Tools I Built for the Public

Corn yield forecasts move markets. Every year, substantial sums in commodity trades are made or lost in the weeks before USDA releases its official monthly estimates, and for most of that window, the public is flying blind. Large institutional players manage this uncertainty with proprietary forecast models built on satellite data, weather analytics, and their own internal data pipelines. Farmers, regional co-ops, and smaller traders have more limited access to this kind of intelligence, leaving them more passive in the market and often having to comply with price signals set by larger players.

This project is an attempt to close that gap. I collect data from open satellite archives, weather reanalysis, and USDA crop progress reports, and apply state-of-the-art published methods to generate county-level corn yield forecasts for the 12 top-producing states in the US Corn Belt, updated throughout the growing season and freely accessible to anyone. I also publish my own national-level corn yield forecast, serving as an independent second signal alongside the official USDA estimates. The goal is twofold: to provide a genuine public good by democratizing early-season yield intelligence, and to benchmark published forecasting methods against each other in a real, operational setting rather than the controlled conditions of academic papers.

Alongside the forecasts, I built a monitoring dashboard that aggregates the full picture at county level: historical and current vegetation indices, weather anomalies, and drought indices — all the leading indicators that impact final yield. It is designed to be a no-friction first stop for anyone who wants a quick read on what is happening in the Corn Belt right now, without the overhead of pulling and processing the underlying data themselves.