In the past three weeks, three major price events have reshaped positions across agricultural & soft commodity markets. Each unfolded faster than the official reporting cycle could track. Each was physically observable before it was officially confirmed.

Cocoa: the collapse official surveys missed

In January 2025, cocoa traded above $12,900 per metric tonne on ICE New York – a record that reflected two consecutive years of drought-driven harvest failure across West Africa and one of the largest supply deficits on record: 494,000 tonnes short in a single season. By early March 2026, that same contract was trading below $3,900 – a 48% decline that returned prices to October 2023 lows.

The swing was not invisible in advance. Production area recovery in Côte d'Ivoire and Ghana – visible in vegetation health indices and rainfall anomaly data weeks before any official crop survey – pointed to an improving supply environment months before the market fully repriced. Ghana's farmgate price was cut 29% on February 12, confirming what the ground signal had already been showing. Official harvest estimates arrived after the fact.

US wheat: a geopolitical shock through the fertiliser supply chain

On February 28, US-led strikes on Iranian infrastructure triggered a blockade of the Strait of Hormuz. Within seven days, US Gulf Coast urea prices had risen 32% – from $516 to $683 per tonne. CBOT May wheat futures surged 33 cents to approximately $6.35 per bushel, the highest level since the height of the 2022 energy crisis. Nearly one million metric tonnes of fertiliser cargo was stranded in the Gulf within a fortnight.

The mechanism is direct. One third of global seaborne fertiliser moves through the Strait of Hormuz, including 44–50% of global urea exports. Spring planting windows for US wheat and corn – March and April – require inputs to be in position now. Urea cost for US farmers reached the equivalent of 126 bushels of corn per tonne in early March, up from 75 bushels in December. The planting decision is being made under conditions that no seasonal forecast, no USDA projection, and no official data product anticipated.

Vietnam robusta: harvest progress outpaces the market

Vietnam's 2025/26 robusta harvest is tracking toward 29.4 million bags — a 6% year-on-year increase and a four-year high. On March 15, ICE London robusta fell $170 per tonne in a single session, touching a seven-month low. The production recovery in Vietnam's Central Highlands – disrupted the prior season by drought and Typhoon Kalmaegi – had been progressing in satellite-derived vegetation data and harvest throughput signals for weeks before that single-day move materialised.

The pattern holds across all three stories: the physical signal moved before the official signal.

The gap financial markets tolerate

The World Bank described the current period as marked by "the highest level of commodity price volatility in at least half a century." Oxford Economics characterised 2026 as a year of "widening dispersion" – sharp divergence in performance where sector-specific fundamentals, not a single macroeconomic narrative, dictate outcomes.

This is not unusual. Agricultural & soft commodity markets have always been subject to weather, geopolitics, biology, and land-use dynamics that official reporting cannot anticipate. What is unusual is that the gap between what is physically observable and what is officially reported remains as wide as it was decades ago – in markets where positions, procurement decisions, and capital allocation turn on information speed.

Jonathan Horn, Treefera's founder and CEO, spent years as Head of Risk at J.P. Morgan. The data gap he observed was not abstract. Trading desks and risk committees managing significant exposure to commodity-sensitive assets were making decisions from data that arrived months after the conditions it described. Nature-based assets were not being treated with the same rigour as other asset classes. The information infrastructure hadn't kept pace.

What changed – and what has accelerated rapidly – is the convergence of satellite observation, AI-native processing, deep scientific modelling, and financial-grade risk analytics. The signals that were always there, embedded in vegetation health indices, soil moisture readings, trade flow data, and atmospheric anomalies, can now be processed at field resolution and delivered at the speed markets require.

Seeing everything, everywhere, all at once – and making sense of it – is no longer hypothetical.

What first-mile intelligence adds

Each of these stories carries a first-mile signal that was readable before the market repriced it.

For Ghanaian cocoa, Treefera's production area model maps 2.7 million hectares of detected cocoa at pixel level, and computes area-weighted Harmattan stress scores against the specific districts where cocoa actually grows – not regional administrative averages that dilute the signal by including non-cocoa land. Across the 2024/25 season, those scores were severe: Ahafo at 0.934, Ashanti at 0.896, Bono at 0.947, Western North at 0.881, with approximately 58 consecutive dry days across the belt. By January–February 2025, the 2025/26 picture was already forming in the same data – scores had dropped to 0.303, 0.127, 0.126, and 0.124 respectively, with belt-wide consecutive dry days falling to approximately 19. The supply recovery that eventually drove prices from above $10,000 to below $4,000 was visible in production-weighted stress data while the market was still in uncertainty mode, futures still above $10,000.

For US wheat, Treefera's yield model operates at weekly cadence with county-level resolution, applying stress signals against the crop's biological development stage. A rainfall deficit in January carries a different yield consequence than the same deficit during heading – the model knows the difference. As the Hormuz fertiliser shock works through to planting decisions, the crop's in-season stress profile is precisely the signal USDA's monthly WASDE cycle is structurally too slow to capture.

For Vietnam robusta, Treefera's production area model maps coffee cultivation at field resolution across the Central Highlands, the region responsible for the majority of Vietnam's output. The harvest recovery that drove the March 15 single-day move of $170 per tonne was physically progressing in vegetation data for weeks before it reached the price.

Coverage spans corn, wheat, coffee, cocoa, sugar and cattle, updated weekly, with monthly written insights translating field-level signals into portfolio-level context. Official data confirms what the first mile already showed.

Starting next week

Treefera's bi-weekly Market Intelligence newsletter launches on Wednesday 25th March. Each issue brings first-mile signals to the commodity stories making news – not commentary after the fact, but the intelligence layer that shows what the data looked like before the market moved.

Sign up here