Control error under multiple testing

You have 1 primary and 12 secondary metrics with correlations, and you plan 4 weekly looks. a) Propose a hierarchy that preserves strong FWER control for confirmatory endpoints and FDR control for exploratory ones. b) If there are k=5 confirmatory metrics at α=0.05, compute the Holm–Bonferroni adjusted threshold for the smallest p‑value. c) For the remaining k=8 exploratory metrics, show how to apply BH at q=0.10 and discuss how to combine with group‑sequential alpha spending or always‑valid p‑values to avoid double‑counting looks.