How to manage a multifactor-driven crude oil market more effectively? A revisit based on the multiple criteria perspective

Abstract

With the complexity of the international crude oil market deepening, it is of profound significance for information recipients and actors to clarify the operation mechanism of the multi-factor-driven international crude oil market, grasp its crucial drivers, and establish a reasonable and effective analysis and early warning system. This paper reviewed the results of existing studies and proposed a collection of daily frequency proxies from five perspectives based on the principles of plurality, completeness and rationality. What's more, a three-dimensional assessment strategy was also developed based on causality, predictability, and necessity, which complemented and extended existing methodologies and findings on crude oil market drivers and theoretically quantified the economic and statistical properties of various proxies. After that, the most critical agents were extracted by relying on the logic of multi-criteria decision-making, which solved the problem of scattered attention in the analysis of the crude oil market. Finally, based on machine learning and artificial intelligence, a hybrid forecasting model that blended key driving agents with both error accuracy, as well as directional accuracy, was constructed. Taking Brent crude oil, the benchmark for more than two-thirds of the world's crude oil, as an example, the findings verified the importance and necessity of correctly grasping the key drivers and confirmed that this study can provide a more scientifically sound research methodology and theoretical basis for crude oil market analysis and early warning based on limited attention.