Copper (HG=F) Deep Dive | 2026-09-27 — Long Into the 6.93 Range Top, 6.69 Invalidates | 2026-09-27

Call: Bullish copper (HG=F, front-month reference HGZ26.CMX) on a 1–2 week horizon, with invalidation on a settle below 6.69. Three reasons: (1) price structure — settle 6.766 [2026-09-25] sits at the 72.2% position of the 20-day 6.3445–6.9285 channel, above pivot P 6.762, with 5D +1.11% and 20D +1.15% (settle) and the last completed weekly bar (2026-09-14–18) up 2.19% w/w to 6.6915; (2) inventories — COMEX registered 475,500 short tons [2026-09-25] down 1,600 d/d and SHFE warrants 16,620 MT [2026-09-24] down 2,468 (-12.93%) d/d, with only LME adding 325 MT to 251,500 MT; (3) positioning — managed-money net 78,058 lots at the 69th percentile of its 3-year range (as-of 2026-09-22), elevated but not extreme, with CTA proxy +62. The bear counterweight is the curve flip to CONTANGO (M1–M2 -0.034, -0.5%; roll yield -6.06%) and a 15-day-stale COT print showing net -4,464 w/w. Copper/gold at 1.57 (95th percentile 1Y) says the pro-growth leg is already priced.

1. Bottom Line & Directional Bias

Call: Bullish copper (HG=F; front-month reference HGZ26.CMX) over a 1–2 week horizon. Invalidation: a daily settle below 6.69, which would put price back under the 20-day mid-channel and negate the constructive inventory and positioning setup.

Three reasons. First, price structure: the 2026-09-25 settle of 6.766 sits at the 72.2% position of the 20-day 6.3445–6.9285 channel, above pivot P 6.762, with 5D +1.11% and 20D +1.15% (settle) — a market grinding higher, not fading. Second, visible inventories are tightening at the margin: COMEX registered 475,500 short tons [2026-09-25] fell 1,600 d/d and SHFE warrants 16,620 MT [2026-09-24] fell 2,468 (-12.93%) d/d, with only LME warehouse stock adding 325 MT to 251,500 MT. Third, positioning is supportive without being stretched: managed-money net 78,058 lots, 27.36% of open interest, at the 69th percentile of its 3-year range (as-of 2026-09-22).

The offsetting evidence is real and must be respected: the curve has flipped FLAT → CONTANGO (M1–M2 -0.034, -0.5%; roll yield -6.06%), which removes the carry tailwind and makes every long a payer, and the copper/gold ratio at 1.57 sits in the 95th percentile of its 1-year range — the pro-growth trade is crowded at the macro level. We stay long, but with a defined stop and a target short of the 52-week high.

2. Price Action & Technical Analysis

The 2026-09-25 final settlement was 6.766, down 0.35% on the day (settle). The last five settled bars show a market that probed 6.9285 on 09-22, failed, and has since coiled in a tight 6.723–6.803 band: 09-21 C 6.7625, 09-22 C 6.836, 09-23 C 6.7535, 09-24 C 6.79, 09-25 C 6.766. The 20-day range is 6.3445–6.9285, and at 6.766 price sits at the 72.2% position — upper third, but not at the ceiling. The 52-week range is 4.6695–6.9285, meaning the 09-22 high was an exact test of the 52-week high and the market has so far failed to clear it.

Momentum is modest but positive: 5D +1.11% and 20D +1.15% (settle). ATR14 is 0.164, i.e. 2.42% of price as a full daily range — a wide tape where a 0.1–0.15 intraday swing is noise, not signal. RV20 is 25.8% annualized, which is the realized backdrop against which any option expression should be judged.

Pivots from the settle-based snapshot: P 6.762, R1 6.801, S1 6.727, R2 6.836, S2 6.688. Price settled essentially on the pivot, with R1 6.801 the first real hurdle and R2 6.836 the level that would re-open the 6.9285 high. To the downside, S1 6.727 is the first shelf and S2 6.688 is the line that matters — it sits just under our 6.69 invalidation and coincides with the 20-day mid-channel.

On the weekly frame, the last completed bar (2026-09-14–2026-09-18) opened 6.5525, high 6.725, low 6.3445, closed 6.6915, +2.19% w/w — a constructive completed week that reclaimed the 6.65 area. The current week (from 2026-09-21, five sessions) is unfinished and last printed 6.766 (+1.11%); no weekly-close conclusion can be drawn from it. The report-date bar is an unfinished Globex/Asia bar and is not a close.

View: constructive while above 6.727; a settle above 6.836 opens 6.9285, and a settle below 6.688 breaks the structure.

3. Supply-Demand Balance & Fundamental Drivers

Visible copper inventories are mixed but net-tightening at the margin. COMEX registered stocks were 475,500 short tons as of 2026-09-25, down 1,600 short tons (-0.34%) from 477,100 on 2026-09-24. LME warehouse stock was 251,500 MT as of 2026-09-25, up 325 MT (+0.13%) d/d, but the 20-report change is +15,925 MT — the LME has been the accumulation venue, not the draw. SHFE warrants were 16,620 MT as of 2026-09-24, down 2,468 MT (-12.93%) d/d, a sharp single-day decline that points to Chinese downstream restocking ahead of the National Day holiday window.

