GSX Weekly Commodity Comment

Key Takeaways:

This week’s market tone has shifted from reacting to a fresh geopolitical shock to testing how long the current risk premium can last. The Strait of Hormuz remains the main macro driver, with crude oil still elevated enough to keep fuel, freight, insurance, fertilizer and synthetic fibre costs firmly in the soft commodity conversation. But the picture is becoming more two-sided: the same energy shock that is supporting input costs is now starting to pressure demand, refinery margins and broader confidence.

For cotton and coffee, markets remain caught between cost-led support and demand-side caution. Cotton still has a more constructive forward setup, supported by tighter balance sheets, West Texas moisture risk and the possibility of fibre substitution away from polyester. However, export demand for old crop has been disappointing and contributing to the market going limit down last Thursday...

Coffee has been trending lower as the basis levels on commercial grades have softened, however it must be noted that the inventories at destination continue to decline which is one constructive signal for flat prices and spreads

The key question for the week ahead is whether the Hormuz premium lasts long enough to trigger another round of inflation across freight, energy and consumer prices, or whether improving flows and demand rationing begin to cap the rally. It also appears the bond market has reached a point of fatigue and yields across the G7 have been rallying due to the lack of a clear resolution to the Iran conflict. This may just force the U.S. Administration’s hand much as it did during the “Liberation Day” meltdown in April 2025.

Cotton has moved from breakout to retest. July 2026 cotton closed at 82.33 c/lb on 19 May, down 137 points on the day, while December 2026 closed at 83.16 c/lb and March 2027 at 83.90 c/lb. The important change from last week is that July has given back the upper-80s move and is now back near the old 82 to 84 c/lb breakout area. That does not invalidate the broader move, but it does mean the market is now testing whether the breakout can hold rather than simply extending higher.

The rally is still not trading as a simple crude-oil proxy. On 19 May, crude was only marginally higher at US$103.59, while cotton corrected by as much as 137 points across the board. That shows the market is now balancing energy-led support against a firmer dollar, profit-taking, crop progress and the need for demand confirmation. Energy and synthetic-fibre costs remain part of the bull case, but cotton now needs support from exports, weather or new fund buying to regain momentum.

Spot market indicators remain firm, even though futures have pulled back. USDA’s Weekly Cotton Market Review showed seven-market spot quotations averaging 81.12 c/lb for the week ending 14 May, up from 79.04 c/lb the prior week and 63.21 c/lb a year earlier. Daily quotations reached a season high of 82.98 c/lb on 11 May before easing to 79.15 c/lb on 14 May. Spot transactions slipped slightly to 17,387 bales, compared with 18,774 the prior week, while the ICE July settlement ended the week at 83.94 c/lb versus 83.00 c/lb the previous week.

The May WASDE remains supportive, but a portion of that support has already been priced. USDA’s May cotton outlook projects 2026/27 global consumption at 121.7 million bales, a six-year high, while production is forecast to fall to 116.0 million bales and ending stocks to decline to 71.8 million bales. That is a constructive balance-sheet setup, but USDA also notes the two-sided demand risk: higher oil prices can support cotton against polyester, while higher inflation and higher cotton prices can also moderate textile and apparel demand.

The macro cotton chart depicts a WXY correction beginning in May 2022 and completing in February 2026, at the 100% Fibonacci extension near $60. From that base a strong impulsive advance developed. Pinpointing the market's current position within the structure off that low is complicated by the sharpness of the advance, +47% in 90 days. However this is where zooming in to the tactical chart provides additional context.

The rally from the February low appears to have terminated, with our standard Fibonacci extension target (red box) of $85–90 acquired this week. Further extension to the next Fibonacci level above cannot be excluded, however this looks less and less likely after the strong rejection we have since since striking $88. Wave (v) itself has subdivided cleanly into a full EW count, adding conviction that a top has now been reached. A break below $82 provides initial confirmation of the newly bearish outlook; with conviction increasing upon losing $77 – where wave (iv) terminated. From there we look toward Fib retracement levels (in orange) for support, particularly the 61.8% retracement at $70.

The positioning story has become more extended, but also more fragile. The latest CFTC futures-and-options data, for 12 May and released on 15 May, showed managed money holding 83,901 long contracts and 24,331 short contracts, leaving a net long of 59,570 contracts. That is up 8,386 contracts on the week and takes fund length beyond last week’s already aggressive level.

