COT May 26 – Treasury Curve Split & 12 Extremes

Twelve markets reached 95th+ or 5th- percentile positioning extremes this week. That's not unusual breadth by itself, but the composition is worth noting: three Treasury futures at opposing extremes, both precious metals near multi-year lows, and four agricultural markets clustered at 97th+ percentile net long.

For readers new to this data: the Commitment of Traders (COT) report tracks how hedge funds, commercial hedgers, and retail traders are positioned in futures markets. Extreme percentile readings indicate crowding conditions — not price signals, but observations about where speculative capital has piled up or withdrawn.

This analysis uses data from the CFTC Commitment of Traders Report. Positioning percentiles are calculated across two lookback windows: 2-year (104 weeks) as the primary actionable reference, and 10-year (520 weeks) for structural context. Full Methodology → | Download this week's data (JSON) | View Interactive Dashboard →

Top 10 Extreme Positions

Rank	Market	Net Position	2Y Pct	10Y Pct	Z-Score
1	30Y Treasury	-199,251	1st	4th	-3.09
2	5Y Treasury	-1,323,127	99th	22nd	1.80
3	Brazilian Real	71,651	99th	99th	1.41
4	Gold	154,260	1st	27th	-1.63
5	Nasdaq 100	-11,282	1st	2nd	-2.03
6	Natural Gas	-203,181	1st	3rd	-2.10
7	Silver	22,223	1st	26th	-1.75
8	Soybean Meal	154,594	99th	97th	2.21
9	Bitcoin	2,282	98th	99th	1.94
10	Copper	73,040	98th	99th	2.06

Treasury Curve: Opposite Extremes at Each End

30Y Treasury net shorts hit -199,251 contracts this week — a 1st percentile reading over two years (4th percentile over 10 years) with a Z-score of -3.09. Within this 104-week sample, readings beyond -3.0 have been uncommon. The position has built consistently: -171k vs the 4-week average, -164k vs 13 weeks. Non-commercial traders now hold net shorts equal to 9.9% of open interest.

Meanwhile, 5Y Treasury positioning sits at the opposite extreme: -1.32 million contracts net short, 99th percentile over two years (though only 22nd percentile over 10 years — this level has precedent in earlier cycles). The crowding ratio is -19.3% of OI, nearly double the 30Y concentration.

The middle of the curve is unremarkable: 2Y and 10Y both sit near their 2-year medians (52nd and 53rd percentile respectively). So this isn't a generalised bond market crowding story. It's a curve positioning split — ultra-long duration極度net short at a 3-sigma extreme, intermediate duration extremely net short within the recent sample but structurally mid-range.

Metals: Both Gold and Silver Near Two-Year Lows

Gold positioning dropped to 154,260 contracts net long — 1st percentile over two years, 27th over 10 years. Silver followed the same pattern: 22,223 net long (1st percentile 2Y, 26th 10Y). Both have built these reduced net long positions over multiple weeks rather than spiking this week.

The crowding ratios tell the positioning concentration story: Gold net longs represent 43.6% of open interest, Silver 21.8%. These are still sizable net long exposures in absolute terms, but they're among the lowest relative to recent history. Commercials hold offsetting net shorts: -185,766 in Gold (hedging intensity 43.6% of OI), -40,893 in Silver.

Worth noting what this does NOT tell you: whether open interest itself is rising or falling. OI history is not currently tracked in this dataset, so I can't assess whether this represents capital leaving the complex or simply a rotation in who holds which side.

Copper sits at the other end: 98th percentile net long over two years (99th over 10 years), with net longs representing 26.5% of OI. So within metals, you've got precious at multi-year positioning lows and industrial at multi-year highs.

Grains: Clustered Long Extremes in Four Markets

Soybean Meal reached 99th percentile net long (97th over 10 years), Cotton 97th percentile (79th over 10 years), Wheat (SRW) 96th percentile, and Soybean Oil 93rd percentile. All four show sustained builds across both 4-week and 13-week averages — this isn't tactical event positioning, it's multi-month accumulation.