The geographic split matters for the trade: the tightness is in the US (COMEX) and China (SHFE), while LME is the release valve. That configuration typically supports the front of the US curve relative to London, and it is consistent with the 52-week-high test on COMEX. It is not, however, a global deficit signal — the LME build of +15,925 MT over 20 reports is the counterweight.

On the physical demand side, the headline flow is constructive: Aurubis and Prysmian signed their largest-ever long-term copper wire rod supply agreement (SMM, 2026-09-27), a multi-year offtake commitment that speaks to grid and electrification demand visibility. Secondary supply is also scaling — Metycle secured a $150 million credit facility to scale secondary copper and aluminium supply, and BioMetallica is planning an integrated urban-mining facility in the Philippines for e-waste and industrial scrap recovery. Both are medium-term bearish for refined primary demand, but neither moves the next two weeks.

Macro transmission is through rates and the dollar, not through copper-specific news. The 10-year Treasury yield at 5.184 (+0.43%, 2026-09-25) and DXY at 100.97 (-0.32%, 2026-09-25) are pulling in opposite directions; the SMM morning comment explicitly flagged rate-hike expectations pressuring copper prices overnight. The curve has flipped FLAT → CONTANGO (M1–M2 -0.034, -0.5%; roll yield -6.06%), which is the single most important fundamental-structure change this week: it says spot is no longer scarce relative to forward, and it imposes a carry cost on longs.

View: the inventory mix (COMEX and SHFE drawing, LME building) plus long-dated offtake commitments support a grind higher, but contango caps the upside velocity. Net constructive, not a squeeze.

4. Positioning & Fund Flows

CFTC managed-money positioning (as-of 2026-09-22, the latest report available and 15 days old — not the current week): open interest 301,657, longs 96,421, shorts 13,899, net 82,522, up 17,416 w/w. The following week's row (2026-09-29) shows OI 301,201, longs 92,094, shorts 14,036, net 78,058, down 4,464 w/w. Net as a share of OI was 27.36% on 2026-09-22, at the 69th percentile of its 3-year range — elevated, but below the 69th percentile of 2026-09-08 and well short of an extreme reading. The CTA trend-following proxy is +62, unchanged across the four weeks, and the hedge ratio is 56.61%.

This is a constructive but not crowded long base. The 2026-09-15 row (net 65,106, netPct 22.49%, crowd 63.36) shows the market added length into the 09-22 high, then trimmed modestly into 09-29. Price up with open interest up is active buying, not divergence; the small net reduction into a flat-to-lower price week is ordinary de-risking, not a positioning reversal. We would only call this crowded if netPct pushed toward the top decile of the 3-year window, and at 69.05 it has not.

On the volatility side, copper's RV20 is 25.8%. The cross-asset implied-vol backdrop is benign: ^VIX 14.87 (9th percentile 1Y), ^GVZ 22.44 (13th percentile 1Y), ^VXSLV 35.99 (-1.81 pts), ^OVX 55.09 (58th percentile 1Y). With copper realized vol at 25.8% and broad equity and gold implied vol in the bottom quartile of their 1-year ranges, optionality across the complex is cheap — which argues for expressing the long via options or a defined-risk structure rather than a naked futures stop.

View: positioning supports further upside but does not fuel a squeeze; with RV20 at 25.8% and cross-asset IV cheap, convexity is the better expression than outright length.

5. Cross-Asset Relative Value

The copper/gold ratio is 1.57, at the 95th percentile of its 1-year range and the 52nd percentile of its 3-year range (spreads table, settle 2026-09-25). Read correctly: copper has been strongly outperforming gold over the past year, and the ratio is near the top of its 1-year distribution. That is a pro-growth signal that has already been paid for — it is not a fresh buy signal, and it is the main reason we size this long modestly rather than aggressively. On a 3-year view the ratio is mid-range, so this is a 1-year extension, not a multi-year extreme.

The rates and dollar backdrop is mixed for copper. The 10-year at 5.184 (+0.43%, 2026-09-25) is a headwind for all real assets and was cited in the SMM morning comment as pressuring copper overnight. DXY at 100.97 (-0.32%, 2026-09-25) is a modest tailwind. The two roughly offset, which is consistent with copper's low realized vol (RV20 25.8%) and tight recent range.

Within the industrial complex, the contango structure (M1–M2 -0.034, -0.5%) is copper-specific and bearish relative to a backwardated market; it means copper longs pay carry while holders of prompt-tight commodities are paid. That is a relative-value argument for preferring copper exposure via equities or options over a rolling futures long, and it is why our strategy section uses a defined-risk structure.

View: copper/gold at the 95th 1-year percentile says the growth trade is consensus; the contango says carry is negative. Own copper, but own it with convexity, not with roll exposure.

