Financial Mathematics, Volatility and Covariance Modelling

11Forecasting realized volatility measures with multivariate and univariate models

The case of the US banking sector

Gianluca Cubadda, Alain Hecq, and Antonio Riccardo

1 Introduction

Realized volatility measures, such as the realized variance (RV) or the bipower variation, are estimates of asset volatilities within a short period, for instance one day, using intra-day returns. The 5-minute Realized Variance (RV5), a benchmark often considered in empirical finance (see Liu et al., 2015), is obtained as $RV{5}_t\equiv {\sum }_{j=1}^M{r}_{t,j}^2,$;where r_{t, j} are the high frequency returns, observed for M intra-day 5-min periods. Both computational simplicity and theoretical foundations make realized volatility measures very attractive among practitioners and academics ...