We used the multivariate t copula, which can capture the tail de-
pendence to modeling the dependence structure of the risk in portfolio analysis.
Multivariate t copula based on GARCH model was used to explain portfolio risk
structure for high-dimensional asset allocation issue. With this method we used
the Monte Carlo simulation and the results of multivariate t copula to estimate
the expected shortfall of the portfolio. Finally, we obtained the optimal weighted
for conditional Value-at-Risk (CVaR) model with the assumption of multivariate
distribution to illustrate the potential model risk among portfolios returns.