A new approach to portfolio analysis of financial market risks by random set tools is considered. Despite many attempts, the consistent and global modeling of financial markets remains an open problem. In particular it remains a challenge to find a simple and tractable economic and probabilistic approach to market modeling. This paper attempts to highlight fundamental properties that a market model should possess. The paper suggests a random set approach as a probabilistic base of this model. Using this approach it is possible to establish a corresponding interactive market dynamics that involves a minimal number of sets. These sets include the set of capital surpluses, the set of capital within assets and the set of capital deficits. Several interesting properties related to random volatility of assets quality, probabilities of quality categories and defaults and matrices of transition probabilities of switching among categories can be derived. In addition the random set approach allows to derive the so called transition set-matrices, random set invariants of capital redistribution processes. Empirical evidence will be given that confirm these random set findings. The approach is also illustrated by collapses in U.S. financial markets in 90's and can be used to explain Russian default'98.