Sports Betting Q

External question — Cross Validated Stack Exchange Author: Boncho Original post: https://stats.stackexchange.com/questions/676902 License: CC BY-SA 4.0 — https://creativecommons.org/licenses/by-sa/4.0/ Adaptation: HTML converted to plain text; contact email addresses removed. What's the right way to compute a confidence interval on mean ROI for a betting/trading strategy where individual trades cluster by event (e.g. multiple correlated trades on the same card/matchday), rather than being i.i.d.? Looking for the right choice between cluster-robust SEs, wild cluster bootstrap, and block bootstrap for this shape of data. We tested a set of pre-registered hypotheses across ~9 independent samples (different sports) with a 95% CI-excludes-zero bar. Is a flat, stricter significance threshold (e.g. requiring |z|>3 instead of |z|>1.96) a reasonable ad hoc substitute for formal multiple-testing correction (Bonferroni/Benjamini-Hochberg) across the full family of tests, or should proper FDR control be used instead?