RMV - package for regularized multivariate volatility modeling

This Matlab package builds upon the MFE toolbox (Sheppard, 2013) and provides functions for (feasible) regularized multivariate volatility modeling. It support both CAW (see function caw_regularized()) and BEKK (see function bekk_regularized()) models. Each of these functions find the optimal degree of regularization for both diagonal and off diagonal parameters via a pseudo out-of-sample evaluation and then estimate the optimally regularized model on the whole supplied dataset. As of now, it only supports order p=q=1. For more information, see the accompanying article “Unrestricted, Restricted, and Regularized Models for Forecasting Multivariate Volatility” by Stanislav Anatolyev and Filip Staněk.