Title: Optimised hybrid CNN bi-LSTM model for stock price forecasting

Authors: Deepti Patnaik; N.V. Jagannadha Rao; Brajabandhu Padhiari; Srikanta Patnaik

Addresses: GIET-University, Gunupur, Rayagada, Odisha, 765 022, India ' GIET-University, Gunupur, Rayagada, Odisha, 765 022, India ' IIMT Bhubaneswar, Khurda, Odisha, 751 003, India ' IIMT Bhubaneswar, Khurda, Odisha, 751 003, India

Abstract: Financial markets are considered the backbone of a country's economy. This article focuses on the stock price forecasting using deep learning models. Here, a hybrid model, i.e., convolutional neural network, bidirectional long short-term memory network has been proposed and its parameters are optimised by self-adaptive multi-population elitist JAYA algorithm. Stock prices of more than 13 years of various challenging stock exchanges of the globe such as: Standard & Poor 500, NIFTY 50, Nikkei 225, Dow Jones are used here for analysis purposes. The performance parameters such as root mean square error, mean absolute percentage error and mean absolute error are used for analysing the model. The proposed hybrid model is also compared with state-of-art models and it is found that this proposed model out performs the existing models.

Keywords: forecasting; convolutional neural network; bidirectional long short-term memory; LSTM; hybrid model; evolutionary computation; SAMPE Jaya algorithm; RMSE; MAE.

DOI: 10.1504/IJIE.2024.139747

International Journal of Intelligent Enterprise, 2024 Vol.11 No.3, pp.248 - 273

Received: 02 Sep 2023
Accepted: 25 Oct 2023
Published online: 05 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article