Tearing apart the soybean demand stack

Soybeans.

Bloomberg is running "China buys soybeans as trade thaw continues." CNBC has a panelist explaining how the tariff removal is "unlocking pent-up demand." Both headlines are technically true. Both miss the machine entirely.

Soybeans pushed above $12 a bushel on Wednesday — a seven-week high — and the press hung it on a single hook: China is buying again. That's one valve. There are three. And the other two are louder.

The old trade was simple: China buys, price goes up. I ran that playbook myself for years and it worked fine until 2018 blew it apart. What's different now is the demand base has structurally widened underneath everyone. Renewable diesel mandates are pulling soybean oil at rates nobody modeled five years ago. And a heat dome is rolling into the western Midwest right as 34% of the crop hits bloom.

Three demand channels stacking on top of each other. One supply model that hasn't been updated since May. That gap is the whole story.

Story off. Numbers on.

Wednesday, USDA confirmed the sale of 472,000 metric tons of soybeans to China — 136,000 MT for the current marketing year, 336,000 MT for 2026/27. That's the biggest single-day export sale to China since November 2025. State-owned Cofco booked at least six cargoes for September–October loading, then came back overnight and booked five more.

That crush number is the one nobody on the equity side is watching. EPA finalized the Renewable Fuel Standard in March: 9.07 billion RINs of biomass-based diesel required in 2026, jumping to 9.20 billion RINs in 2027. Soybean oil is the primary domestic feedstock. Last year's crush already set a record at 2.61 billion bushels. This year's projection — 2.75 billion — would shatter it.

Meanwhile, China agreed to purchase at least 25 million metric tons of U.S. soybeans annually through 2028. The 10% reciprocal tariff is being removed. That's not a headline — it's a demand valve that was closed for most of 2025, now fully open and flowing.

USDA projects 2026/27 ending stocks at 310 million bushels, down from 340 million in the current year. That's already below the 10-year average. And they built that number on 53 bushels-per-acre trend yield — a figure they haven't touched since May.

That balance sheet is built on a world where all three demand channels pull simultaneously and the supply side delivers a perfect crop. The crop is not perfect. Not this week.

The USDA's 53 bushels-per-acre yield estimate has not moved since the May WASDE. They held it in June. Analysts expected them to hold it again in yesterday's July 10 report — and they did. That number won't get a weather-based adjustment until August at the earliest.

Here's the problem. A heat dome is pushing 95 to 105°F across the northern Plains and Upper Midwest starting Sunday. The extended forecast stays hot and dry through July 21, squarely over the western Corn Belt and Plains. Soybeans are 34% into bloom. Nine percent are already setting pods. This is the pollination window — the two weeks where heat stress does irreversible damage.

Good-to-excellent ratings already slipped a point last week to 65%. That's not alarming on its own. What's alarming is the direction — and the fact that the most damaging weather hasn't arrived yet. It starts tomorrow.

I've been on the wrong side of a weather trade before. 2012, specifically — I got short beans early that June because condition ratings looked adequate and trend yield seemed safe. Two weeks of heat turned "adequate" into a major yield miss that sent beans above $17. Trend yield is an assumption until the thermometer makes it fiction. The USDA doesn't guess ahead of the data. Which means the model stays wrong until August, and by then, the bushels are already gone.

Managed money is net long only 2.88% of open interest on soybeans as of June 30. That's practically flat. It means the speculative community hasn't built the weather trade yet. They're reading the USDA's comfortable numbers. They haven't checked the 10-day forecast.

Run the math. USDA is projecting a 4.435 billion bushel crop at 53 bpa. If heat and dryness shave that to 51 bpa — not a disaster, just a bad two weeks — you lose roughly 170 million bushels of production. Ending stocks don't drop from 310 million to 140 million gracefully. That's a stocks-to-use ratio that makes buyers stop negotiating price and start negotiating delivery dates.

The sequencing matters. First: weather premium builds as the next two Crop Progress reports show condition deterioration. Managed money — sitting at a nearly flat 2.88% net long — starts adding. They have enormous room to run. Second: the August 12 WASDE forces USDA's hand on yield. If they cut even a bushel per acre, the ending stocks math goes from comfortable to critical, and every trader with a calculator sees it simultaneously.

Third — and this is the wave that gets loud — China's 25-million-ton annual commitment collides with a shrinking U.S. exportable surplus. Crushers, already locked into record biodiesel volumes by the RFS mandate, can't yield bushels to the export market. Export and crush compete for the same pile. When the pile shrinks, both bid harder.

Where does the capital go? Not into the broad ag ETFs — those are diversified across livestock, wheat, and corn, and they'll dilute the soybean move. The edge, if there is one, is in new-crop November bean futures, which already pushed above $12 but analysts see $12.40 to $12.60 as the next range and $13-plus if the August yield cut materializes. Crush-margin positions — long beans, short products — benefit directly from the tightening. And producers sitting on unpriced 2026 inventory are watching the best pricing window they've had since spring.

The mainstream narrative says China tariffs drove this rally. The machine says something bigger: three demand channels pulling simultaneously on a supply estimate that's running a month behind the weather. In my experience, when the model and the sky disagree, the sky wins. It just takes until August for USDA to admit it.