Unlike last week, this was almost entirely fresh length rather than short covering. Managed money added 8,781 longs, while shorts actually increased slightly by 395 contracts. That is still supportive while prices hold the higher range, but it raises liquidation risk if the market cannot rebuild upside momentum. Cotton is now clearly a fund-length market, so the rally increasingly needs confirmation from West Texas weather, stronger export demand or another leg higher in crude-linked fibre substitution.

Commercial positioning remains the caution point, although the pressure did not intensify in the same way as the prior week. Producer, merchant, processor and user accounts were 82,163 contracts long and 220,638 contracts short, leaving them net short 138,475 contracts. Their outright short position increased by 3,453 contracts, but longs also rose by 3,724 contracts, so the net commercial short was broadly steady rather than materially larger. That still shows heavy producer and merchant hedging into the rally, but it is not a fresh escalation in commercial selling.

The key takeaway is that the fund bid is still there, but the market is no longer under-owned. If weather risk in Texas persists and exports improve, the length can keep working in cotton’s favour. If rainfall improves or export demand stays only average, the size of the managed-money long becomes a downside risk, especially now that futures have pulled back toward the old breakout area.

The May WASDE is still the latest official balance sheet, so the balance-sheet story is largely unchanged from last week: supportive, but not bullish enough by itself to carry cotton without help from weather or demand. USDA’s May report is the first official 2026/27 forecast, and it notes that the new-crop projections are still tentative because spring planting is still underway.

For old crop, USDA puts the 2025/26 U.S. cotton balance at production of 13.90 million bales, exports of 12.00 million bales, domestic use of 1.60 million bales and ending stocks of 4.40 million bales. That keeps the old-crop stocks-to-use ratio heavy at roughly 32%. For 2026/27, USDA projects planted area at 9.64 million acres, harvested area at 7.38 million acres, yield at 866 lb/acre, production at 13.30 million bales, exports at 12.30 million bales and ending stocks at 3.90 million bales. That pulls stocks-to-use down to roughly 28%, which is constructive but still leaves weather and abandonment risk as the main upside trigger.

The global balance is tighter. USDA projects 2026/27 world production at 116.04 million bales, consumption at 121.69 million bales and ending stocks at 71.84 million bales, down from 77.27 million bales in 2025/26. That tightening is supportive, but Brazil remains the key competitive headwind: USDA has Brazil’s 2025/26 crop at 19.50 million bales with exports of 14.70 million bales, then projects 2026/27 exports at a record 15.00 million bales despite lower production of 17.50 million bales. Brazil and the U.S. together are projected to account for about 63% of world cotton exports in 2026/27, so U.S. export upside will still depend heavily on basis competitiveness and broader demand growth

Planting pace has improved and is still not the immediate problem. NASS reported the U.S. cotton crop 41% planted as of 17 May, up from 29% the prior week and slightly ahead of the five-year average of 40%. Texas is exactly on pace at 34% planted versus a 34% average. Georgia is a little behind at 38% versus 41%, and Louisiana remains behind at 50% versus 62%, although that is a large recovery from 25% the prior week. Mississippi, Arkansas and Tennessee are running ahead of average at 58%, 66% and 73%, respectively.

The issue is still establishment and abandonment risk, especially in Texas and the broader dryland Southwest. NASS now shows Texas topsoil moisture at 61% short or very short, up from 54% in last week’s note, while subsoil moisture is 63% short or very short, up from 60%. That means planting progress looks acceptable on paper, but the crop is still going into a moisture-stressed profile.

Drought data are a little more mixed. The latest Texas data, valid 12 May, show 65.3% of the state in drought, including 29.1% in severe drought, 13.5% in extreme drought and 1.1% in exceptional drought. That is a modest improvement in total drought coverage versus last week, but the Drought Monitor summary still says exceptional drought was introduced from the Texas Panhandle into northwest Oklahoma and that extreme drought expanded in northern Texas and western Oklahoma. In other words, statewide drought coverage has improved, but the key cotton-production risk area is not out of trouble.

USDA’s May harvested-area assumption already builds in some abandonment risk: 9.64 million planted acres and 7.38 million harvested acres imply national abandonment of roughly 23.5%, and USDA says harvested area is based on regional 10-year average abandonment with the Southwest adjusted for moisture conditions. That is manageable, but if West Texas dryland conditions fail to improve, the market will keep a weather premium in place.

