Multimedialne seminarium z ekono- i socjofizyki

A novel class of the conditional distribution of the Copula-based M-GARCH models is studied. Initially we consider the multivariate distribution, defined as a product of independent univariate skewed Student-t components. Then, an orthogonal transformation is applied in order to model heavy tails and asymmetry along free set of coordinate axes. We apply the Bayesian approach to model comparison and verify the empirical importance of proposed generalisations. Also the posterior inference about some bivariete processes on the Warsaw Stock Exchange is presented.