Volume 20: Sustainable Energy Solutions for a Post-COVID Recovery towards a Better Future: Part III

Crude Oil Market Intraday Risk Prediction Based on Generalized Heterogeneity Autoregressive and Threshold Kernel Variation Method Zhi-Yi Zheng, Lu-Tao Zhao

Abstract

In recent years, the advancement of trading technology and the acceleration of information transmission have intensified the intraday volatility of the oil market. To identify the volatility and risk of the intraday market accurately and effectively, this paper proposes a method for intraday risk prediction based on generalized heterogeneity autoregressive for high-frequency spot volatility modeling. First, use the threshold kernel variation method to separate jumps and characterize the spot volatility, then redefine the heterogeneity characteristics of the high-frequency intraday market to construct the optimal generalized heterogeneity autoregressive model, and finally predict and assess the intraday market risk. The results show that the intraday jumps of the high-frequency crude oil futures mostly occur in the event window of geopolitical news and EIA announcements, and there are short-term jump aggregations. Separating the jumping components can establish a more accurate prediction model for the fluctuation process. The model proposed takes into account the characteristics of intraday heterogeneity and finds that weekly fluctuations have no information contribution to high-frequency traders. Compared with the ARMA and GARCH models, it ensures the validity and accuracy of the results. With easy operation and scalability, it is an effective risk management tool for crude oil intraday market transactions.

Keywords value at risk, high-frequency spot volatility, crude oil futures market, market heterogeneity, jump recognition