Title: Revisiting the evidence for rough volatility in daily realized variances
Authors: Hiroyuki Kawakatsu - Dublin City University (Ireland) [presenting]
Abstract: The evidence for rough volatility in the daily log realized variance series is reexamined. For the class of ARMA(1,1) models with fractional Gaussian noise, there are two nearly observationally equivalent models, one with short memory and one with long memory. The practical difference between the two models is explored through comparisons of persistence in autocorrelations and (pseudo)out-of-sample performance.