Integrating LSTM and CNN for Stock Market Prediction: A Dynamic Machine Learning Approach

Sujata  Joshi; Bangaru Lakshmi Mahanthi; Pavithra G; Kiran Sree Pokkuluri; Swapnil S. Ninawe; Rani Sahu

doi:10.37965/jait.2025.0652

Integrating LSTM and CNN for Stock Market Prediction: A Dynamic Machine Learning Approach

Authors

Sujata Joshi Department of Computer Science and Engineering, Nitte Meenakshi Institute of Technology, Bangalore, India https://orcid.org/0000-0002-1889-0326
Bangaru Lakshmi Mahanthi Department of Computer Science, GST, GITAM (Deemed to be University), Visakhapatnam., India https://orcid.org/0009-0005-6872-8385
Pavithra G Dayananda Sagar College of engineering, Bengaluru, Karnataka, India https://orcid.org/0000-0002-8938-6303
Kiran Sree Pokkuluri Department of Computer Science and Engineering, Shri Vishnu Engineering College for Women, Bhimavaram, India https://orcid.org/0000-0001-8601-4304
Swapnil S. Ninawe Department of Electronics and Communication Engineering, Dayananda Sagar College of Engineering, Bengaluru, Karnataka, India https://orcid.org/0000-0002-6573-334X
Rani Sahu Department of Computer Science and Engineering, IES College of Technology, Bhopal, Madhya Pradesh, 462026, India https://orcid.org/0000-0001-7844-2085

DOI:

https://doi.org/10.37965/jait.2025.0652

Keywords:

predictive analytics, stock market, LSTM, CNN, Hybrid LSTM-CNN, machine learning

Abstract

This paper investigates the application of machine learning models, including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and a hybrid LSTM-CNN model, for predicting stock prices of companies listed on the National Stock Exchange (NSE). The proposed hybrid model leverages CNN’s capability for spatial feature extraction and LSTM’s proficiency in modelling temporal dependencies, effectively addressing the complex and volatile nature of stock price movements. Using a comprehensive dataset of historical stock prices and trading volumes from various sectors, the hybrid model achieved superior performance with a 15% improvement in RMSE compared to standalone CNN and LSTM models. Results demonstrate its robustness, particularly in volatile market conditions, showcasing its potential for accurate and reliable predictions. This study contributes a novel hybrid approach that integrates spatial and temporal learning to enhance stock market prediction, offering valuable insights for investors and financial analysts while providing a scalable framework for broader financial applications.