To estimate value at risk (VaR) and conditional value at risk (CVaR) of nonferrous metals futures, this study estimates the shape parameter and scale parameter in peaks over threshold model (POT) model by particle swarm optimization algorithm (PSO). Hill plot and mean excess function (MEF) plot together determine the appropriate thresholds of POT model for different nonferrous metals futures. The volatility of the return is computed respectively by generalized Pareto distribution (GPD) model and GARCH-GPD model. The results show that VaR and CVaR based on PSO-GARCH-GPD are effective to measure...
