between the existing hybrid models and proposed hybrid model have also been done with experiments to justify.