U.S. Cotton Export demand turned less supportive in the latest report. USDA/AMS reported current-crop upland net sales of just 47,700 RB for the week, a marketing-year low, down 61% from the previous week and 66% below the prior four-week average. Upland exports were better than sales but still softer at 290,300 RB, down 11% from the prior week and 12% below the four-week average. The main shipment destinations were Vietnam, Turkey, Bangladesh, China and Pakistan.

New-crop upland sales also slowed to 29,700 RB, mainly to Vietnam, Mexico, Bangladesh, Guatemala and Indonesia, partly offset by reductions for Pakistan. Pima was also softer: current-crop net sales were 9,300 RB, down 19% week on week and 51% below the four-week average, while Pima exports were 12,100 RB, down 18% from the prior week but still 12% above the four-week average.

In 480-lb bale terms, the National Cotton Council puts weekly sales at 58,700 bales and shipments at 311,500 bales for the week ending 7 May. Cumulative 2025/26 sales are now 11.651 million bales, slightly behind last year’s 11.811 million at the same point, while cumulative shipments are 8.563 million bales versus 8.721 million last year. New-crop cumulative sales are 1.485 million bales.

The broader picture is still mixed. Shipments are good enough to keep USDA’s 12.0 million bale old-crop export target alive, but sales have not yet confirmed a stronger demand trend. Cumulative sales are also slightly behind last year’s 11.811 million bales at the same point. We are therefore keeping our full-year U.S. export forecast around 11.5 to 11.6 million bales for now, with upside risk if shipments stay above 300,000 bales per week and net sales recover.

The broader export picture is therefore still mixed. Shipments above 300,000 bales are good enough to keep USDA’s 12.0 million bale old-crop export target alive, but sales have not confirmed a stronger demand trend. I would keep the full-year U.S. export forecast around 11.5 million bales for now, with upside if shipments remain above 300,000 bales per week and net sales recover. Vietnam remains the anchor buyer, accounting for 33% of MY2025 commitments, followed by Pakistan at 12%, Turkey at 9%, and Bangladesh and India at 7% each.

The weather signal is wetter than last week, but it is not cleanly bearish for cotton because the rainfall focus is uneven. WPC’s latest Day 2 Excessive Rainfall Outlook, valid 20-21 May, has a Slight Risk for the Edwards Plateau region of southwest Texas, with the risk area shifted slightly south and guidance showing more than a 50% probability of rainfall exceeding 5 inches in 24 hours inside the Slight Risk area. WPC also trimmed the Marginal Risk area in the Southern Plains and Central High Plains because of the southward trend. That means the rainfall could pressure weather premium if it reaches the right cotton areas, but it may not fully solve West Texas High Plains abandonment risk.

The longer-range outlook also leans wet. CPC’s Week-2 Hazards Outlook, valid 27 May through 2 June, keeps a moderate risk of heavy precipitation for the Southern Plains and Lower Mississippi Valley on 27 May, a slight risk from the Central and Southern Plains through the Southeast for 27-29 May, and flooding possible for much of eastern Texas plus parts of Arkansas and Louisiana. This pattern could improve drought conditions if rain reaches dry cotton areas, but the highest flooding risk appears more eastern Texas and Lower Mississippi Valley than the core dryland West Texas production zone.

Coffee remains two-sided, but the tone has shifted heavier. The latest settled close on Tuesday, May 19, looked like a short-covering bounce rather than a bullish reset: July arabica closed at 270.15 c/lb, up 5.95 c/lb, and July robusta at $3,345/t, up $39/t, after arabica briefly hit a 1.5-year nearest-futures low and robusta a one-month low.

The front-end inventory story is still supportive. ICE arabica certified stocks fell to 458,735 bags, a 2.75-month low, while ICE robusta stocks dropped to 3,631 lots, a two-year low. That keeps prompt cover and nearby spreads sensitive to any further stock draw.

The problem is that the broader supply tape has turned more bearish. Sucafina’s May 20 report said July arabica had collapsed 18.1 c/lb to $2.64/lb by Monday’s close after stops were triggered near 265 c/lb. It also flagged stronger origin selling, better roaster cover, and weaker U.S. demand, with net imports plus inventory change down 1.9% on a 12-month basis through March.

