A specific problem in financial optimal design is addressed in this paper: the construction of collateralized debt obligation (CDO) transactions with minimum overlapping between two tranches. A CDO is divided into tranches, each consisting of a subset of the underlying credit derivatives. Minimizing overlap between two tranches maximizes diversification so that investors can spread their risks or choose the preferred level of risk.
In the most general form, the problem is stated as a generalization of the balanced incomplete block design (BIBD), also arising in combinatorial design theory. In Section 3, the authors suggest the use of constraint programming for solving small or medium instances of this problem. In Section 4, they present a heuristic procedure for the larger instances arising in financial applications.
The paper is clearly written—even a user that is not an expert in financial applications will be able to read it with no particular difficulties; various examples are included. The computational experiments on large instances are limited, and the paper lacks a comparison with other possible approaches, such as integer programming. Section 2.3 on BIBDs is also extremely interesting.
As reported in the final section of the paper, the proposed strategy has eliminated the need for ad hoc manual permutations when designing CDO squared transactions. From a more general point of view, it would be interesting to include as a criterion not only the number of overlapping derivatives, but also some measure of risk.