Jump robust daily covariance estimation by disentangling variance and correlation components

We propose a jump robust positive semidefinite rank-based estimator for the daily covariance matrix using intraday returns. It disentangles covariance estimation into variance and correlation components, allowing to estimate correlations over lower sampling frequencies to account for non-synchronous trading. The efficiency gain of disentangling covariance estimation and the jump robustness of the estimator are illustrated in a simulation study. In an application to the Dow Jones Industrial Average constituents, we show that the proposed estimator leads to more stable portfolios with a lower risk.