CMStatistics 2018: Start Registration
View Submission - CMStatistics
B1739
Title: A parallel regularized optimization approach to hedging a portfolio of collateralized mortgage obligations Authors:  Yury Gryazin - Idaho State University (United States) [presenting]
Abstract: Collateralized Mortgage Obligations (CMO) can exhibit different degrees of cash flow variability depending on tranche structure and realized prepayment speeds. In order to have a meaningful comparison across structures and collateral types, an Option-Adjusted Spread (OAS) methodology is typically used. While OAS provides a meaningful improvement to yield to maturity for CMO, it does not fully account for embedded optionality. We introduce a related methodology which significantly improves valuation metric for CMOs. In this method, the construction of the optimal hedging policy is considered an essential part of the valuation procedure. To ensure the convergence of the resulting ill-conditioned optimization problem, the standard Tikhonov type regularization technique is applied. The convergence of this method and the stability of the underlying control optimization problem are discussed. The numerical results on the hedging of the portfolios of CMO and European swaptions based on parallel Monte-Carlo simulation on multicore PCs are presented.