 |
Center for Nonlinear Studies |
Multivariate models with dependent variables are popular in many fields, such as quantitative finance, engineering, statistics, and medicine. Simulation of copulas can be done by Monte Carlo (MC) methods or quasi-Monte Carlo (QMC) methods. Goodness-of-fit tests can be used to find the best simulation algorithms for copulas. I will introduce a new goodness-of-fit test based on the collision test and low-discrepancy sequences, and present numerical results on option pricing and Value at Risk estimation via copula models. The numerical results show the improvements of QMC simulations of copulas over MC simulations, and the advantages of the new goodness-of-fit test for copulas for large samples.
Host: Peter Brady