Commodities research

Inventory surprise, storage and weather: official point-in-time vintages, first-release capture, and what happens to a seasonal signal net of cost.

Commodity research here follows physical constraint rather than sentiment. Storage capacity, inventory levels, seasonal demand and hedger positioning have official point-in-time records. Agencies publish these quantities and their revisions on a schedule, making vintage control practical.

Vintage discipline is the entire methodological content of this cluster. An inventory figure exists in several versions: the first release, and every revision after it. A study that uses the current value is using a number that was not available for weeks, and because revisions are not random the resulting effect can be manufactured entirely by hindsight. Capturing the first release, and only the first release, is what makes a surprise a surprise.

The results have been unkind. Inventory seasonality, cross-sectional futures carry and cross-asset carry were each tested and published as nulls. The hedging-pressure and storage-weather mechanisms remain at the feasibility stage. One is blocked on a free credential and the other on paid data. A priced blocker is not a result, so both remain labeled as blockers.

The general lesson from this cluster is about cost rather than signal. Seasonal commodity effects tend to be real, small, and slower than their turnover implies; net of roll, spread and the calendar alignment that a twenty-four-hour market imposes on a business-day book, several survived as ideas and none survived as sleeves.

The commodities documents