GSX Weekly Commodity Comment

Key Takeaways:

The cotton and coffee markets remain caught between cost-led support and demand-side caution. Cotton still has the more constructive forward setup, with tighter U.S. and global balance sheets, ongoing West Texas weather risk and potential fibre substitution away from polyester. Export demand has improved from the previous week’s disappointment, but shipments remain soft, so demand is not yet strong enough to drive the market on its own.

Coffee remains under pressure as commercial-grade basis levels soften and the Brazil supply outlook improves. That said, destination inventories remain historically tight, which continues to provide some support for flat prices and nearby spreads, especially if Brazilian flows are delayed or weather risk returns.

The key question for the week ahead is whether the Hormuz premium lasts long enough to trigger another inflation impulse across freight, energy and consumer prices, or whether improving flows and demand rationing begin to cap the rally. Bond markets already look fatigued, with G7 yields rising as investors price in both inflation risk and the lack of a clear resolution to the Iran conflict. That pressure could again force the U.S. Administration’s hand, much as the bond-market reaction did during the “Liberation Day” tariff shock in April 2025.

Cotton has moved from breakout retest to deeper correction. The key change from last week is that July has not merely tested the old 82–84 c/lb breakout area; it has now broken below it. July 2026 cotton closed Friday, 22 May, at 77.42 c/lb, while December 2026 closed at 79.33 c/lb and March 2027 at 80.19 c/lb. Compared with the 19 May closes used last week, that leaves July down roughly 491 points, December down 383 points, and March down 371 points. On 26 May, the broader cotton benchmark was indicated around 78.22 c/lb, up on the day but still well below the mid-May upper-80s high and below the old breakout zone.

Spot indicators have also weakened materially. USDA’s Weekly Cotton Market Review showed seven-market spot quotations averaging 76.45 c/lb for the week ending 21 May, down 467 points from 81.12 c/lb the previous week, though still above 62.81 c/lb a year earlier. Daily spot quotations ranged from a high of 78.91 c/lb on 18 May to a low of 73.19 c/lb on 21 May. The most striking change was liquidity: spot transactions fell to just 6 bales, compared with 17,387 bales the prior week. ICE July ended the USDA review week at 77.98 c/lb, down from 83.94 c/lb the week before. That means the cash market is now confirming near-term futures weakness rather than resisting it, although season-to-date spot transactions remain well ahead of last year at 1.49 million bales versus 957,062 bales.

Conviction in the medium-term top discussed last week has increased, as our stated levels to confirm the bear case were indeed breached. We can now consider how deep this downside may unfold, tracking it as an ABC to the 70-73 region.

Since striking the 89 red resistance box, CT has dropped back to its 0.382 retracement (at 77)—a likely area for a dead cat bounce to begin from.

The current price action presents a highly technical setup. After being swatted at 89 resistance, the market produced a clean (a)-wave down to the 0.382 extension. The green box represents confluence of Fibonacci support.

Looking for a (b)-wave rally here, after which we'd expect a sharp (c)-wave down to follow. Standard target for that (c) is the 0.618 retracement area - around 70. However, we will further refine targets as price action unfolds for increased accuracy.

The latest CFTC futures-and-options data, for 19 May and released 22 May, showed managed money holding 82,849 long contracts and 20,804 short contracts, leaving a net long of 62,045 contracts. That is up 2,475 contracts from the prior week’s 59,570-contract net long. The market therefore remains clearly fund-long, but the composition has changed: this was not fresh length. Managed money actually cut 1,052 longs, while shorts fell by 3,527 contracts, meaning the increase in the net long came from short-covering rather than new bullish buying.

Commercial positioning has eased slightly, but it remains a major caution point. Producer, merchant, processor and user accounts were 80,570 contracts long and 213,877 contracts short, leaving them net short 133,307 contracts. That is less short than the previous week’s roughly 138,475-contract net short, because commercial shorts fell by 6,761 contracts while longs fell by only 1,594 contracts. So this is not a fresh escalation in commercial selling; it is a modest reduction in hedge pressure. Even so, the outright commercial short remains large, equal to 42.5% of open interest, and still shows heavy producer and merchant selling into the rally.

