Does public corporate information produce a systematic, statistically reproducible underreaction? This desk measures that claim. It does not sell a score.
Companies
—
Priced events
—
Dev 2016–24
—
Holdout 2025
—
Pipeline starting. Integrity pending.
Verdict
The engine is running on a repaired clock: SEC acceptance time, NYSE session mapping, Yahoo prices (Stooq HTML rejected), causal analogs, and a frozen 2025 holdout. Results appear here when the sample is priced.
Validated
0
FDR + 2025 sign
Exploratory
0
dev only
Failed / tiny
0
not a product
Ranked hypotheses
Development 2016–2024. Holdout is shown, never used to rewrite the rule.