GSX Monthly Coffee Report - February 2026

The prevailing market view has been that, as New York transitioned from a relatively balanced 2025/26 crop year to an anticipated 2026/27 surplus of roughly 5%, prices would need to adjust lower.

It seems like a lifetime ago, but in early January, NY traded at $3.80 and encountered Central America’s willingness to sell in size. Even then, with CAM differentials sharply breaking, the market still managed a late January probe above $3.70. Then Brazilian weather forecasts normalized.

February, however, presented a markedly different landscape, a reminder that in commodities, timing is half the trade. A combination of speculative and index fund selling led to nearly 7 million bags of paper selling, erasing about 25% of New York’s value.

Since February was dominated by spec selling, let’s review the CIT shifts from 1/27 to 2/24:

Net commercial buying of +23.6k was 91% gross short liquidation. The roaster community took advantage of this sharp break to extend paper cover. Roaster pricing, given up to trade houses, would reduce gross shorts. This is imprecise, but assuming a linear relationship between selling size and price, 1k lots of spec/index selling corresponds to roughly a 4 cent drop.

The KK26 arb narrowed by about 45% of the NY decline. LDN futures/options disaggregated data showed Managed Money sold 14.7k lots, with 26% long liquidation and 74% new shorts, while PMPU net bought 14.5k, with 65% new longs and 35% short covering.

As expected, physical differentials firmed.

End-January certs increased from 435,721 bags to 477,229 bags (Honduras +55k, Nicaragua +44k), while Brazil certs declined by 47k bags, as firmer FOB differentials left their certs at a discount value. Pending gradingsat the end of February were 149.2k bags.

The KK26 paper arbitrage narrowed by 16 cents during February, yet Brazil’s internal farmgate arb remains wide. Conilon’s discount to low-grade arabicas tightened from R$600/bag to R$500 but remains about one-third cheaper in percentage terms. Robustas continue to underpin domestic consumption, which is estimated to be roughly 80 to 90% robusta-based (vs a normal 45-55% range).

“Paper” is the cheapest form of the commodity. Mainstream commercial physical grades trade at positive differentials, with the arguable exception of Conilon 5/6 13 up at -2 cts/lb FOB. No significant origin is anywhere near tenderable parity. Two stoppers at ABN AMRO, 500 each? Cert stock size is disproportionately larger.

Stocks at destination remain historically low. Brazilian farmers are very well capitalized, and shipments are declining. They are also viewed as the clear long in the market.

We’ve started to make that long-anticipated price transition from 25/26 balance to 26/27 surplus. Futures markets anticipate. They are the price discovery benchmark on the forward curve, but the lead time on that anticipation can vary significantly. It is straightforward to expect Brazil arabica 26/27 surplus pressure five to six months ahead, into the July/August 2026 forward period.

The key question is whether continued non-commercial and index selling will outweigh scale-down roaster buying, potentially firming differentials but eventually forcing Brazilian farmers to release remaining 2025/26 stocks.

Non-commercials have built their gross short to 28k lots, but that is still significantly less than the 60k gross short positioning seen in January 2023 or June 2020. KC2-KC7 is still around 6.5%. We still believe the forward annual inversion has to break toward par to trigger the heavy long-term systems adding perhaps another 30k lots of shorts. Undoubtedly, that additional systems-driven short selling will not be linear. We would continue to closely monitor KC2-KC7, expecting that every 1% of slippage in the annual inversion would trigger more selling.

Conversely, during the second half of February, NY saw 7 or 8 sessions in the 275/285 area, where it seemed to be attempting to build a base for a technical short-covering bounce. We need to determine where origin or trade reappears as good sellers. Considering the extent of the recent decline, that could set up a retracement back to the 300 to 320 area.

Global fundamentals point to a 5% surplus, equivalent to 8 to 10 million bags, primarily in arabica. The forward NY/London arbitrage does not appear to reflect that surplus.

The K27 NY/K27 LDN ratio is about 1.75, higher than the nearby K26 ratio. Since roughly 70% of OI is usually positioned in the two lead liquid months, we can reasonably assume this is due to new spec short positioning. However, on a decadal lookback, a NY/LDN ratio of 1.75 is relatively neutral. NY certainly is not cheap. The end-February K27 NY/K27 LDN arb of $1.13 may need a price adjustment to accommodate that projected arabica surplus.

It is logical that the first place to adjust consumption would be within the Brazilian domestic market. Every 10% shift in internal robusta/arabica usage is 2 million bags or more. However, trying to bridge that one-third discount would imply significant shifts.

The onset of conflict in the Middle East will shift market attention toward crude oil, freight, the U.S. dollar, and equities in the short term. Outcomes of geopolitical events are hard to forecast. However, in the longer term, one concern is that overall commodity supply chain risk has increased. Markets will be watching for any increased hoarding of strategic commodities. The relevance to coffee prices over the longer term would be any significant increase in general commodity-related investment.