6. Historical & Seasonal Patterns

Seasonality for the same calendar start (next 20 sessions, last 15 years): mean +1.48%, median +0.82%, up in 8 of 15 years. Best case 2017 +8.95%, worst 2012 -4.9%. This is context only — a 15-observation sample with a hit rate barely above half and a mean dragged by a single outlier year.

The honest read: the median (+0.82%) is the better central estimate than the mean (+1.48%), and an 8/15 hit rate is a coin flip with a slight positive skew. Seasonality is a mild tailwind that supports holding a long, but it is not strong enough to justify initiating size on its own, and the -4.9% worst case is well beyond our 6.69 invalidation from the current 6.766 settle — a reminder that the seasonal distribution has a fat left tail.

View: seasonality adds a small positive tilt to the long but carries no independent weight; the trade stands on structure, inventories and positioning.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 50%: grind higher toward 6.9–6.95. Trigger: price holds above pivot P 6.762 and clears R1 6.801 on a settle, with COMEX and SHFE inventories continuing to draw. Target 6.9–6.95, just shy of the 6.9285 52-week high. Action: hold the long, trail the stop up behind S1 6.727 as R1 converts to support. This is the path consistent with our call.

Bull case — 30%: breakout through the 52-week high. Trigger: a daily settle above R2 6.836 followed by a settle above 6.9285, ideally on a Chinese Manufacturing PMI print above the 50.1 forecast (BJT 09-30 09:30) and a soft Core PCE (BJT 09-30 20:30). Target 7.05–7.15, an extension leg above the range. Action: add on the retest of 6.9285 as support, move the stop to breakeven, and take partial profit into 7.05. Note that contango means this leg must be driven by spot tightness, not carry.

Bear case — 20%: failure at the range top and a break of structure. Trigger: a daily settle below S2 6.688, which is also our invalidation, most likely on a hot Core PCE (forecast 0.3% m/m, surprise if outside ±0.1%) that pushes the 10-year further above 5.18 and lifts DXY, or a Chinese PMI miss. Target 6.55, then 6.45. Action: exit the long on the settle, stand aside; do not fade the move, because the 20-day low at 6.3445 is the next structural reference and the seasonal worst case (-4.9%) maps near it.

Probabilities sum to 100%. The base case is our call; the bull and bear cases are the weighted paths around it, not alternative conclusions.

8. Trading Strategies & Risk Management

Strategy 1 — Long HGZ26.CMX on a pullback to the pivot (primary). Entry 6.76–6.78, stop 6.64 (below S2 6.688 and roughly one ATR14 of 0.164 from entry), target 6.95, horizon 1–2 weeks, conviction 7. Size at half normal risk budget given the 95th-percentile copper/gold ratio and the contango carry drag. This aligns with the house view (entry 6.7, stop 6.55, target 6.95, 1–2w) but sits higher in the range because the 2026-09-25 settle of 6.766 has already left the 6.7 entry behind; the house stop at 6.55 is wider than our structural invalidation and we prefer the tighter 6.64.

Strategy 2 — Defined-risk call spread for convexity. With RV20 at 25.8% and cross-asset implied vol cheap (^VIX 9th percentile, ^GVZ 13th percentile 1Y), buy a 6.85/7.05 call spread on HGZ26.CMX expiring in 3–4 weeks, risking no more than 0.5% of book. This avoids the -6.06% roll yield embedded in the contango and caps the loss if the 6.688 break occurs. Horizon 2–4 weeks, conviction 6.

Risk management: no adds below 6.727; a settle below 6.688 exits both structures. Do not hold either position through the BJT 09-30 20:30 Core PCE without reducing size, given the 10-year at 5.184 and the explicit rate-hike pressure cited in the copper morning comment.

9. This Week's Data Calendar

China PMIs are the dominant copper catalyst: Manufacturing PMI (BJT 09-30 09:30 | ET 09-29 21:30), forecast 50.1 vs prior 49.8, surprise if outside ±0.3; Non-Manufacturing PMI forecast 49.2 vs 49.0; RatingDog Manufacturing PMI forecast 51.7 vs 51.5 (BJT 09-30 09:45 | ET 09-29 21:45). US Core PCE m/m (BJT 09-30 20:30 | ET 09-30 08:30), forecast 0.3% vs prior 0.2%, surprise if outside ±0.1%, alongside Final GDP q/q at 1.5%. JOLTS (BJT 09-29 22:00 | ET 09-29 10:00) and ISM Manufacturing PMI (BJT 10-01 22:00 | ET 10-01 10:00), forecast 54.8 vs 54.6, round out the week.

This report is generated automatically from public quantitative and macro data for research and market tracking only. It does not constitute investment advice or a recommendation to trade.

Copper (HG=F) Deep Dive | 2026-09-27 — Long Into the 6.93 Range Top, 6.69 Invalidates | 2026-09-27

今日風險地圖

單品種深度報告 深度分析

往期研報封存

⟠ 本月策略回測表現

單品種深度報告深度分析