2026 crop size: every forecast
Every published forecast of the 2026 US corn and soybean yield on one page: USDA’s latest, the Pro Farmer tour, DTN and our own ratings model, each with the date it was said and a link to where. Nobody is scored until USDA prints the January final. How much the weekly ratings are worth, state by state, is in the research section further down.
Data: each forecast from the source that published it, linked beside it.
Every 2026 forecast, low to high
Read against the other published forecasts: corn: ours is number 3 of 4, low to high, 7.4 over the lowest; soybeans: ours is the highest of the 4, 1.4 over the lowest.
Corn bu/acre
Pro Farmer Crop Tour173.2 171.5–175.0 · 2026-08-21
DTN Digital Yield Tour178.5 · 2026-08-10
▸ AGSIST nowcast180.6 173.7–187.5 · 2026-10-04
USDA, October WASDE181.2 · 2026-10-09
4 forecasts on the table, 8.0 apart end to end.
Soybeans bu/acre
DTN Digital Yield Tour52.1 · 2026-08-10
USDA, October WASDE53.1 · 2026-10-09
Pro Farmer Crop Tour53.3 52.2–54.4 · 2026-08-21
▸ AGSIST nowcast53.5 51.1–55.9 · 2026-10-04
4 forecasts on the table, 1.4 apart end to end.
Everything here was published by somebody else and is linked to where they published it. Our own figure is read from the nowcast file below, so the two always match. Nobody is scored until USDA prints the January final. Cross-checked against the crop tour page so the two cannot print different numbers for the same figure (USDA corn, USDA soybeans, Pro Farmer corn).
AGSIST Yield Nowcast: where these fits point right now
ratings week ending 2026-10-04, the last week with enough history · no nowcast from the Oct 11 ratings: this late in the season only 9 past years can be checked, 10 needed
Corn: 2026 national yield
180.6 bu/ac
80% band ±6.9 · our straight-line trend 188.7 (2010–2025 finals)
▼ 8.1 below our trend · ▼ 2.4 below USDA’s 183.0 trend
Two trends: ours is a straight line through USDA’s national yields (2010–2025 finals). USDA’s 183.0 is its weather-adjusted 2026 trend (Feb 2026 Outlook Forum, kept in the May WASDE).
backtest average, not this week’s number (earlier years only, 2011–2024): MAE 4.2 vs 5.5 trend-only (24% better), n=12 yrs, 18 states
Soybeans: 2026 national yield
53.5 bu/ac
80% band ±2.4 · our straight-line trend 53.9 (2010–2025 finals)
≈ on our trend
backtest average, not this week’s number (earlier years only, 2011–2024): MAE 1.3 vs 2.4 trend-only (46% better), n=13 yrs, 18 states
On the record: direction vs USDA’s next print
Locked from the live model before each WASDE, graded by subtraction after it. Record so far: 3 of 6 correct (3 reports; 2 calls on a defective input, graded as called). Corn and soybeans on one report come from one model run, so the independent sample is the number of reports.
2026-10-09 beans ▲ UP model 53.7 · ratings week of Sep 13 · locked Oct 6 · vs USDA 52.8USDA 52.8 → 53.1 CORRECT
2026-10-09 corn ▲ UP model 182.6 · ratings week of Sep 13 · locked Oct 6 · vs USDA 178.5USDA 178.5 → 181.2 CORRECT
2026-10-09 beans ▲ UP model 53.7 · ratings week of Sep 13 · locked Sep 15 · vs USDA 52.7superseded: locked against 52.7, not the print in force; re-locked 2026-10-06 before the report, same call
2026-10-09 corn ▲ UP model 182.6 · ratings week of Sep 13 · locked Sep 15 · vs USDA 180.7superseded: locked against 180.7, not the print in force; re-locked 2026-10-06 before the report, same call
2026-09-11 beans ▲ UP model 54.1 · ratings week of Aug 16 · locked Aug 18 · vs USDA 52.7USDA 52.7 → 52.8 CORRECT
2026-09-11 corn ▲ UP model 182.6 · ratings week of Aug 16 · locked Aug 18 · vs USDA 180.7USDA 180.7 → 178.5 INCORRECT
2026-08-12 beans ▲ UP model 53.8 · ratings week of Aug 9 · locked Aug 11 · vs USDA 53.0USDA 53.0 → 52.7 INCORRECT defective input
2026-08-12 corn ▲ UP model 183.6 · ratings week of Aug 9 · locked Aug 11 · vs USDA 183.0USDA 183.0 → 180.7 INCORRECT defective input
Every state’s fit on this page, pointed at this week’s ratings and weighted by planted acres. The band is the 80th percentile of the model’s own misses at this week in a pseudo real-time backtest: each past year is estimated from earlier years only, the way it would have looked that season, never from the years after it. It earns its width from that backtest, and the number of years behind it is printed on each card: a dozen or so, so the band is coarse. The trend is a straight line through every USDA final since 2010. Each weekly number is kept and graded against USDA’s final yield in January, misses included.
Do crop ratings predict yield? The researchHow much the weekly Good+Excellent rating has explained of each state’s final yield, every week since 2000.
