📅 2026-10-01 · COMEX精銅 · 單品種深度報告舊覆蓋策略

數據截止: 2026-09-30 · 結算口徑: 交易所最終日結算價

Copper (HG=F) Deep Dive | 2026-10-01 — Constructive Base, Buy the S1 Shelf | 2026-10-01

Call: LONG copper (COMEX HGZ26) on a 1–3 week horizon, with the view invalidated on a settle below 6.55 (S2). Three reasons underpin the call. First, the pullback is a positioning flush, not a fundamental break: the 5D change is -1.95% (settle) while the 20D change is still +0.32% (settle) and price sits mid-channel at the 47.4% position of the 6.35–6.93 20-day range. Second, the visible inventory picture is tight — LME warehouse stock 249,400 MT (-1,075 d/d), SHFE warrant 10,011 MT (-721 d/d, -6.7%) — even as COMEX registered at 471,400 short tons (+1,100 d/d) absorbs metal. Third, the copper/gold ratio at 1.58 sits in the 96th percentile of its 1-year range, confirming copper's relative strength versus gold, and the 20-session seasonal window is positive in 11 of the last 15 years (median +1.88%). The main risk is the CTA proxy pinned at 98 with net length at the 69th crowding percentile — a further long liquidation toward 6.55 is the invalidation path. ATR14 is 0.134 (2.02% of price), so stops must sit beyond the S2 shelf, not inside daily noise.

1. Bottom Line & Directional Bias

Call: LONG copper (COMEX HGZ26) on a 1–3 week horizon. Invalidation: a daily settle below 6.55 (S2).

Three reasons. First, the recent weakness is a positioning flush rather than a change in the physical picture: the 5D change is -1.95% (settle) while the 20D change remains +0.32% (settle), and price at 6.62 sits at the 47.4% position of the 6.35–6.93 20-day channel — mid-range, not broken. Second, visible inventories are tight at the margin: LME warehouse stock 249,400 MT (-1,075 d/d) and SHFE warrant 10,011 MT (-721 d/d, -6.7%), with the SHFE drawdown accelerating from the prior session's -2,693 MT. Third, the copper/gold ratio at 1.58 is in the 96th percentile of its 1-year range, i.e. copper has been structurally strong versus gold, and the 20-session seasonal window is positive in 11 of the last 15 years (median +1.88%).

The bear case is not dead: the CTA trend proxy is pinned at 98 and net length sits at the 69th percentile of its 3-year crowding range, so a break of the 6.55 shelf would force a mechanical exit. That is precisely why the invalidation is set at the S2 shelf rather than at the pivot.

2. Price Action & Technical Analysis

HG=F settled at 6.62 on 2026-09-30, +0.27% on the day (settle). In early Asian trade on 2026-10-01 06:55 the last print was 6.63 (+0.09% vs settle), with an Asia session range of 6.63–6.64 — a narrow, quiet tape, labelled as Asia and not to be confused with the settled bar.

Momentum is mixed by design. The 5D change is -1.95% (settle) and the 20D change is +0.32% (settle). That combination — negative short window, positive longer window — is the signature of a pullback inside an intact range rather than a trend reversal. The 20-day channel is 6.35–6.93, and at 6.62 the market sits at the 47.4% position of that channel, essentially dead-centre.

Volatility is elevated but not disorderly. ATR14 is 0.134, i.e. 2.02% of price, expressed as a full daily range — not a ± band. RV20 is 26.4% annualized. With ATR at roughly 0.134, a stop placed less than one ATR from entry would sit inside ordinary daily noise; the S2 shelf at 6.55 is the nearest structurally meaningful level below the market.

Daily pivots from the settle-based snapshot: P 6.64, R1 6.68, S1 6.59, R2 6.73, S2 6.55. Price at 6.62 is trading just below the pivot, between S1 and P — a mildly soft intraday posture that is consistent with the 5D drawdown. The first upside hurdle is P at 6.64, then R1 6.68; the first support is S1 6.59, then S2 6.55.

On the weekly timeframe, the last completed weekly bar (2026-09-21 to 2026-09-25) opened 6.72, high 6.93, low 6.7, closed 6.77, +1.11% w/w. The current week (from 2026-09-28, three sessions in) is not closed and shows a last print of 6.62, -2.13% — no weekly-close conclusion can be drawn from an unfinished bar. The 52-week range is 4.78–6.93, so the market is trading in the upper third of its annual range and the 6.93 high is the reference ceiling.

View: the tape is a mid-range consolidation with a soft short-term tilt; the constructive read requires price to reclaim P 6.64 and hold above S1 6.59.

3. Supply-Demand Balance & Fundamental Drivers

COMEX registered copper inventory stood at 471,400 short tons on 2026-09-29, up 1,100 short tons d/d (+0.2%). This is the one visible pool still building, and it is the main counterweight to the bullish case: metal is being delivered into the US exchange rather than drawn from it.

