Fintech Thrust Seminar | Rough Volatility Models for Realized Volatility of Cryptocurrency
Supporting the below United Nations Sustainable Development Goals:支持以下聯合國可持續發展目標：支持以下联合国可持续发展目标：
Rough Volatility Models for Realized Volatility of Cryptocurrency
Abstract:
This paper introduces a new model for log RV of cryptocurrency. The specification is a discretized fractional Ornstein-Uhlenbeck (fOU) process with a time-varying persistency that depends on realized quadraticity, as in the HARQ model of Bollerslev, Patton, and Quaedvlieg (2016). Under the in-fill asymptotic scheme, the process has a local-to-stochastic-unit-root specification, with the error term being a fractional Gaussian noise so that fractional integration is possible. Methods are proposed to estimate parameters in the model. The asymptotic theory is developed for the estimators. Empirical estimates from RV of 110 cryptocurrencies suggest strong evidence of time-varying persistency and roughness in the fractional Brownian motion. When using the proposed model to forecast the RV of 110 cryptocurrencies, we find evidence of superior forecasting performance of the proposed model relative to other popular models in the literature for most of the RV series.
Xiaohu WANG is an associate professor at the School of Economics, Fudan University. His research focuses on financial econometrics, econometric theory and empirical asset pricing. He primarily works on topics related to rough volatility models, bubble testing, structural breaks, and estimation and inference of continuous-time models. His papers have been published on the Journal of Econometrics, Econometrics Journal, Journal of International Money and Finance and Econometric Reviews.