Vine copula is useful in modelling high dimensional distribution and dependence. Generalized additive models is a flexible method for doing regression. In this thesis, we first fit static R-vine copula models to the S&P 100 data using different vine structure selection methods. Based on the analysis of the static R-vine copula, we estimate the dynamic vine copula models for the S&P 100 data with the GAM vine copula models. We also explore the usefulness of GAM-HAR models for Kendall's τ forecasting using lagged covariates on different time scale.
«
Vine copula is useful in modelling high dimensional distribution and dependence. Generalized additive models is a flexible method for doing regression. In this thesis, we first fit static R-vine copula models to the S&P; 100 data using different vine structure selection methods. Based on the analysis of the static R-vine copula, we estimate the dynamic vine copula models for the S&P; 100 data with the GAM vine copula models. We also explore the usefulness of GAM-HAR models for Kendall's τ forecastin...
»