Dynamic risk resonance between crude oil and stock market by econophysics and machine learning

Abstract

Changes in crude oil prices have a strong impact on global stock markets. As the largest importer of crude oil, China’s international crude oil market and stock market have a strong risk of contagion. It is of great significance to study the stochastic resonance effect between crude oil market and Chinese stock market. This paper proposes the dynamic risk resonance analysis method based on econophysics and machine learning. From the perspective of stochastic resonance in nonequilibrium statistical physics, we set up a new periodic capital asset pricing model based on econophysics and asset pricing theory, and introduce signal power amplification to describe the risk resonance effect of crude oil price and Chinese stock market. Then, we analyze the dynamic risk resonance between crude oil and stock market impacted by systemic risk and investor heterogeneity. The results demonstrate that: (1) Based on machine learning and predictive testing, the in-sample and out-of-sample forecasting results show that the proposed new periodic model has better fitting performance than the benchmark model; (2) There are optimal system parameters and period information to minimize the risk resonance effect between two markets; (3) The existence of optimal investor heterogeneity leads to inverse resonance behavior between the two markets, which greatly reduces the contagion of market risks.