A0652
Title: Modelling high dimensional realized correlations
Authors: Roxana Halbleib - University of Freiburg (Germany) [presenting]
Ilya Archakov - York University (Canada)
Jasper Rennspies - University of Freiburg (Germany)
Abstract: A novel approach is proposed to model and forecast high-dimensional realized correlation matrices by using the matrix logarithm transformation introduced by a prior study. To reduce dimensionality, the focus is on extracting factors from the transformed correlation series, i.e., gamma series, according to a particular sector or industry. The empirical results show that the new approach is easily implementable for large-dimensional matrices and outperforms standard methods.
