How Gro Is Generating Its Supply/Demand Forecasts for Friday Release

Talk to our our team about Gro's offering
Talk to our team

Gro Intelligence will release on Jan. 11 its report on supply and demand forecasts for agricultural crops worldwide to fill the void left by the delay of the USDA’s closely watched WASDE report.

Gro’s forecasts will be released on the company’s data platform at noon EST Friday. Gro has been offering free subscriptions to our web app since shortly after the US government shutdown began on Dec. 22. Interested users can go to to access all of Gro’s data, including predictive-model outputs free of charge.

Gro has one of the world’s most comprehensive agricultural data platforms, making it uniquely positioned to reproduce the large number of robust forecasts contained in the USDA’s monthly World Agricultural Supply and Demand Estimates report, or WASDE. Gro will be forecasting demand for corn, wheat, and soybeans. Gro will forecast production and usage for major world crops, including corn, soy, and wheat, for the 2018/19 marketing year. It will also estimate production for other crops such as cotton, rice, and sugar. In all, Gro will be releasing over 1,000 forecast models covering 35 crops grown around the world.

The USDA announced last week it would delay releasing the January WASDE report, which was due to come out on Jan. 11, because of the US government shutdown. Gro’s own report will follow a similar format to the analysis and statistical tables in WASDE. Although the USDA report is considered the gold standard for data among agricultural market participants, the Gro estimates will provide markets with the closest alternative to the delayed WASDE. The estimates will leverage Gro’s existing, market-tested, predictive-modeling framework along with the hundreds of trillions of data points available in our platform.

Regularly published trade estimates, or market polls, only cover small portions of the WASDE, both in terms of numbers of crops and variables forecasted. They are impossible to confirm because industry analysts do not provide their individual methodologies. Gro’s estimates will not only cover all major crops in WASDE and introduce forecasts for over 20 crops not covered in WASDE, but, consistent with our mission and our approach to date, will be entirely reproducible using Gro data through both the Gro web app and the Gro API. Gro’s estimates leverage more than four years of work harvesting and normalizing vast amounts of data from a diverse range of global sources and building a predictive engine that allows us to generate quantitative forecasts seamlessly and in a reproducible format.

Gro will be publishing the details behind our forecasts in the coming weeks so that anyone can recreate our results using the data in our platform. Gro is one of the only companies in the world with such a comprehensive data platform, encompassing more than 40 million unique agriculture-related data series and nearly 500 trillion data points. We believe we are the only company that is entirely open about its methodology.

With data updates from the US government on hold, the Gro estimates will analyze country/crop pairings using five distinct methodologies for both supply and demand.

One set of forecasts leverages Gro’s machine-learning-based yield and production forecast models, which update daily on a district level for in-season crops. These models, whose baseline methodology was first released to the public in 2017, have been extensively backtested for performance.

Another set of production forecasts, which mostly applies to Southern Hemisphere countries for which Gro does not yet have in-season daily yield models, uses an array of satellite-generated knowledge to assess the current weather’s impact on a crop.

A third set of forecasts relies on country-specific data sources that are available in Gro and that the USDA typically uses to generate WASDE estimates. For example, ROSSTAT, the Russian statistics service, recently updated data for that country’s wheat crop, and CONAB, an aggregator of Brazilian agricultural information, has recent data on corn and soybeans.

Our fourth set of forecasts only applies to crops in Northern Hemisphere countries that are between seasons and for which production estimates are unlikely to have changed significantly in recent weeks (growing seasons can be tracked using crop progress calendars available in Gro). Gro may choose to leave data for these crops unchanged from what was reported in the last WASDE report issued on Dec. 11, as we believe the USDA would also have left them unchanged if the government shutdown had not delayed the report.

A last set of forecasts uses linear regression to model annual crop demand, based on the pace of higher frequency US demand statistics for the current marketing year. In addition, various non-US data sets that are available in Gro are used to gather new weekly and monthly data to better inform cumulative year-over-year comparisons with conservative assumptions about future progress.

Gro will continue to publish its global crop supply and demand estimates free of charge until the US government reopens. We will also be publishing a summary of our methodology on our website in the next couple of weeks. For paid users with API access, we will also publish code to replicate models in our API library.

About Gro Intelligence:

Gro Intelligence is a data and analytics company that has built one of the world’s most comprehensive agriculture-related data platforms. Gro uses its technology to harvest, translate, and transform approximately 40 million unique data sets and nearly 500 trillion data points across multiple types of data and crop types on a global basis. Gro has also built a suite of machine-learning-based predictive models allowing users to have access to an unprecedented number of data sets and forecasts for production, consumption, trade, and climate globally. Gro has simplified and streamlined the discovery process of data and added a layer of intelligence on top of it.

Customers primarily access the data and Gro’s various analytical models by signing up for free through the Gro website and directly through Gro's API, which is also available free-of-charge to interested parties for a 90-day trial period. Please go to for more information or contact us at

Get a demo of Gro
Talk to our enterprise sales team or walk through our platform