Datasets
Commodities & agro data API (EIA · USDA · Brazil)
Commodity fundamentals are published by agencies on strict calendars, which makes them ideal point-in-time data and terrible spreadsheet data. EIA inventories arrive weekly, WASDE lands monthly at a fixed hour with estimates that get revised for years, and Brazilian agro statistics follow their own harvest-driven schedule in a different language.
Hindcast is building these into one queryable layer where every value is stamped with its release, and revisions are added rather than substituted. For anything with a scheduled release the distinction between the reference period and the publication moment is the whole game — a WASDE estimate for a crop year means something different the day before and the day after the report.
The Brazil angle is deliberate. Brazilian production and export data moves global grain and protein markets, and it is systematically underrepresented in agent-accessible sources, which is exactly the gap this catalog exists to close.
hindcast — session
previewhindcast> what did the last WASDE change versus the prior month?
→ tool: wasde_release { commodity:"...", release:"latest", diff:"previous" }
✓ changed line items with both vintages side by side
source: USDA WASDE, each release retained as its own vintage
What's covered
- EIA energy series including inventories and production, by release date
- USDA WASDE estimates with every monthly vintage retained
- Brazilian production, export and harvest data aligned to publication
- Weather series joined to producing regions rather than to countries
- Reference period and release timestamp stored separately
- Revisions added as new observations, never overwriting the first print
Join the waitlist — this page's dataset is already selected.
Questions
Why keep every WASDE vintage instead of the final numbers?
Because markets trade the estimate, not the eventual truth. Studying how prices responded to a report requires the report as published; using final revised figures answers a question nobody could act on.
How is weather data tied to commodities?
By producing region rather than by country, since national averages hide the localized conditions that actually move a crop. Regions are joined to the same entity and geography identifiers used elsewhere in the catalog.
Related
Join the waitlist. Vote the roadmap. First in, first served.