View in EDS

Integrating Technical, Temporal, and Sentiment Cues: A Hybrid Deep Learning Approach Optimized by WWO for Stock Market Risk Prediction.

Saved in:
Bibliographic Details
Title: Integrating Technical, Temporal, and Sentiment Cues: A Hybrid Deep Learning Approach Optimized by WWO for Stock Market Risk Prediction.
Authors: Veena, D., G. L., Prakash, S., Savitha, S., Roselin Mary, Kishore, D., Valluri, Prasanthi
Source: International Journal of Intelligent Engineering & Systems; 2025, Vol. 18 Issue 10, p901-915, 15p
Subject Terms: RISK assessment, DEEP learning, MATHEMATICAL optimization, SENTIMENT analysis, PREDICTION models, ECONOMIC indicators, FINANCIAL databases, STOCKS (Finance)
Abstract: Numerous economic, political, and psychological variables impact the ever-changing stock market. The non-linear and stochastic character of financial data has made risk prediction a formidable obstacle for quite some time. In order to better anticipate stock market risks, this study presents a new hybrid deep learning-based model that combines Vision Transformer (ViT), Bidirectional Long Short-Term Memory (BiLSTM), besides Convolutional Neural Networks (CNN). Utilizing the Water Wave Optimization (WWO) technique allows for the efficient optimization of hyper parameters and model weights. The suggested approach captures geographical representations besides temporal connections by utilizing technical indicators, sentiment investigation from news sources, and historical stock data. The following benchmark models are used for comparative analysis: ViT, baseline BiLSTM, LeNet, AlexNet, GoogLeNet, VGGNet, ResNet, and CNN. In addition, the optimal performance of the model is contrasted with that of several other algorithms, including Tree-Seed Algorithm (TSA), Ant Colony Optimization (ACO), Crow Search Optimization (CRO), and Black Widow Optimization (BWO). Various metrics for evaluation are calculated, including precision, score, Receiver Operating Characteristic (ROC)-Area Under the Curve (AUC), and loss values. When compared to other models, the suggested one performs better experimentally in terms of stability, reduced prediction errors, and generalizability. The proposed model achieved 98.7% accuracy, 97.9% precision, and 98.2% recall on the benchmark dataset, demonstrating superior performance. As a result of this study, investors and policymakers would have better resources at their disposal to make informed decisions in the financial sector, thanks to the advancements made in intelligent risk management systems. Timely interventions and portfolio modifications are made possible by the system's promising real-time risk assessment. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Intelligent Engineering & Systems is the property of Intelligent Networks & Systems Society and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:Numerous economic, political, and psychological variables impact the ever-changing stock market. The non-linear and stochastic character of financial data has made risk prediction a formidable obstacle for quite some time. In order to better anticipate stock market risks, this study presents a new hybrid deep learning-based model that combines Vision Transformer (ViT), Bidirectional Long Short-Term Memory (BiLSTM), besides Convolutional Neural Networks (CNN). Utilizing the Water Wave Optimization (WWO) technique allows for the efficient optimization of hyper parameters and model weights. The suggested approach captures geographical representations besides temporal connections by utilizing technical indicators, sentiment investigation from news sources, and historical stock data. The following benchmark models are used for comparative analysis: ViT, baseline BiLSTM, LeNet, AlexNet, GoogLeNet, VGGNet, ResNet, and CNN. In addition, the optimal performance of the model is contrasted with that of several other algorithms, including Tree-Seed Algorithm (TSA), Ant Colony Optimization (ACO), Crow Search Optimization (CRO), and Black Widow Optimization (BWO). Various metrics for evaluation are calculated, including precision, score, Receiver Operating Characteristic (ROC)-Area Under the Curve (AUC), and loss values. When compared to other models, the suggested one performs better experimentally in terms of stability, reduced prediction errors, and generalizability. The proposed model achieved 98.7% accuracy, 97.9% precision, and 98.2% recall on the benchmark dataset, demonstrating superior performance. As a result of this study, investors and policymakers would have better resources at their disposal to make informed decisions in the financial sector, thanks to the advancements made in intelligent risk management systems. Timely interventions and portfolio modifications are made possible by the system's promising real-time risk assessment. [ABSTRACT FROM AUTHOR]
ISSN:2185310X
DOI:10.22266/ijies2025.1130.57