Paper Title: "Apply Multivariate Time Series Approaches for Forecasting Vietnam Index 30"
Paper Link: Springer Link
Student Contributors:
- Nguyen Thanh Noi – 19521979 – Information Systems 2019: Lead author.
- Vo Quang Huy – 19521640 – Information Systems 2019: Co-author.
- Nguyen Minh Nhut – 17520867 – Information Systems 2017: Co-author.
Supervisor: Assoc. Prof. Dr. Nguyen Dinh Thuan
Paper Summary:
The stock market is an attractive channel for many investment funds. The stock indices of several companies with the largest capitalization are essential indicators for the economic situation. The VN 30 index in Vietnam is calculated from the 30 companies with the largest capitalization and liquidity. Forecasting for these market indices is always a significant challenge. According to recent research, many prediction models for a variable, including LSTM and GRU, have been proposed to achieve good performance in analyzing time trends. However, stock indices are influenced by many factors that a univariate time series model cannot represent enough information. Therefore, this paper researches and applies multivariate time series with various methods from statistical VAR and machine learning regression to deep learning methods such as the LSTNet model combining CNN and LSTM, MTGNN applying graph neural networks, and DSTP using attention mechanism to weight dependencies for predicting the VN30 Index with multivariate time series data of the most significant and influential stocks on the current Vietnam stock market. The results of the study show that the LSTNet model achieves the best forecast among the experimented multivariate models with a 1.12% MAPE and outperforms univariate models.
"We would like to express our gratitude to Assoc. Prof. Dr. Nguyen Dinh Thuan – Lecturer of the Information Systems Department for dedicated guidance, identifying limitations, and providing possible improvements, assisting the team during the research and publication of this paper."
The 10th International Conference on Future Data and Security Engineering (FDSE) is a leading forum designed for researchers and students interested in advanced and practical activities related to data, information, knowledge, and security techniques to explore advanced ideas, present and exchange research results, and discuss new issues and future directions for research and development in data, information, knowledge, and security techniques. At FDSE, researchers and students can share research solutions for contemporary data and security technical and societal issues, identify new issues and directions for future research and development.
FDSE Proceedings have been indexed in Scopus, EI Compendex, DBLP, and listed in Thomson Reuters' Conference Proceeding Citation Index (CPCI).
For more details, please visit: UIT Fanpage Facebook Post
Written by: Hai Bang
Translated by: Ngoc Diem
