Analysis on the Applicability of RNN, LSTM, and GRU Deep Learning Algorithms for Stock Price Prediction

Tianya Xu

2022

Abstract

There are many studies based on deep learning algorithms to predict stock prices. Although the prediction results are good in the experimental environment, the accuracy drops dramatically in the actual stock market. Most scholars want to solve the problem by enhancing the algorithmic model. But the author assesses the applicability between algorithm and stock data as another reason for that problem, and hopes to find out whether there is a matching problem between the algorithm and data by analyzing the prediction result of different types of stock data based on the different algorithms. This paper performs stock price prediction based on RNN, LSTM, and GRU algorithms on four stocks with different fluctuation types and determines the applicability of the three algorithms by analyzing the regression evaluation index of prediction results. The result shows that the fluctuation of stock price has a significant impact on the accuracy of the three algorithms. The LSTM algorithm fits best for the fluctuation type that stock price showing large cyclical fluctuations, whose correlation coefficient reaches at 0.8067, while the GRU algorithm fits best for the fluctuation type that shows slump in stock price, whose correlation coefficient reaches at 0.8072.

Paper Citation

in Harvard Style

Xu T. (2022). Analysis on the Applicability of RNN, LSTM, and GRU Deep Learning Algorithms for Stock Price Prediction. In Proceedings of the International Conference on Big Data Economy and Digital Management - Volume 1: BDEDM, ISBN 978-989-758-593-7, pages 301-305. DOI: 10.5220/0011175000003440

in Bibtex Style

@conference{bdedm22,
author={Tianya Xu},
title={Analysis on the Applicability of RNN, LSTM, and GRU Deep Learning Algorithms for Stock Price Prediction},
booktitle={Proceedings of the International Conference on Big Data Economy and Digital Management - Volume 1: BDEDM,},
year={2022},
pages={301-305},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011175000003440},
isbn={978-989-758-593-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the International Conference on Big Data Economy and Digital Management - Volume 1: BDEDM,
TI - Analysis on the Applicability of RNN, LSTM, and GRU Deep Learning Algorithms for Stock Price Prediction
SN - 978-989-758-593-7
AU - Xu T.
PY - 2022
SP - 301
EP - 305
DO - 10.5220/0011175000003440