USDA puts the 2025/26 U.S. cotton balance at 13.90 million bales of production, 12.00 million bales of exports, 1.60 million bales of domestic use and 4.40 million bales of ending stocks. That keeps old-crop stocks-to-use heavy at roughly 32%. For 2026/27, USDA projects 9.64 million planted acres, 7.38 million harvested acres, an 866 lb/acre yield, production of 13.30 million bales, exports of 12.30 million bales and ending stocks of 3.90 million bales. That lowers the stocks-to-use ratio to 28.1%, which is constructive but not tight enough to remove the need for a weather or demand catalyst.

USDA has already built some abandonment risk into the new-crop forecast. The May balance sheet uses March Prospective Plantings for planted area, while harvested area is based on regional 10-year average abandonment, with the Southwest adjusted for moisture conditions. That means part of the West Texas risk is already in the official numbers. The market will therefore need evidence of worsening dryland conditions, not just “dryness” in general, to justify a larger supply premium.

The global balance remains tighter than the U.S. balance. 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 is a constructive world setup, with consumption exceeding production by more than 5.5 million bales. However, Brazil remains a major competitive headwind: USDA has Brazil’s 2025/26 crop at 19.50 million bales with 14.70 million bales of exports, then projects 2026/27 production at 17.50 million bales but exports rising to 15.00 million bales. Brazil and the U.S. together are projected to export 27.3 million bales, about 63% of world cotton exports, so U.S. export upside still depends heavily on basis competitiveness and broader import demand.

Demand confirmation is still only partial. The latest USDA export sales week showed old-crop upland net sales of 131,800 running bales, up noticeably from the prior week and 16% above the four-week average, while new-crop sales were stronger at 216,000 running bales, mostly to Pakistan. But exports of 289,400 running bales were unchanged from the prior week and 11% below the four-week average. Total U.S. export commitments were reported at 10.994 million running bales, 1% below last year and behind the normal pace relative to USDA’s projection. That is not demand failure, but it is not yet strong enough to offset the chart damage.

USDA/NASS has now released the 26 May Crop Progress report, covering the week ending 24 May. The U.S. cotton crop was 53% planted, up from 41% the prior week and exactly in line with the five-year average of 53%. It is also slightly ahead of last year’s 50% pace.

The issue remains establishment and abandonment risk, not acreage getting planted. Texas topsoil moisture improved slightly but is still stressed: 26% very short and 30% short, or 56% short/very short, down from 61% the previous week. Texas subsoil moisture is still poor at 30% very short and 32% short, or 62% short/very short, only marginally better than last week’s 63%. Nationally, the moisture trend improved: U.S. topsoil rated short/very short fell to 38% from 44%, and subsoil short/very short fell to 41% from 45%.

At the statewide Texas level, drought coverage is still broad but not worsening in the latest official data. Drought.gov shows 65.3% of Texas in D1–D4 drought, with 21.8% in moderate drought, 29.2% in severe drought, 13.3% in extreme drought and 1.1% in exceptional drought, with the data valid 19 May. That is less severe than the cotton-weighted view because cotton acreage is concentrated in drier production areas.

The weather signal has turned more mixed-to-helpful for West Texas, but not cleanly bearish yet. WPC’s Day 1 Excessive Rainfall Outlook, valid 26–27 May, has a Slight Risk covering central to West Texas and notes potential 1–3 inch rainfall totals across the southern Texas Panhandle, Permian Basin and western Rolling Plains, with heavier local totals possible farther south and east. Day 2 shifts the heavier-rain focus toward the eastern half of Texas, the Hill Country, I-35 corridor, East Texas and the upper Texas coast.

