Skip to content

Scenario Generation for Time Series and Curves: A Comparison of Nonparametric and Semiparametric Bootstrap

Home / Papers / Scenario Generation for Time Series and Curves: A Comparison of Nonparametric and Semiparametric Bootstrap

Scenario Generation for Time Series and Curves: A Comparison of Nonparametric and Semiparametric Bootstrap

Authors: Nicola Baldoni, Michele Sparviero, Lorenzo Viola Date: 2026-06-10 Paper ID: openalex:2606.11859

Summary

This paper evaluates and contrasts nonparametric and semiparametric bootstrap techniques for generating stochastic asset-class trajectories. While traditional nonparametric stationary bootstrap methods successfully maintain historical return distributions, they often fail to generate paths that are economically plausible, particularly in fixed-income settings. By combining parametric models—such as autoregressive or mean-reverting structures—with the resampling of historical residuals, the authors show that semiparametric approaches can effectively bridge the gap between distributional accuracy and path-wise realism.

Key Contributions

  • Systematic comparison of nonparametric versus semiparametric bootstrap methods for stochastic trajectory generation in quantitative finance.
  • Demonstration that semiparametric bootstrap methods preserve both historical return distributions and path-wise economic realism for interest rates and yield curves.
  • Validation that integrating parametric term-structure models with residual resampling produces more coherent yield-curve simulations compared to traditional stationary bootstrap approaches.

Metadata & Links

url
https://arxiv.org/abs/2606.11859
paper_id
2606.11859
paper_source
openalex
domain
time-series
tags
architectures
datasets
concept_slugs
dataset_slugs
skill
TimeSeriesSkill
created_at
2026-06-13T06:28:08Z