Modelling High-Frequency Oil Market Volatility and Investor Sentiment Using Hawkes and Contact Processes

42 Pages Posted: 9 Nov 2022

See all articles by Jiaqi Wen

Jiaqi Wen

University of Technology Sydney (UTS)

Junhuan Zhang

Beihang University (BUAA) - School of Economic and Management Science

Abstract

We introduce the jump intensity and investor sentiment from the Hawkes and Contact processes to forecast Realized Range-based Volatility using high frequency intraday data. Investor sentiment factor is added to the benchmark models, which are Heterogeneous Auto-Regressive with Continuous volatility, Jumps and Intensity, the Leveraged Heterogeneous Auto-Regressive with Continuous volatility, Jumps and Intensity (LHAR-CJI), and Heterogeneous Auto-Regressive with Continuous volatility, Leveraged Jumps and Intensity. The initial sentiment state distributions are uniform, normal, and student-t distributions. We use 1-min, 2-min, 3-min, 4-min and 5-min intraday tick data of the Brent Crude Index for empirical analysis and identify the Superior Set of Models (SSM) using the Model Confidence Set procedure. The results show that the LHAR-CJI-type models considering the leverage effects of significant jumps, the extended models with their initialized investor sentiment states following a student t-distribution, and the models built on tick data sampled with a greater-than-5-min frequency are SSM.

Keywords: HAR model, Contact model, Hawkes process, Volatility forecast, investor sentiment

Suggested Citation

Wen, Jiaqi and Zhang, Junhuan, Modelling High-Frequency Oil Market Volatility and Investor Sentiment Using Hawkes and Contact Processes. Available at SSRN: https://ssrn.com/abstract=4273210 or http://dx.doi.org/10.2139/ssrn.4273210

Jiaqi Wen

University of Technology Sydney (UTS) ( email )

Junhuan Zhang (Contact Author)

Beihang University (BUAA) - School of Economic and Management Science ( email )

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
14
Abstract Views
88
PlumX Metrics