A2024
Title: Exploiting ubiquitous local optima for forecasting emerging market credit spreads
Authors: Gary Anderson - CEMAR LLC (United States) [presenting]
Alena Audzeyeva - Keele University (United Kingdom)
Abstract: A novel approach for determining support vector regression (SVR) kernel parameters in the presence of multiple local optima is proposed. In contrast to existing approaches focusing on identifying a single best tuning parameter setting, an impractical goal in many financial market applications, the framework employs a global optimization algorithm to produce a collection of competitive SVR kernel parameter candidates, and applies the model confidence set test to select the most accurate subset from the collection of promising candidates. The approach is used to predict credit spreads for four mature emerging market sovereign borrowers. Combining forecasts into simple forecast combinations and contrasting them against random forests, standard SVR, and conventional benchmark model forecasts, substantial gains in forecast accuracy are found when forecast combinations from the most accurate model sets are used. Furthermore, analysis of the forecasting results provides useful insights into credit risk pricing by international investors.
