The pre-registered evaluation window has closed. H-EX1 finished significantly negative out-of-sample: n=309, Δ −3.0pp vs the same-day close, pooled 95% CI [−4.4, −1.5], cluster CI [−6.0, −0.6], 5 of 16 names favoring it. No selection filter showed an edge. Across 11 live rules there are 6 positive point estimates against 5.5 expected by chance, and both significant results are negative.
This is the pre-registered null, confirmed — the process working as designed. Per the registry, monetization is No-Go and the model stays labelled unvalidated. Scope: this refutes this screen's exit rule and current filters at this sample and in this regime — it is not a claim that no edge could ever exist. The boxes below keep recomputing live from the public log, so they will drift past the numbers frozen here.
Monetization is downstream of this gate, never a parallel track — there is nothing to sell until the pre-registered edge survives the forward test. The honest prior is that it won't, and that's a success of the process, not a failure. Every box below is computed live from the public log; nothing is a matter of opinion. See the full framework in the method.
Loading the live leaderboard…