Subjective views and stress-tests for optimization of credit risk of loan portfolio

I apply the methods to express subjective views in the spirit of Black-Litterman in the optimization of CVaR of a loan portfolio in order to explore the effects of the stress-tests of parameters defining the risk and return of a loan portfolio. The loan portfolio returns distribution is characterized by the joint default distribu- tion, where each individual loan has a high probability of small profit (no default) and a low probability of big losses (default). Normality assumption does not hold thus the standard mean-variance optimization and Black-Litterman views cannot be applied. I simulate the joint distribution of the times to default and express views on different properties of the joint distribution of discrete default events: means, correlation, tail dependence. I optimize the CVaR of the original and stressed portfolios and compare the weights.