Brazil is the main pressure point. Cecafé said April exports reached 3.122 million bags, up 0.6% y/y, but revenue fell 17.7%. Arabica availability is still tighter, but conilon/robusta flow is improving sharply, with April canephora exports up 374% y/y. Rabobank now pegs Brazil’s 2026/27 crop at 73.3 million bags, while StoneX is at 75.3 million and Marex at 75.9 million. The nuance is that Espírito Santo robusta may be slightly smaller and better quality, but Conab still sees national canephora at 22.1 million bags, up 6.4%.

Vietnam is the clearer bearish signal for robusta: Jan-Apr exports rose 15.8% y/y to 810,000 tonnes, or about 13.5 million bags, while export value fell 7%. Colombia remains a washed-arabica support, but April was less severe: production was down only 1% y/y to 697,000 bags, while exports fell 15% to 682,000 bags.

Weather and freight are now volatility overlays, not the base case. Brazil harvest weather has mostly improved, although INMET flags locally intense rain in Zona da Mata, northern Rio de Janeiro, and southern Espírito Santo. Net-net: certified stocks can still squeeze the front end, but rallies now face heavier resistance from Brazil selling, Vietnam flow, better roaster cover, and a 2026/27 surplus narrative.

ICE front month:

The multi-year advance in KC Arabica from the 2019 lows, does not display the internal characteristics of a five-wave Elliott Wave impulse. The price action is more consistent with a three-wave pattern – a (B)-wave.

Under this scenario, the entire move from 2019 is labelled A-B-C, the sub-waves of a (B)-wave rally. This kind of structure up, portends a deep (C)-wave down. The standard Fibonacci target for this decline at ~126 becomes support, however low conviction until tactical analysis aligns.

Within the bigger trend outlined above, the April spike to Fibonacci resistance above 300 offered a clean entry point for shorts. That level has since given way and downside momentum has resumed. We are tracking completion of wave v of (iii), targeting roughly the 250–260 region. Following the breach of the 200-day moving average in February, price has traded below that average without recovery. The degree of separation is becoming extended — a factor that warrants tactical awareness given the potential for mean-reversion.

The positioning picture has turned less supportive in arabica. The latest CFTC futures-and-options data for 12 May showed managed money in Coffee C holding 44,974 long contracts and 19,946 short contracts, leaving funds net long 25,028 contracts. That is down by about 6,007 contracts from the prior week’s 31,035-contract net long. The mix matters: longs were cut by 1,316 contracts, while shorts increased by 4,691 contracts, so this was not just passive length reduction. Funds are still leaning bullish, but the fund bid has clearly weakened.

Commercial positioning is less cautious than last week. Producer, merchant, processor and user accounts were 47,372 contracts long and 71,717 short, leaving them net short 24,345 contracts. That compares with a net short of 30,898 contracts the prior week, so the commercial net short narrowed by roughly 6,553 contracts. Longs increased by 5,491 contracts and shorts fell by 1,062, it is likely that roasters were active buyers on the dip below 280 c/lb through the COT reporting date, while origin selling remained limited. That is an important change from last week’s report.

The caveat is that the COT report is already stale. July arabica had fallen nearly another 15 c/lb since the 12 May reporting date, and so Monday marked the fifth consecutive lower close in New York and the third consecutive lower close in London. July arabica briefly traded below 265 c/lb for the first time since 4 August, while robusta briefly fell below $3,300/t after having traded as high as $3,599/t on 13 May. That means actual arabica fund length may already be lower than the CFTC snapshot suggests, or the short side may be larger.

Robusta positioning has turned more supportive on paper. The latest robusta COT shows managed money gross length rising to 30,976 lots, while shorts covered 3,209 lots during the week. That lifted the managed-money net long by about 4,784 lots to 13,789 lots. This reverses last week’s fading-fund story, but it is not a clean physical-demand confirmation: there has likely been heavy origin selling into the rally from $3,400/t to $3,500/t, while roaster buying remained largely absent.

Net-net, arabica still has fund support, but the positioning cushion has thinned. It is no longer an overcrowded 31,000-contract net-long market, and the larger short component could fuel short covering if nearby tightness bites. But funds are still net long more than 25,000 contracts, so a failure to reclaim 280/285 c/lb, and eventually 300 c/lb, leaves arabica vulnerable to further liquidation if Brazil harvest weather stays comfortable, the real weakens, or new-crop flow confirms the large-crop forecasts. Robusta is physically tighter and now positionally more constructive, but the renewed fund net long also means rallies need follow-through from spreads or physical demand to avoid becoming another sellable bounce.