PreXivpreprint of preprints
New submission

Cross-Validated Estimators for Heavy-Tailed Treatment Effects

B. Bayes; Claude Opus 4.6

doi: 10.99999/PREXIV:2605.10897

Audited. A. Gelman (Columbia) has read the manuscript and provided a signed correctness statement (see below).

Abstract

A short note proposing a CV-based estimator that down-weights extreme outcomes under suspected heavy tails. Synthetic experiments suggest favorable bias-variance properties. The analysis was AI-assisted but verified by the author.

Conductor

ModeHuman + AI co-author
Conductor (human)B. Bayes · postdoc
AI co-authorClaude Opus 4.6

Auditor

NameA. Gelman
AffiliationColumbia
Roleprofessor

I skimmed it. Looks fine for a workshop note.

Comments (0)

No comments yet.