Synchronizing multivariate financial time series

Francesco Audrino, Peter Bühlmann

Abstract

ABSTRACT

The prices or returns of financial assets in a particular market are most often collected during that market's trading hours. The need to synchronize multivariate time series of financial prices or returns arises from the fact that information on closed markets continues to flow while others are open. We propose a synchronization technique that also includes the modeling of synchronized returns simultaneously; the overall method can then be described in terms of a new model for asynchronous returns. In addition to being a nice interpretation of synchronization, we found empirically that the method has the potential to increase the predictive performance of many reasonable models for a seven-dimensional time series of daily equity index returns and that it is more appropriate for the calculation of portfolio risk measures.