Most of the existing studies on portfolio loss focus on default correlation and tail dependence. Few of them consider the dynamics of risk structure and the endogenous recovery rate. We introduce a new approach to the evaluation of multi-period portfolio credit risk using a firm’s internal valuation information. The underlying idea of our framework is to construct a portfolio loss model with dynamic risk structure which is linked to a stochastic state model through a factor copula approach. And we demonstrate the examples of the model application in CBO tranches pricing.