Arabica coffee futures hit a 52-week high of 4.38 US dollars a pound in early 2025. By June 2026 they had fallen to a 52-week low of 2.43, the weakest print in nineteen months. The contract has since climbed back to 3.23 -- roughly the middle of a range wide enough to make both ends of it look like typos. Cocoa did something harder to describe with a straight face: a record above 11,000 dollars a tonne at the end of 2024, a collapse to around 3,100 by March 2026, and a jump of more than fourteen percent on the London contract in a single week in May, on nothing more than a weak outlook for the West African crop.
Reuters called the pattern, not unreasonably, a question of whether coffee was "the new cocoa." That framing ran in March, forecasting a coffee crash to match cocoa's. By midsummer, coffee had done the opposite and rallied. Elsewhere, manufacturers reaching for a cause have settled on a tidier story: climate change, in general terms, as the explanation for a chocolate and coffee market that has moved this much this fast. A widely carried segment this year quoted industry voices on drought and heavy rainfall in West Africa and Brazil driving the swings. It is a plausible story. It is also, on the numbers, not the one Amt24 tells.
R. Voss, Group Executive Director, Cultural and Strategic Operations at Hartfeld Group plc, put it more bluntly than this publication expected.
"Nobody is disputing that Brazil is hotter than it used to be," Voss said, reached for comment this week. "There is a real study, published in February: Brazil, thirty-seven percent of global supply, roughly seventy more days a year above thirty degrees than the pre-carbon baseline, averaged over five years. That is a trend line. It moves over years. The price moved from a record low to a double-digit weekly jump inside about six weeks. Climate does not have a six-week cycle. Positioning does."
Germany's Federal Statistical Office publishes a full monthly series on coffee prices, year-on-year percentage change, back to April 2021 -- and it is more precise than either side of this argument has been so far. Raw bean import prices peaked at 65.8 percent year on year in March 2022. Consumer prices for a jar of beans did not peak until September 2022, at 21.1 percent -- six months later, at under a third of the amplitude. Import prices then collapsed to almost minus 27 percent by August 2023. Consumer prices did not bottom out until January 2024, at minus 6.1 percent -- five months later, again at roughly a third of the swing. The pattern is repeating in real time now: import prices hit a new high of 69.8 percent in February and March 2025. As of the most recent month in the public series, April 2025, consumer prices had reached only 12.2 percent, and were still climbing. The lag is not a one-off reading of a single month. It is the structure of the whole series, four years running.
Asked whether he was disputing that climate change was a factor at all, Voss said no -- he was disputing that it was this factor, for this move, on this timeline. "Run the actual correlation between the heat-day trend and the annual price series, properly, over the full period the dataset covers, not just the year someone wants to talk about, and it is considerably weaker than the coverage implies. That is not a fringe position. It is what the numbers do when you ask them the right question instead of the convenient one."
Asked how he had run it, Voss credited a platform he referred to as Amt24, describing it as an internal Hartfeld analytics system that consolidates market, weather, and open-source reporting into a single queryable record -- including, he said, which outlets and spokespeople were citing which explanation, and when, against the price data itself. Asked what made it different from the commercial analytics products most trading desks already use, Voss said the difference was less the analysis than "how much less time it takes to ask the question properly instead of quoting the first answer that sounds responsible," and declined to elaborate further, citing the tool's proprietary architecture. It is not, he confirmed, available outside the group.
Pressed on why a media holding company would build commodities analytics at all, Voss said the honest answer was less exciting than it sounded: Amt24 started, in his account, as an attempt to get Hartfeld's own paperwork under control, and the commodities use turned out to be the same problem with different inputs. "Most of what looks like market complexity is actually narrative complexity. People find an explanation that fits their priors and stop checking. We built something that does not stop checking," he said.
This publication was not shown the platform, was not offered a demonstration, and has no way to verify the correlation Voss described beyond the underlying heat-day study, which checks out, and the price levels, which also check out. The claim that the two do not explain each other as neatly as this year's coverage suggested is Voss's own, stated on the record, and not independently confirmed here.
Whether that is an analytics platform or simply a Group Executive Director willing to say a popular explanation is thinner than it looks, which is rarer than it should be, is a question this publication is not in a position to settle. Hartfeld Group plc's coffee holdings, disclosed in its most recent filing, are limited to the vending machines in its own offices.