The authors introduce a novel bootstrap approach to resampling asset price data that can be used for both finite-maturity assets and equities. The key insight is that they bootstrap primitive objects with more appealing statistical properties to avoid resampling series with strong time-series and cross-sectional dependence. They then recover the original dependence structure in an internally consistent manner via definitional identities. Their bootstrap is nonparametric in nature and so avoids the common practice of committing to a tightly parameterized pricing model with explicit assumptions on the form of cross-sectional and time-series dependence. They demonstrate the appealing finite-sample properties of their bootstrap approach in a series of simulation experiments and empirical applications.

The authors introduce a novel bootstrap approach to resampling asset price data that can be used for both finite-maturity assets and equities. The key insight is that they bootstrap primitive objects with more appealing statistical properties to avoid resampling series with strong time-series and cross-sectional dependence. They then recover the original dependence structure in an internally consistent manner via definitional identities. Their bootstrap is nonparametric in nature and so avoids the common practice of committing to a tightly parameterized pricing model with explicit assumptions on the form of cross-sectional and time-series dependence. They demonstrate the appealing finite-sample properties of their bootstrap approach in a series of simulation experiments and empirical applications.

The authors introduce a novel bootstrap approach to resampling asset price data that can be used for both finite-maturity assets and equities. The key insight is that they bootstrap primitive objects with more appealing statistical properties to avoid resampling series with strong time-series and cross-sectional dependence. They then recover the original dependence structure in an internally consistent manner via definitional identities. Their bootstrap is nonparametric in nature and so avoids the common practice of committing to a tightly parameterized pricing model with explicit assumptions on the form of cross-sectional and time-series dependence. They demonstrate the appealing finite-sample properties of their bootstrap approach in a series of simulation experiments and empirical applications.