Against that, the rest of the visible complex is tightening. LME warehouse stock was 249,400 MT on 2026-09-30, down 1,075 MT d/d (-0.4%), though the 20-report change is +15,550 MT, meaning the LME draw is recent and the broader month still shows net accumulation. SHFE warrant was 10,011 MT on 2026-09-30, down 721 MT d/d (-6.7%), following a 2,693 MT draw the prior session (10,732 MT on 2026-09-29). The SHFE warrant level is small in absolute terms and the two-day draw is meaningful — Chinese exchange metal is being consumed.

The read-through: the global visible stock picture is bifurcated. US exchange stocks are rising while LME and especially SHFE stocks are falling. That pattern is consistent with a tariff- or logistics-driven relocation of metal into the US rather than a genuine global surplus. For price, the marginal signal is the LME and SHFE draw, which is why the fundamental tilt remains constructive even with COMEX building.

On the news flow, the 48-hour headlines are supply-side and mostly supportive of the medium-term thesis rather than the immediate price: Rio Tinto secured Bell Bay smelter operations through end-2031, removing a near-term curtailment risk; Atico secured a US$111.4 million financing package for the La Plata copper-gold project, adding future supply; and the US DOE awarded $29.5M for 17 national lab mining technology projects, a longer-dated supply-efficiency signal. None of these change the next-20-session balance. The nickel headlines (Indonesia HMA and ore prices declining) are a reminder that the base-metals complex is not uniformly tight, but nickel weakness does not transmit directly to copper.

Macro transmission: US 10-year yield at 5.293 (+0.72%) and DXY at 101.46 (+0.09%) are both mild headwinds for a dollar-denominated industrial metal. A 5.29% 10-year is restrictive for global industrial demand, and that is the principal macro risk to the constructive view. However, the copper/gold ratio at 1.58 (96th percentile, 1Y) shows copper has been winning the relative-growth argument against gold despite that backdrop.

View: physical tightness outside the US and a small SHFE warrant base support the long side; the COMEX build and a 5.29% 10-year cap the upside.

4. Positioning & Fund Flows

CFTC managed-money positioning has swung hard in both directions over the last four reports. Net length went from 72,882 (2026-09-01) to 82,154 (2026-09-08, +9,272) to 65,106 (2026-09-15, -17,048) and back to 82,522 (2026-09-22, +17,416). Open interest rose from 282,640 to 301,657 over the same span. The most recent week added both length (96,421) and reduced shorts (13,899), a clean bullish repositioning.

Crowding metrics: netPct 27.36% on 2026-09-22, crowding percentile 69.05 on a 3-year window. That is elevated but not extreme — it is below the 80th percentile threshold that would justify calling the trade crowded. The CTA trend-following proxy is pinned at 98 across all four weeks, and the hedge ratio is 56.61%. The CTA reading is the key vulnerability: with trend followers maximally long, any sustained break of support triggers mechanical selling, which is exactly the path to the 6.55 invalidation.

On volatility, RV20 is 26.4%. The available implied-vol proxies are for other assets — ^OVX 52.24 (52nd percentile), ^GVZ 23.74 (20th percentile), ^VXSLV 37.87, ^VIX 16.34 (33rd percentile) — and none of them is a copper IV, so no direct copper implied-versus-realized conclusion can be drawn. What the cross-asset vol complex does show is that equity and gold optionality are cheap relative to their own histories (VIX 33rd, GVZ 20th percentile), i.e. macro event risk is not being priced aggressively into those markets.

View: positioning is supportive but no longer a tailwind of the same magnitude; the marginal buyer is the physical market, not the fund community.

5. Cross-Asset Relative Value

HG/GC ratio at 1.58 (ratio ×1000), 1-year percentile 95.63%, 3-year percentile 53.44%. The interpretation matters: a high 1-year percentile means copper has been outperforming gold over the past twelve months — this is a pro-growth signal, not a copper-weakness signal. The 3-year percentile at 53.44% shows the ratio is only mid-range over a longer horizon, so the recent copper strength is a recovery within a broader neutral band rather than a stretched extreme.

Against the dollar, DXY at 101.46 (+0.09%) is a mild headwind but not a dominant one; the ratio's strength has been achieved despite a firm dollar and a 5.29% 10-year yield, which strengthens the signal that copper's relative bid is growth- and supply-driven rather than a pure dollar play.

The term structure is in contango: M1-M2 -0.019 (-0.29%), roll yield -3.5%, slope 0.0448. A shallow contango of this magnitude is a carry cost for long positions, not a directional signal — it does not cap price. It does mean that holding a long futures position for an extended period bleeds roughly 3.5% annualized in roll, which argues for a shorter holding horizon (1–3 weeks) rather than a multi-month carry trade.

View: copper's relative-value position versus gold is a tailwind for the long thesis; the contango argues for tactical rather than strategic length.

