Multi-sequential Neural Network Models for Stock Price Forecasting

Sheba Elizabeth Thomas; Rubell Marion Lincy George; Nevin Selby; Aditya Taparia; Jobish Vallikavungal

doi:10.47852/bonviewJCCE52025731

Authors

Sheba Elizabeth Thomas Indian Institute of Information Technology Kottayam, India https://orcid.org/0000-0002-0854-7102
Rubell Marion Lincy George Indian Institute of Information Technology Kottayam, India https://orcid.org/0000-0003-4394-248X
Nevin Selby Indian Institute of Information Technology Kottayam, India https://orcid.org/0009-0009-2102-6094
Aditya Taparia Indian Institute of Information Technology Kottayam, India https://orcid.org/0009-0004-4603-0085
Jobish Vallikavungal School of Engineering and Sciences, Tecnologico de Monterrey, Mexico https://orcid.org/0000-0003-4238-5829

DOI:

https://doi.org/10.47852/bonviewJCCE52025731

Keywords:

stock price forecasting, long short-term memory, financial market, investment decision-making, deep learning, time series

Abstract

Deep learning techniques are transforming stock market forecasting by significantly improving the accuracy of predicting price movements and market patterns. In this research work, we propose two novel hybrid architectures – MSLSTM (Multivariate Sequential Long Short-Term Memory) and MSLSTMA (Multivariate Sequential Long Short-Term Memory Autoencoder). Both models leverage the Long Short-Term Memory (LSTM) ability to capture complex temporal dependencies in sequential financial data. Our method performs well in a variety of industries and surpasses a few other LSTMs and their autoencoder-augmented variations, as well as conventional deep learning models like CNNLSTM. MSLSTMA achieves the highest sector-wise winning rates, notably 95.83% in the telecommunication sector, 87.5% in technology, and 83.33% in industrials. Model performance was rigorously evaluated using mean squared error, root mean squared error, mean absolute error, and mean absolute percentage error. Across all these metrics, MSLSTMA consistently delivered the lowest error rates, showcasing its superior accuracy and practical relevance for real-world financial forecasting. MSLSTM also demonstrated strong performance, with winning rates of 50% in consumer staples and 45.83% in the healthcare sector. This research introduces an effective and scalable tool for investors, with the potential to enhance investment decisions through precise stock price prediction.

Received: 18 March 2025 | Revised: 17 June 2025 | Accepted: 11 July 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Author Contribution Statement

Sheba Elizabeth Thomas: Methodology, Validation, Investigation, Data curation, Writing – review & editing, Visualization, Supervision. Rubell Marion Lincy George: Conceptualization, Validation, Resources, Writing – original draft, Project administration. Nevin Selby: Conceptualization, Software, Formal analysis, Writing – original draft, Project administration. Aditya Taparia: Software, Formal analysis, Supervision. Jobish Vallikavungal: Methodology, Validation, Investigation, Resources, Data curation, Writing – review & editing, Visualization, Supervision.