Realised Volatility Estimation Shortcuts: An Empirical Analysis

This paper features an empirical evaluation of Yang and Zhang’s (2000) short-cut method for calculating drift-independent realised volatility, based on high, low, open, and close prices using a historical time series. They suggest that this method is unbiased in the continuous limit, independent of the drift, and consistent when faced with opening price jumps. Souto and Moradi (2024) published Python script to implement this method and claimed that this was the first open-source code made available, that automates its estimation from high-frequency intraday stock data. We convert their code to R script and undertake an evaluation of the method using high-frequency hourly data on Bitcoin, originally sourced from Binance, for a period from January 1, 2018, to May 29, 2025. The method is benchmarked against the Bipower variation method of Barndorff-Nielsen and Shephard (2004a, 2004b). The ordinary least squares regression analysis reveals a close relationship between the methods. We further explore the persistence of volatility estimates by regressing the absolute values of daily Bitcoin returns on lags of themselves. The results suggest significant persistence out to seven daily lags. This finding supports the range enhanced GARCH model for modelling cryptocurrencies of Fiszeder et al. (2024). (The R code developed is attached in the Appendix).