YieldCast: Best Track Record In the Corn Industry Last 9 Years

YieldCast is a stand-alone service of T-storm Weather that forecasts the U.S. corn and soybean yield via robust statistical models.  The report will be released every Wednesday at 12 PM Central Time from April 17 until October 9 in 2019.  Forecast accuracy and subscription information are below.

The real test of the accuracy of U.S. corn and soybean yield forecasts takes place before the USDA issues their first forecast in the August Crop Production report.  During critical periods leading into that report, yield forecasters do not have the benefit of “anchoring” to USDA estimates.

The performance of YieldCast corn forecasts in early August is truly outstanding.  Over the last 9 growing seasons, before the release of the August Crop Production report, the YieldCast forecast of U.S. corn yield averaged #1 closest to the final yield of approximately 35 market analysts reported via Bloomberg and Reuters, including:

#1 most accurate of 32 forecasts in 2010
#2 most accurate of 31 forecasts in 2011
#1 most accurate of 40 forecasts in 2015
#3 most accurate of 34 forecasts in 2016
#1 most accurate of 29 forecasts in 2017

Equally notably, YieldCast forecasts in early August averaged closer to the final corn yield than projected by the USDA in the August Crop Production report.  And, YieldCast forecasts in early October were nearly identical to USDA October Crop Production reports.  See Figure 1, 2, 3, 4.

2018 provides another example of the value YieldCast provides with our October forecast having missed the true yield by less than 1 bpa (Figure 5):

In early August last year, there was a somewhat wide difference between the YieldCast, market, and USDA forecasts leading into the August Crop Production report, and the YieldCast forecast was too low (Figure 2).  However, USDA and market forecasts shot much higher through October, while YieldCast forecasts were fairly steady at approximately 175 bpa.  It took all the way until January for the final yield to be realized at 176.4 bpa, showing the market and USDA became much too optimistic with forecasts that were 5.2 and 4.3 bpa too high, while YieldCast was only 0.9 bpa too low.

YieldCast soybean forecasts have not been as accurate as for corn through early August, but averaged similar to the market and USDA over the last nine seasons, and were more accurate than the market average in 3 of the last 4 years after we adjusted the models to reflect changing technology.

If you are not subscribed to YieldCast and would like information for a subscription, please visit the YieldCast webpage, where full charts of our forecasts since 2010 for both corn and soybeans are available.

 

Figure 1

corn_mape_aug

 

 

Figure 2

corn_timeline_aug

 

Figure 3

corn_errors_aug_all_since_2010

Figure 4

corn_mape_oct

Figure 5

corn2018