The concentration varies: Cotton crowding ratio is 26.5% of OI, Soybean Meal is less specified in the summary data but sits at 154,594 contracts net long. Commercial offsetting positions range from -29.2% of OI in Cotton to smaller hedging ratios in the soy complex.

Coffee C remains at the 6th percentile (multi-week persistence, suppressed from headline findings per the data rules), Cocoa at 13th, Sugar mid-range at 34th. So this isn't a broad agricultural crowding wave — it's specific to the soy complex and cotton.

FX, Crypto, and the Alignment Signal

Australian Dollar sits at 89th percentile over two years (96th over 10 years), with net longs representing 19.9% of OI. Brazilian Real hit 99th percentile in both windows — extreme positioning with multi-year conviction. Bitcoin reached 98th percentile (99th over 10 years), though the net position is small in absolute terms: 2,282 contracts, representing 10.6% of OI.

One market flagged the alignment signal this week: Feeder Cattle, where both large speculators (+1,598 net, 2.7% of OI) and commercials (+468 net, 0.8% of OI) are positioned net long. Because futures markets are zero-sum, this means small speculators hold the entire net short position. The market sits at the 20th percentile — not an extreme reading, so this is a structural observation rather than an amplified crowding flag.

When large specs and commercials align on the same side, small speculator positions have historically sometimes coincided with periods of elevated positioning volatility — though the timing and magnitude cannot be determined from positioning data alone. This configuration is not a directional indicator, just a note about who holds what.

Conditional Observations: Three Markets

30Y Treasury — 1st percentile net short, Z-score -3.09, sustained build over 13 weeks:

• If net shorts continue building from this 3-sigma level while yields fail to make new highs, then this may be consistent with a crowding condition that has historically sometimes preceded tactical covering — though extreme positioning can persist for extended periods when macro narratives remain intact.
• If positioning stabilises near current levels while yields consolidate, then this 1st percentile reading (in the 2-year window) may simply represent a structural shift in how hedge funds position duration risk — the 10-year percentile (4th) suggests this level has precedent.
• If net shorts unwind rapidly over the next 2-4 weeks, then the speed of entry (sustained build vs. spike) suggests this is conviction-based positioning rather than tactical, meaning unwinding may be slower and more price-sensitive than event-driven spikes typically are.

Gold — 1st percentile net long, Z-score -1.63, multi-week reduction:

• If net longs continue declining from this 2-year low while price holds or rises, then this may be consistent with speculative capitulation patterns that have historically sometimes coincided with intermediate trend changes — though positioning alone does not confirm price direction.
• If positioning stabilises near current levels (1st percentile 2Y, 27th percentile 10Y), then this may represent a return to a longer-term structural range rather than an extreme washout — the 10-year context suggests net longs at 154k have precedent.
• If net longs spike back above the 4-week average (+154k vs. current levels) within the next 1-2 reports, then the NC crowding ratio (43.6% of OI) indicates the trade remains concentrated enough to move quickly on macro catalysts — though the direction of that move cannot be determined from positioning data.

Soybean Meal — 99th percentile net long (97th over 10 years), Z-score 2.21, sustained build:

• If net longs continue building from this 99th percentile level, then the sustained nature of the build (consistent across 4-week and 13-week averages) suggests structural conviction rather than tactical positioning — historically, conviction-based extremes have sometimes persisted longer than event-driven spikes.
• If price stalls or reverses while positioning remains at this extreme, then the concentration (99th percentile in both windows) may indicate vulnerability to unwinding — crowded consensus trades are particularly sensitive to price disappointment.
• If net longs unwind sharply over the next 2-3 weeks, then the speed of the prior build suggests this is multi-month accumulation, meaning unwinding may be gradual and price-dependent rather than a single-week flush.

This analysis is for educational purposes only and does not constitute financial advice.

Next week's report covers the period ending June 2, 2026.

Explore More: Interactive Dashboard | Methodology | Download Data | Previous Analysis