The High Plains field-level read is also mixed. Plains Cotton Growers/Texas A&M’s 26 May update says parts of the region received small to significant rainfall, soil temperatures are warm enough for planting, and rain chances should help surface moisture, but moisture remains highly variable. It also notes that areas south of Lubbock remain in D0/D1 dryness and that rain could complicate completion of planting before insurance deadlines in the Panhandle.

CPC’s medium-range outlook keeps some rainfall relief in the forecast rather than showing a quick return to dry conditions. For 1–5 June, CPC’s 6–10 day table shows above-normal precipitation favored for North, South and West Texas, as well as Oklahoma and Kansas. The 8–14 day outlook for 3–9 June still keeps above-normal precipitation probabilities for North, South and West Texas, with West Texas temperatures moving back toward near normal.

U.S. cotton export demand turned less bearish and more mixed in the latest report. After last week’s marketing-year-low sales figure, USDA/FAS reported current-crop upland net sales of 131,800 RB for the week ending 14 May, up noticeably from the previous week and 16% above the prior four-week average. The rebound was led by Pakistan at 65,300 RB, followed by Vietnam at 26,100 RB, Turkey at 20,100 RB, Malaysia at 5,300 RB and China at 3,400 RB, partly offset by reductions for Peru and South Korea.

Shipments remained good but not accelerating. Upland exports were 289,400 RB, essentially unchanged from the previous week but still 11% below the prior four-week average. The main destinations were Vietnam at 110,800 RB, Turkey at 28,700 RB, Pakistan at 26,000 RB, Mexico at 22,100 RB and Bangladesh at 21,200 RB. So the export side is still doing enough to support USDA’s old-crop export target, but shipment momentum is not strong enough by itself to offset the recent futures-market weakness.

New-crop sales were the strongest part of the report. Upland sales for 2026/27 rose to 216,000 RB, almost entirely led by Pakistan at 206,100 RB, with smaller sales to Indonesia, Turkey and Mexico. That is a clear improvement from last week’s 29,700 RB new-crop figure, but the concentration matters: the market would take more confidence from sustained, diversified buying across Vietnam, Bangladesh, Turkey, China and Mexico rather than a single Pakistan-led week.

Weekly all-cotton sales are 145,500 bales and shipments are 308,300 bales for the week ending 14 May. Cumulative 2025/26 sales are now 11.796 million bales, still slightly behind last year’s 11.942 million at the same point, while cumulative shipments are 8.872 million bales versus 9.069 million last year. New-crop cumulative sales rose to 1.716 million bales, helped by the Pakistan-led buying in the latest week.

The broader export picture is therefore better than last week but still not outright bullish. The sales rebound removes some immediate concern after the marketing-year low, and shipments above 300,000 bales in NCC terms keep USDA’s 12.0 million bale old-crop export target in play. However, cumulative sales and shipments remain slightly behind last year, and the latest old-crop demand was still heavily dependent on Pakistan, Vietnam and Turkey. We are lowering the working full-year U.S. export view slightly from 11.5–11.6 million bales to around 11.3–11.4 million bales, with upside toward USDA’s 12.0 million if weekly shipments remain near or above 285,000–300,000 bales and net sales stay positive. If it is the case that we realized 11.3 million instead of the forecasted 12 million bales of exports this will materially effect the U.S. Balance leading to ~700k bales more carry in for the next marketing year.

Vietnam remains the anchor current-crop customer, accounting for 33% of MY2025 commitments, followed by Pakistan at 12%, Turkey at 9%, and Bangladesh and India at 7% each. The new-crop book looks different: Mexico at 17% and Pakistan at 16% are currently the top MY2026 customers, followed by Turkey, China, Indonesia and Vietnam. That reinforces the point that new-crop demand is improving, but the market still needs confirmation from broader Asian mill buying.

Near-term weather has turned less bullish for cotton, because WPC now has a heavy-rain setup reaching central to West Texas, including potential 1–3 inch totals across parts of the southern Texas Panhandle, Permian Basin and western Rolling Plains.

