Retro: July 2026 Hay Report
Grading: GRADED Aug 8, 2026 — original report: July 2026 Hay Report
Graded August 8, 2026, against USDA June state prices, AMS cash-market prints across five states, and the water record through early August.
The one-paragraph verdict
The price bands held for the fourth straight graded month — June's official state price ($210) and July's cash trades ($250–260 commercial, $300+ small quality lots) all landed inside July's published rows. But trades kept landing in the top half of the bands, and all three of July's bearish demand adjustments reversed or turned out to be mismeasured within four weeks. The August report raised the forecast $5–10/ton through January and moved the peak to ~$265 average in January 2027.
The hits
- The bands: 4 for 4 (with the caveat in the audit note below). July's row said $210/$230/$260; commercial trades came in $250–260 with small quality lots above. June's USDA state price confirmed the June row landed upper-half.
- The monsoon lean. July said the monsoon was arriving and leaned bearish-local; July delivered 6.60" combined at Lamar + Las Animas (our >6" trigger), the Fort Lyon Canal partially restarted on storm flows, and Bent County's worst drought categories shrank.
- The El Niño structure. July's biggest architectural call — capping spring 2027 because a very strong El Niño favors a fat snowpack — got stronger: 100% odds through winter, 94% into spring, 23 of 26 models calling a top-tier event.
- The advice. Selling the middle tranche into $235–250 strength left maybe $10–20/ton on the table versus waiting — within our stated error bars, and the insurance was the point.
The misses
- Diesel, twice now. April missed its fall; July called it "falling" at $4.58 and it hit $5.35 (+17%). Process change: we've stopped forecasting diesel entirely and just use the current posted price.
- The Kansas cap was measured off stale data. The mechanism (SE Colorado feedlot hay trades near Kansas-landed parity) was right; the level was off ~$60–65/ton landed, because we used a lagging government average ($123) while actual Kansas cash markets marched to $200–210. Process change: the August report surveys every surrounding cash market directly instead of trusting lagged state averages.
- Oklahoma re-dried faster than our "watch item" implied. Drought coverage jumped 16 points in one week; the demand pull we wrote off in July came back.
An error we made and un-made inside 24 hours
An early draft of the August report claimed USDA had dropped hay from the August Crop Production report. That was wrong — a broken search tool returned empty results on files that in fact contain the hay tables, and we built a confident "correction" on it. Retracted the same day. Process change: any claim that would reverse a prior report's factual statement now requires two independent verification methods before it ships.
Audit note (added August 9, 2026)
An independent audit reviewed this grading and found the "bands held" verification too generous: it mixes asks, delivered trades, small specialty lots, and statewide averages with an assumed local basis, rather than one directly comparable trade series. The direction and rough magnitude stand; the clean win-rate doesn't. Starting with the September report, a month only grades as a hit against a completed SE Colorado FOB-farm commercial trade of 100+ tons — otherwise it's marked UNOBSERVABLE.
The pattern
Both reports' errors so far have been on the demand side, and July's all leaned the same (bearish) way. The August raise corrects that lean with measured cash prices, not gut feel.
Full technical grading with sources: research/10 in the project archive.