Every Monday afternoon the market trades the condition report hard. This section works out how much that number has been worth in past years: the share of final-yield variation the Good+Excellent rating has explained, for every state, at every week of the season, since 2000. week 40 of 2026 (the last week with a fit; week 41 has under 15 paired years): corn ratings explained 36% of how Iowa’s final yield differs from trend (n=22 yrs), 70% of Illinois’s (n=22 yrs), 90% of Tennessee’s (n=22 yrs) · by week 39 Illinois peaks at 76% (n=26 yrs) · fits vs detrended yield, 2000–2025, n≥15 · data refreshed 2026-10-16
…
Loading…
How much this week’s corn rating is worth, by state. Tap a state
mostly noiseweakmeaningfulstrongno fit
Every R² on this page is an in-sample fit over each state’s own paired years (the count is shown as n or Years). A peak is the best of the weeks tested, so it is flattered by the choice.
36%
Iowa corn, last fitted week (wk 40)
of yield deviation from trend explained by G+E · n=22 yrs · not the same years as week 39 (22 vs 26), so the two are not like for like
64%
Iowa corn by week 39
the season seals it late · n=26 yrs
90%
Tennessee, wk 40
the clearest read in the country right now · n=22 yrs
n≥15
years per fit, minimum
thin fits refused
How the answer builds through the season: Iowa corn
gold outline = current week
Each bar: R² of that week’s G+E share vs final yield (detrended), across all years with data. Green ≥50% · gold ≥25% · grey = mostly noise.
Every state, corn: ranked by what this week’s number is worth
Click a state for its full-season curve; click a column to sort. Slope: bu/ac of yield deviation per 1-point G+E move, at that state’s peak week.
| State | This week’s R² | Peak R² (best week tested) | Slope @ peak | Years |
|---|---|---|---|---|
| 90% | 90% wk 40 · n=22 | +0.71 | 22 | |
| 89% | 90% wk 37 · n=26 | +1.03 | 22 | |
| 89% | 89% wk 39 · n=26 | +0.83 | 22 | |
| 84% | 85% wk 35 · n=26 | +0.77 | 22 | |
| 77% | 80% wk 38 · n=26 | +0.90 | 22 | |
| 76% | 81% wk 34 · n=26 | +0.76 | 22 | |
| 76% | 76% wk 40 · n=22 | +0.69 | 22 | |
| 73% | 79% wk 38 · n=26 | +0.71 | 22 | |
| 72% | 76% wk 39 · n=26 | +0.89 | 22 | |
| 70% | 76% wk 39 · n=26 | +0.86 | 22 | |
| 68% | 76% wk 37 · n=26 | +0.54 | 22 | |
| 68% | 73% wk 39 · n=26 | +0.60 | 22 | |
| 62% | 71% wk 38 · n=26 | +0.69 | 22 | |
| 51% | 53% wk 39 · n=26 | +0.51 | 22 | |
| 42% | 44% wk 39 · n=26 | +0.55 | 22 | |
| 41% | 48% wk 39 · n=26 | +0.45 | 22 | |
| 36% | 64% wk 39 · n=26 | +0.66 | 22 | |
| 11% | 21% wk 25 · n=26 | +0.27 | 22 |
What this can’t tell you. R² is explanation, not prediction, and a late-season
rating "predicting" yield is partly just the season being over. Fits are against detrended yield (state
yields drift up ~2 bu/ac/yr on genetics alone; without detrending, that drift fakes signal; the raw fit is
kept in the data for anyone who wants it). Weeks with under 15 paired years are left blank rather than thinly fitted.
And 26 years is still a small sample: treat differences of a few points as noise.
Do something with it: see how this week’s rating
ranks against every year since 2000 on Crop conditions rank · check how much of the
story the market already owns on What’s Priced In · turn a yield guess
into a plan with the yield estimator and break-even.
Questions people actually ask
Do USDA crop condition ratings actually predict final yield?
Partially, late, and it depends where you farm. In week 40 (the last week with a fit; week 41 has under 15 paired years), the Good+Excellent share explains about 36% of how Iowa's final corn yield differs from its trend (22 years). Iowa's fit peaks at week 39, at about 64% (26 years). The two weeks rest on different years (22 vs 26), so the gap between them is partly a change of sample, not only of signal. In Tennessee, week 40's ratings explained 90% (22 years). This page computes the number for every state and week instead of asserting it.
Why do the ratings explain so little in early summer?
Because the yield isn't made yet. July corn in Iowa can rate 80% Good+Excellent and then hit August heat during grain fill. Ratings measure how the crop looks; yield is decided by what happens next. The R-squared rises through the season because each week more of the yield is already made.
What does the slope number mean?
It's the historical relationship: how many bushels per acre of yield deviation (from that state's long-term trend) has come with each 1-point move in Good+Excellent share, at that week. Illinois at week 39: about +0.86 bu/ac per G+E point.
Why detrend the yields?
State yields rise about two bushels per acre per year from genetics and agronomy alone. Without removing that trend, the fit confuses 'years got later' with 'ratings got better' and reports fake signal. We fit ratings against each year's deviation from trend; the raw fit is also kept in the data.
So should I trade off Monday's crop conditions number?
This page shows exactly how much that number has historically been worth in your state at this week, often less than the market reaction implies. When the market moves more than that, it is reacting to more than the rating. See what's already priced on the What's Priced In page.
Source: USDA NASS weekly crop condition (Good+Excellent share) and annual state yield (final YEAR figures only), 2000–present. Fit recomputed weekly in season. Data refreshed 2026-10-16. AGSIST is free and sells nothing on this page.
Made by Sigurd Lindquist, CCA, Chetek, Wisconsin.