# Independent reproduction protocol CUPED-PRE-20260831

**Published:** 2026-08-31  
**Status:** Open protocol; no qualifying independent report received as of 2026-08-31

1. Disclose identity, affiliation, compensation, conflicts, data rights, experiment context, and decision authority.
2. Pre-register unit, assignment, primary, covariate name, pre-period, estimand, missing-history treatment, guardrails, and stopping rule.
3. Test timestamps, treatment leakage, correlation, covariance, theta estimation, null coverage, new-user share, balance, and row retention.
4. Publish de-identified data or sufficient statistics, unadjusted and adjusted code, both intervals, sensitivity checks, failures, and limitations.
5. Keep the unadjusted result; hold CUPED where pre-period proof or population coverage fails.

A qualifying report must be authored outside InfiniSynapse, use independently controlled assignment and outcome data, expose reproducible calculations, and document statistical review. Running the first-party checker alone does not qualify.

Send a public report URL or documented contradiction to [zhuhl@infinisynapse.com](mailto:zhuhl@infinisynapse.com).