That could reduce some West Texas abandonment premium if it verifies over dryland cotton acres. However, it is not cleanly bearish because the Day 2 rain focus shifts more toward the eastern half of Texas and the Lower Mississippi Valley, so the key High Plains cotton areas may not all receive enough moisture. The market implication is: rain verification pressures weather premium; missed or uneven rain keeps support under cotton.

Coffee remains two-sided, with the market now caught between tight nearby inventories and a heavier new-crop supply story. July arabica settled Tuesday, May 26, at 274.00 c/lb, up 1.65 c/lb, while July robusta settled at $3,519/t, up $63/t. The bounce looks tactical rather than structural, driven by low exchange stocks, Vietnam rainfall concerns, and El Niño risk for Brazil flowering later in the year.

The front-end stock story is still supportive. ICE arabica certified stocks fell to 446,816 bags, a 3.25-month low. ICE robusta stocks remain historically tight, although they have recovered slightly from the recent two-year low. That keeps nearby spreads sensitive to any fresh stock draw, especially in robusta, but the inventory signal is more of a squeeze risk than a full bullish reset.

Brazil remains the main cap on rallies. Conab now estimates the 2026 crop at 66.7 million bags, up 18% year on year, with arabica at 45.8 million bags and conilon at 20.9 million bags. Private estimates are higher, with EISA at 75.8 million bags and expectations for record exports near 50 million bags in 2026/27. The nuance is that Brazil is not yet flooding the market, since old-crop stocks are still tight, but the trade is already pricing stronger second-half export flow.

Vietnam is the clearer bearish signal for robusta. Jan to Apr exports rose 15.8% year on year to 810,000 tonnes, while April shipments alone were up sharply. That flow weighs on London, even though short-term rainfall concerns in the Central Highlands helped trigger this week’s bounce.

Colombia remains supportive for washed arabica, but not enough to change the broader balance. April production was nearly flat year on year at 697,000 bags, while exports fell 15% to 682,000 bags. Weather and harvest delays still support differentials, but the story is more about temporary tightness than a new deficit shock.

Net-net: coffee is in an inventory trap. Low certified stocks can still punish shorts and squeeze the front end, but Brazil’s larger crop, Vietnam’s export flow, and better roaster cover limit upside follow-through. Arabica needs to hold above the upper 270s, and robusta needs to sustain trade above the mid-$3,500s, before the bounce looks like more than short covering.

ICE front month:

We continue tracking the move down off the Oct 2025 high as a five-wave impulse, currently close to completing wave (iii). A common termination point for third waves is the 1.618x Fib, residing in the 254 region. This is the region we must monitor closely for signs of a counter-trend rally beginning, that could play out over several weeks.

This past week's rally has unfolded as corrective rather than impulsive—price action is overlapping. Drilling into 1-hour bars confirms the thesis. For now, we track this as an ABC pattern, terminating at a common Fibonacci extension level—the 1.382 extension near 275. This compels us to continue looking lower in the short term, watching 254 support below us to trigger the coming wave (iv) rally.

The latest positioning data is less supportive for arabica. CFTC futures-and-options data for 19 May showed managed money in Coffee C net long 17,908 contracts, down from 25,028 the previous week. Longs fell by 2,452 contracts, while shorts rose by 4,669, so the move reflects both liquidation and fresh short selling.

Commercials were less net short, with producer, merchant, processor and user accounts net short 19,862 contracts, versus 24,345 a week earlier. Commercial longs rose sharply, suggesting industry buying on weakness, although the rise in commercial shorts also points to some renewed origin hedging.

The report is already stale, but the direction is clear: arabica’s fund cushion has thinned materially, with net length now down from more than 31,000 contracts two reports ago to below 18,000. That leaves the market vulnerable to further liquidation unless prices can recover 280/285 c/lb, and eventually 300 c/lb.

Robusta positioning is also less supportive than last week. Managed-money net length fell to about 11,097 lots, though the market remains physically tighter than arabica. Rallies therefore still need confirmation from spreads or physical demand, rather than relying only on fund buying.