A New Data-Driven Fixed-Income Risk Framework
Leveraging Advanced Statistical Methods to Construct Robust Issuer Credit Curves and Market Surfaces as the Basis for Granular, DTS-style Risk Modeling
Modeling potential losses of a credit-risky bond portfolio based on granular, issuer-level return data is notoriously difficult. A myriad of data-quality concerns arise, driven by a vast, frequently illiquid market for which evaluated pricing is often stale, inconsistent or simply missing. Many issuers have only a small number of bonds outstanding. In fact, generally less than half of the issuers in USD high yield index portfolios have more than one bond outstanding that meets standard requirements for inclusion in a model estimation universe (sufficient maturity, etc.). Thus great care must be used to extract signal from data noise.
Managing Director, Research at Qontigo
© 2020 Qontigo GmBH