The Cross-Section of Option Returns : Deriving Inferences in Sparse Models

This paper derives accurate inferences about the contribution of a high-dimensional set of option and stock characteristics to the cross-sectional variation in delta-hedged option returns. Unlike the extant literature that is largely focused on the construction of predictive models, we apply sparse modeling combined with a nonparametric bootstrap approach to perform variable selection and model estimation simultaneously. We document evidence on deficiencies of conventional approaches in high dimensional settings and the post-selection estimators of sparse models in making inferences. Further, we provide accurate confidence intervals for estimating the marginal contributions of numerous variables to the cross-sectional variation of option returns