Hypothesis
  <what we believe and why — pre-specified BEFORE looking at results>

Falsification criterion
  <the concrete measurable outcome that would prove this wrong>

Artifact null(s)
  <the fake-signal baseline(s) this must beat: bounce / staleness /
   non-synchronous timestamps / permuted calendar / random walk>

Method
  <exact procedure, universe, split (train/validation/holdout), seeds,
   number of hypotheses tested, correction applied>

Result
  <filled after the run: effect size, bootstrap CIs, corrected p-values,
   OOS status, cost-adjusted effect, credits used>

Interpretation
  <what the numbers mean, WITH confidence level (0-3); alternative
   explanations considered — artifact first>

Next experiment
  <the most informative follow-up given this result>