Properties of Realized Volatility and Correlation

Abstract

The aim of this study is to measure and characterize one-day and one-week volatility and correlation on a number of German equities. For this purpose we measure the realized volatility from tick-by-tick data by summing up short term squared returns. Volume weighted average prices are used to filter the price series. The distributional properties of returns standardized by realized volatility are shown to be gaussian or close to gaussian. In accordance with the literature we also find that weekly returns are more normally distributed than daily returns. Also, weekly returns standardized by realized volatilities are found to be more normally distributed than daily standardized returns. In comparison to similar studies on U.S. equity data we can conclude that the distributional properties of returns and realized volatilities are alike. After examining the memory of realized volatility we find that it is highly unstable implying poor forecasting performance of parametric volatility models.