Simulation II: Modelling Multivariate Financial Data

We return to the question of sampling for Monte Carlo, and specifically the exact sampling of models of multiple correlated assets, such as equities held in a portfolio. To the end we consider multivariate variants of the Normal and Student’s \(t_\nu \) distributions, considering sampling techniques, heaviness of tails and tail dependence. We also introduce copulas as a tool for understanding the dependence structure of multidimensional data, and for constructing bespoke multivariate distributions. Real-world equity data is used to demonstrate model fitting and the generation of synthetic data.