A novel prediction model to evaluate the dynamic interrelationship between gold and crude oil

Pandit, Sarth; Luo, Xiaojun

Abstract

Global events, such as the pandemic and European conflicts, have caused significant inflation and high volatility in gold and crude oil prices. This has garnered substantial international attention while banks, governments, and businesses are devoted to developing strategies to hedge against the potential impacts of economic uncertainties. Focused on addressing economic uncertainties, this study delves into the crucial role of the relationship between gold and crude oil in shaping global financial and economic dynamics. The primary objective of this study is to conduct a comprehensive analysis and construct a rolling SARIMAX model for predicting the rolling 12-months correlation of the Gold–WTI and Gold–Brent relationship. Monthly data on Gold spot prices, WTI futures, and Brent futures is collected from May 1983 to December 2022. Three sophisticated data analysis techniques, the Rolling Correlation method, the SARIMAX model, and the Rolling Model are integrated to develop the Rolling SARIMAX module. R-square values of this newly developed model achieved 89.8% and 88.4% for predicting the rolling correlation for Gold Spot Price-WTI Futures and Gold Spot Price-Brent Futures, respectively, while the mean absolute percentage error was 10.33% and 10.84%, respectively. The higher accuracy in correlation prediction between gold and crude oil prices can present critical insights for risk management, economic planning, strategic investment, economic cycles, and global economic outlook. This newly developed prediction model adeptly handles both linear and non-linear relationships while adapting to external variables in dynamic market scenarios. Its judicious balance between complexity and practicality positions it as a sophisticated analytical tool with real-world applicability, setting a new benchmark in financial market analysis.

• Journal Article Type: Article
• Acceptance Date: Feb 7, 2024
• Online Publication Date: Mar 16, 2024
• Deposit Date: Mar 20, 2024
• Publicly Available Date: Mar 21, 2024
• Journal: International Journal of Data Science and Analytics
• Print ISSN: 2364-415X
• Electronic ISSN: 2364-4168
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• DOI: https://doi.org/10.1007/s41060-024-00519-8
• Keywords: Applied Mathematics; Computational Theory and Mathematics; Computer Science Applications; Modeling and Simulation; Information Systems
• Public URL: https://uwe-repository.worktribe.com/output/11833950
• Additional Information: Received: 9 March 2023; Accepted: 7 February 2024; First Online: 16 March 2024