6. Historical & Seasonal Patterns

Using the same calendar start and the next 20 sessions over the last 15 years: mean +2.46%, median +1.88%, up in 11 of 15 years. The best instance was 2011 at +15.57% and the worst was 2012 at -7.01%. The sample is small and the dispersion is wide — the standard deviation implied by a +15.57% best and -7.01% worst is large relative to the +1.88% median — so this should be treated as context, not as a standalone edge.

What the seasonality does provide is a probabilistic tilt that aligns with the constructive call: a roughly 73% hit rate (11/15) and a positive median over the next 20 sessions. Combined with the mid-range technical position and the physical draw in LME/SHFE stocks, the seasonal window supports holding length through early October rather than chasing the 5D weakness.

View: seasonality is a modest tailwind for the long side over the next 20 sessions; it does not override the 6.55 invalidation.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 55% probability. Trigger: price holds above S1 6.59 and reclaims the pivot at 6.64. Path: consolidation between 6.59 and 6.73 (R2), with the market grinding back toward the 20-day channel midpoint and the 6.77 level of the last completed weekly close. Target: 6.73–6.77. Action: hold existing length, add on a settle above 6.64. This scenario agrees with the section 1 call.

Bull case — 25% probability. Trigger: a daily settle above R2 6.73, ideally accompanied by a further SHFE warrant draw and a softer dollar. Path: the market retests the 20-day high at 6.93 and the 52-week high at 6.93. Target: 6.93. Action: add to length on the breakout close, trail stops to the 6.64 pivot. The catalyst would most likely be a dovish surprise from the FOMC minutes (BJT 10-08 02:00) or a weak non-farm payrolls print (BJT 10-02 20:30).

Bear case — 20% probability. Trigger: a daily settle below S2 6.55, which would confirm the CTA proxy at 98 unwinding. Path: the 20-day low at 6.35 comes into play, with the 52-week range floor at 4.78 far below. Target: 6.35. Action: exit all length on the settle below 6.55; do not attempt to fade the move. The catalyst would be a hot non-farm payrolls print (forecast 89K vs previous 162K, surprise threshold ±73K) driving the 10-year yield above 5.29% and the dollar higher.

Probabilities sum to 100%. The base case is the section 1 call; the bull and bear cases are probability-weighted paths, not alternative conclusions.

8. Trading Strategies & Risk Management

Strategy 1 — Tactical long (primary). Direction: LONG COMEX HGZ26. Entry: 6.62 (current settle area) or on a reclaim of P 6.64. Stop: 6.52, which sits beyond the S2 shelf at 6.55 and roughly 0.1 below entry — inside one ATR14 of 0.134 but beyond the structural level. Target: 6.73 (R2) for the first tranche, 6.93 (20-day and 52-week high) for the second. Timeframe: 1–3 weeks. Size: half of normal risk budget on the initial entry, adding the second half only on a settle above 6.64. Conviction: 7/10.

Strategy 2 — Breakout continuation (secondary). Direction: LONG. Entry: on a daily settle above 6.73 (R2). Stop: 6.59 (S1), which is beyond the pivot and approximately one ATR below the breakout level. Target: 6.93. Timeframe: 5–10 sessions. Size: quarter of normal risk budget, given the entry is extended relative to the 20-day channel. Conviction: 6/10.

Risk management: the invalidation for the entire thesis is a daily settle below 6.55. The contango roll yield of -3.5% annualized means positions should be held tactically rather than carried for months. The 10-year yield at 5.29% and DXY at 101.46 are the macro variables to monitor; a sharp move higher in either would argue for reducing size ahead of the 6.55 stop.

9. This Week's Data Calendar

BJT 10-01 22:00 | ET 10-01 10:00 — FOMC Member Waller speaks; ISM Manufacturing PMI (forecast 54.8, previous 54.6, surprise outside ±0.2) and ISM Manufacturing Employment (forecast 51.5, previous 51.2, surprise outside ±0.3).
BJT 10-02 20:30 | ET 10-02 08:30 — Non-Farm Employment Change (forecast 89K, previous 162K, surprise outside ±73K), Average Hourly Earnings m/m (forecast 0.3%, previous 0.3%, surprise outside ±0.1%), Unemployment Rate (forecast 4.1%, previous 4.1%, surprise outside ±0.1%).
BJT 10-05 22:00 | ET 10-05 10:00 — ISM Services PMI (forecast 54, previous 55.4, surprise outside ±1.4).
BJT 10-08 02:00 | ET 10-07 14:00 — FOMC Minutes.

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.

單品種深度報告深度分析

COMEX黃金GC=F 2026-10-01

Gold (GC=F) Deep Dive | 2026-10-01 — Long into 4,143 Support, Invalidation 4,132 | 2026-10-01

Copper (HG=F) Deep Dive | 2026-10-01 — Constructive Base, Buy the S1 Shelf | 2026-10-01

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