Forecasting of Share Prices Based on Hybrid Model of CNN and LSTM: A Multi-Factor Approach

Jialin Tian

2024

Abstract

As a matter of fact, hybrid models are widely adopted in pricing models contemporarily. This study uses a hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model to investigate price prediction for stocks, taking macroeconomic and technical indicator effects into account. The dataset includes 40 medical device firms' daily stock values from January 2022 to August 2024. Upon completing training and optimization, the model demonstrated its ability to accurately identify brief variations in stock prices and forecast the closing price of the stock market for the upcoming day by utilizing five days' worth of pertinent data. The model's Mean Absolute Percentage Error (MAPE) is 6.53%, its Root Mean Square Error (RMSE) is 4.95, and its Adjusted R² is 0.9968. However, when the model encounters an extremely high range of stock prices (500-700), its accuracy decreases as the true stock price increases. Future research may aim to enhance stock price prediction especially when experiencing very high share prices and expand the dataset in multiple sections to improve generalizability. This research investigates the utility of CNN-LSTM models in guiding investors to more accurately predict stocks in the healthcare sector (after epidemic), providing relevant references for future research.

Paper Citation

in Harvard Style

Tian J. (2024). Forecasting of Share Prices Based on Hybrid Model of CNN and LSTM: A Multi-Factor Approach. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 222-228. DOI: 10.5220/0013213700004568

in Bibtex Style

@conference{ecai24,
author={Jialin Tian},
title={Forecasting of Share Prices Based on Hybrid Model of CNN and LSTM: A Multi-Factor Approach},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={222-228},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013213700004568},
isbn={978-989-758-726-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Forecasting of Share Prices Based on Hybrid Model of CNN and LSTM: A Multi-Factor Approach
SN - 978-989-758-726-9
AU - Tian J.
PY - 2024
SP - 222
EP - 228
DO - 10.5220/0013213700004568
PB - SciTePress