What actually moved

What is and isn't measurable

• Only materials with a FREE exchange benchmark are measurable. OTC minerals (gallium, germanium, REEs, antimony, tungsten...) have no free futures curve; they are skipped, never faked. That is the majority of action-material pairs.
• Yahoo returns ~1255 daily bars for these futures (~5 years), so actions announced before roughly 2021-09 cannot be measured at all and are absent rather than zero.
• A reaction is DESCRIPTIVE history, not a prediction. The single-name equity event study on this corpus returned null; do not present these as forecasts of what a future measure will do.
• Standard errors are not clustered. Events overlap in time and share materials, so the effective sample is smaller than the nominal count.
• The “beats noise” rate assumes |z| ≥ 2 is a 5% event. Commodity returns are fat-tailed, so the true null rate is higher and that absolute figure flatters us. The relative comparison in finding 1 is the trustworthy one — both arms share the same fat-tailed null.

By material

The measured dislocations

Every event that beat the noise threshold, largest abnormal 21-day move first. Each row links to the filed action with its primary source, so any figure here is checkable back to the measure itself — a cohort statistic alone would not survive that scrutiny.