IPH2O: Island Parallel-Harris Hawks Optimizer-Based CLSTM for Stock Price Movement Prediction

Stock price movement forecasting is the process of predicting the future price of a financial and company stock from chaotic data. In recent years, many financial institutions and academics have shown interest in stock market forecasting. The accurate and successful predictions of the future price o...

Full description

Saved in:
Bibliographic Details
Published in:Annals of data science 2024-12, Vol.11 (6), p.1959-1974
Main Authors: Joel, Linda, Parthasarathy, S., Venkatesan, P., Nandhini, S.
Format: Article
Language:English
Subjects:
Accuracy
Algorithms
Artificial Intelligence
Business and Management
Economics
Error analysis
Finance
Forecasting
Insurance
Management
Noise prediction
Securities markets
Statistics for Business
Stock prices
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Stock price movement forecasting is the process of predicting the future price of a financial and company stock from chaotic data. In recent years, many financial institutions and academics have shown interest in stock market forecasting. The accurate and successful predictions of the future price of stock yield a substantial profit. However, the current approaches are a major challenge due to the dynamic, chaotic, high-noise, non-linear, highly complex, and nonparametric characteristics of stock data. Furthermore, it is not sufficient to consider only the target firms' information because the stock prices of the target firms may be influenced by their related firms. Significant profits can be made by correct forecasting of stock prices, while poor forecasts can cause huge problems. Thus, we propose a novel Island Parallel-Harris Hawks Optimizer (IP-HHO)-optimized Convolutional Long Short Term Memory (ConvLSTM) with an autocorrelation model to predict stock price movement. Then, using the IP-HHO algorithm, the hyperparameters of ConvLSTM are optimized to minimize the Mean Absolute Percentage Error (MAPE). Four different types of financial time series datasets are utilized to validate the performance of the evaluation measures such as root mean square error, MAPE, Index of Agreement, accuracy, and F1 score. The results show that the IP-HHO-optimized ConvLSTM model outperforms others by improving the prediction rate accuracy and effectively minimizing the MAPE rate by 19.62%.
ISSN:2198-5804
2198-5812
DOI:10.1007/s40745-023-00489-x