The market sees futures. We see the physical market underneath.
Futures tell you what the market believes and official statistics tell you what governments counted, eighteen months late. Harmattan measures what is happening at origin now — who is being paid, which buyers have money, and where cocoa is piling up outside normal channels.
Today's readings
Three developments
Next season's projected surplus has all but disappeared
- What happened
- StoneX cut its 2026/27 global surplus from 267,000 t in January to 25,000 t in July while revising the current season up to 422,000 t.
- Why it matters
- The market spent the first half of the year positioned for abundance. A near-balanced next season with a strong El Niño overhead is a different market.
- Watch next
- Main-crop arrivals from October, and the ICCO bulletin in November. Supply at origin →
Ghana's crop estimate moved 130,000 tonnes in three weeks
- What happened
- COCOBOD's internal 2026/27 estimate reached 650,000 t on 20 August against public guidance of 450–550,000 t and StoneX at 586,000 t.
- Why it matters
- The spread between those three numbers is the single largest uncertainty in the balance, and none of them is observed yet.
- Watch next
- The farmgate price announcement and whether licensed buyers are funded when the crop opens. Physical flows →
European grinding is contracting; Asia and origin are absorbing some of it
- What happened
- ECA reported the weakest second quarter since 2020, −4.6% year on year, while Asian grindings rose 25% and Ivorian onshore processing 29%.
- Why it matters
- Demand destruction after three years of record prices is real, but it is moving geographically as much as shrinking.
- Watch next
- ECA Q3 on 15 October — a third consecutive contraction or a floor. Processing and demand →
Four readings from the field, taken the same way every wave
Purchasing Capacity
Can local buyers actually purchase cocoa?
Purchasing clerk funding status, funding recency, and volume turned away for lack of cash. A leading indicator of official arrivals that no public source reports.
- Latest reading
- Wave 1 pending
- Change since last wave
- —
- Coverage
- Ghana, then Côte d'Ivoire and Nigeria
- Publication
- On wave lock · method v0.1
Payment Stress
Are farmers being paid?
Delivery-to-payment duration, share unpaid, and chit usage — measured from transactions, not sentiment.
- Latest reading
- Wave 1 pending
- Change since last wave
- —
- Coverage
- Ghana, then Côte d'Ivoire and Nigeria
- Publication
- On wave lock · method v0.1
Realized Farmgate Spread
What price is actually clearing?
Cash a farmer has in hand against the official producer price, time-discounted for what is still owed. A price paid in forty days is not the official price.
- Latest reading
- Wave 1 pending
- Change since last wave
- —
- Coverage
- Ghana, then Côte d'Ivoire and Nigeria
- Publication
- On wave lock · method v0.1
Held Stock
Where is cocoa accumulating?
Beans held by farmers, in purchasing clerk sheds and at depots — with the stated reason, which is what separates normal inventory from a seizing market.
- Latest reading
- Wave 1 pending
- Change since last wave
- —
- Coverage
- Ghana, then Côte d'Ivoire and Nigeria
- Publication
- On wave lock · method v0.1
From farm to consumer
- 01
Farm
The tree, the pod, the six months from flower to bag.
- 02
Local buying
Purchasing clerks, licensed buyers, the cash that has to arrive first.
- 03
Exporter
Where price risk is taken and hedged on ICE.
- 04
Port
Arrivals, held stock, and the lanes into Europe.
- 05
Processor
Grindings by region — the standard demand proxy.
- 06
Consumer
Price, positioning, and what reaches the farmer.
Why origin data is different
Cocoa's information gap is not an oversight. It exists because the data that moves the price is not published: farmgate prices are administered rather than discovered, purchasing clerk funding is invisible, and cross-border leakage is inferred from price spreads because nobody counts it.
Plot every public cocoa source on two axes — how fresh it is, and how far upstream it sits — and one quadrant is empty. Everything fast is downstream of the farm. Everything upstream arrives in months. The upstream-and-fresh quadrant is what the panel fills.
Coverage, method and provenance
The panel launches in Ghana and extends to Côte d'Ivoire and Nigeria: roughly 400 respondents per wave across farmers, purchasing clerks, depot staff, input dealers and hauliers, the same respondents interviewed repeatedly so the series measures change rather than sampling noise. Weights, caps and thresholds are pre-registered before any data exists. Every public figure on the site carries its source, observation date and licence in the ledger, and the build refuses to publish anything the licence does not permit.
What we will not claim
These measure current physical conditions. They are not forecasts, and any lead-lag relationship to arrivals or price is a hypothesis we publish the data to let you test. The panel is purposive rather than a representative sample, and nothing is seasonally adjusted until three years of history exist.