Better Estimates, Earlier: Gro's Models Accurately Predict Government Data

Gro's customers improve their short- and long-term purchasing and product-demand forecasts with Gro’s model projections that are within 1.2% of the USDA’s eventual estimates nearly six months in advance. Each year, Gro's team of agriculture and climate experts incorporates new sources and historical data to improve the models’ predictive value.

This year, the USDA's September WASDE report underscored how Gro’s machine-learning models can serve as an accurate, early predictor of government indicators - both in the US and around the world.

Gro’s Planting Intentions Model led the USDA’s September WASDE number by six months because it incorporates multiple current and historical datasets. The Gro Yield Forecast Model led the USDA by two months because it updates daily rather than monthly during the growing season. Daily-updating products like Gro’s Yield Forecast Models enable customers to get ahead of government data, which is especially valuable in the three to four weeks between official reports.

1) Gro's Planting Intentions Model Predicts High Corn Planted Area Six Months Ahead of the USDA

In its September 10 WASDE report, the USDA increased its planted area estimate for US corn, bringing the USDA estimate in line with Gro’s Planting Intentions forecast released in mid-March.

Almost six months ago, Gro’s machine learning-based Planting Intentions Model projected strong increases in both corn and soybean planted area in 2021 based on the strong prices offered to farmers before this season’s planting began. The model forecast that US growers intended to plant 93.1 million acres of corn and 88.6 million acres of soybeans this year.

Gro's Planting Intentions Model improves on USDA estimates by:

• Integrating current and historical datasets on local new crop cash price relationships, futures prices, location, and historical planted area to make state-level predictions
• Publishing state-level projections 3-4 weeks before the USDA’s Prospective Plantings report is released

2) Gro's Daily Corn and Soybean Yield Forecast Models Predict US Corn and Soybean Yield Estimates Two Months Ahead of USDA

In its July WASDE report, the USDA estimated corn and soybean yields at 179.5 bushels/acre and 50.8 bu/acre, respectively. In September, they lowered those numbers to align with Gro’s machine-learning yield forecasts.

Almost two months ago, on July 11, Gro’s Corn Yield Forecast Model projected a US corn yield of 176 bu/acre, predicting the USDA’s revised number of 176.3 bu/acre. For soybeans, on July 16, Gro’s Soybean Yield Forecast Model projected 50.54 bu/acre, predicting the USDA’s revised 50.6 bu/acre.

Gro’s Yield Forecast Models improve on USDA estimates by:

• Providing daily updates for the most accurate forecast of yields based on climate and weather conditions seen in the season to date (vs. monthly-only reports from the USDA)
• Including weather forecasts for the next 16 days
• Incorporating near-real-time environmental and growing conditions data on a district, province, and/or national level to give earlier insight into where the USDA is heading

Gro’s machine-learning models provide powerful predictive abilities that incorporate multiple factors to enable better, more accurate decisions months before official government reports.