📅 2026-10-07 · COMEX精銅 · 單品種深度報告

數據截止: 2026-10-06 · 結算口徑: 交易所最終日結算價

Copper (HG=F) Deep Dive | 2026-10-07 — Constructive Base, Long Bias Into 6.72/6.93 | 2026-10-07

Call: LONG copper (HG=F, front-month reference HGZ26.CMX) from the 6.65 settle, with invalidation on a settle below the 6.58 pivot S2 / 6.51 weekly low. Three reasons underpin the call. First, the tape has stabilised: after a -3.21% w/w completed week (2026-09-28–10-02, close 6.55), the last two settled bars have ground higher to 6.65, and the 20-day channel position is 52.2% — mid-range, not extended. Second, visible inventories are not building: LME warehouse stock fell 3,750 MT to 244,900 MT (as of 2026-10-05) and SHFE warrants dropped 6.72% to 10,011 MT (as of 2026-09-30), while COMEX registered stocks were flat at 471,400 short tons (as of 2026-10-02). Third, positioning is supportive but not crowded: managed-money net length is 78,058 contracts, netPct 25.92%, crowding 68.51 on a three-year window, with the CTA proxy at +62. The copper/gold ratio at 1.6 sits in the 98th percentile of its one-year range — pro-growth confirmation, not a contrarian signal. Invalidation: a daily settle below 6.58.

1. Bottom Line & Directional Bias

Call: LONG copper (HG=F; front-month reference HGZ26.CMX), invalidation on a daily settle below 6.58.

Three reasons. (1) Price structure has repaired off the 6.51 low of the last completed weekly bar (2026-09-28–10-02, close 6.55, -3.21% w/w): the last two settled bars closed 6.64 and 6.65, and the settle sits at the 52.2% position of the 20-day 6.35–6.93 channel — mid-range with room to the top. (2) The visible inventory picture is neutral-to-tight: LME warehouse stock 244,900 MT (as of 2026-10-05, -3,750 MT d/d), SHFE warrants 10,011 MT (as of 2026-09-30, -6.72% d/d), COMEX registered 471,400 short tons (as of 2026-10-02, unchanged d/d). (3) Positioning is constructive without being stretched: managed-money net 78,058 contracts, netPct 25.92%, crowding 68.51 (three-year percentile), CTA proxy +62.

The copper/gold ratio at 1.6 (98th percentile, 1Y) confirms the pro-growth read. Invalidation is a settle below 6.58 (pivot S2), which would put the 6.51 weekly low back in play and break the two-session recovery sequence. Horizon: 1–3 weeks.

2. Price Action & Technical Analysis

Settle 6.65 [2026-10-06], +0.13% on the day, +0.7% over five sessions, -2.55% over twenty. The 20-day range is 6.35–6.93, placing the settle at the 52.2% position; the 52-week range is 4.83–6.93, so price is in the upper third of the annual envelope but below the 6.93 ceiling. ATR14 is 0.127, i.e. 1.91% of price as a full daily range — a normal-volatility regime, not a breakout tape. RV20 is 25.8%.

Pivots from the settle: P 6.65, R1 6.69, R2 6.72, S1 6.61, S2 6.58. The immediate task for longs is a settle above R1 6.69, which would open R2 6.72 and then the 6.93 twenty-day high. The first line of defence is S1 6.61; the invalidation line is S2 6.58.

The last five settled bars tell a coherent story: 09-30 close 6.62, 10-01 close 6.54, 10-02 close 6.55, 10-05 close 6.64, 10-06 close 6.65. Two consecutive higher closes off the 6.51–6.52 base, with the 10-06 high at 6.68 pressing R1. In early Asian trade on 2026-10-07 (08:00), the last print was 6.65, +0.05% versus the settle, with an Asian session range of 6.65–6.67 — a tight, quiet session that has not yet challenged R1.

Weekly context: the last completed weekly bar (2026-09-28–2026-10-02) opened 6.77, high 6.77, low 6.51, closed 6.55, -3.21% w/w — a lower weekly close that defines the 6.51 support the market is now working off. The current week (from 2026-10-05, two sessions in) is not closed; at 6.65 it is +1.54% versus the prior weekly close, but no weekly-close conclusion can be drawn from an unfinished bar.

View: constructive while above 6.61; a settle through 6.69 shifts the bias from base-building to trend continuation toward 6.93.

3. Supply-Demand Balance & Fundamental Drivers

Visible copper inventories are the cleanest signal in the current feed, and they are not building. LME warehouse stock was 244,900 MT as of 2026-10-05, down 3,750 MT (-1.51%) from 248,650 MT on 2026-10-02; the twenty-report change is +8,425 MT, so the draw is recent rather than a sustained trend — the market has rebuilt some stock over the past month and is now giving a little back. SHFE warrants were 10,011 MT as of 2026-09-30, down 721 MT (-6.72%) d/d, a small absolute base but a sharp percentage move that signals tightness at the Chinese exchange level. COMEX registered stocks were 471,400 short tons as of 2026-10-02, unchanged d/d — the US delivery point is stable, neither a drain nor a glut.

Read together: the exchange-visible stock picture is balanced with a tightening bias at the margin. LME draws plus SHFE warrant declines outweigh the flat COMEX book. That is consistent with a market that can hold a 6.35–6.93 range without needing a demand shock to defend the lower bound.

Supply-side headlines over the past 48 hours reinforce the same theme. BHP is seeking formal mediation at Escondida, temporarily delaying a strike at the world's largest copper mine (SMM_EN, 2026-10-06 12:00) — a reminder that labour risk at tier-one assets is live, even if the immediate disruption has been deferred. Separately, the Puquios copper project secured authorisation to begin key water infrastructure in Chile (SMM_EN, 2026-10-06 15:08), and Canterra grew its Newfoundland copper resource 55% while grades fell (Mining_Copper, 2026-10-06 14:48) — both are medium-term supply additions, not near-term tonnage. The broader mining tape was hit by an energy shock that wiped $264 billion off top mining stocks (OilPrice_Main, 2026-10-06 09:00), a cost-side and sentiment headwind rather than a physical copper signal.

Macro transmission: the US 10-year yield at 5.269 (-0.79%) and DXY at 101.85 (-0.32%) [2026-10-06] are both marginally supportive for dollar-denominated copper. A softer dollar and stable-to-lower real yields reduce the headwind to industrial metals. The copper/gold ratio at 1.6 (98th percentile, 1Y) says the market is pricing copper strength relative to the defensive metal — a pro-growth configuration.

View: inventories are the swing factor and they are not bearish; the Escondida mediation headline keeps a supply-risk bid under the market. Bias stays constructive above 6.58.

4. Positioning & Fund Flows

CFTC managed-money data show net length at 78,058 contracts as of 2026-09-29, down 4,464 w/w from 82,522 on 2026-09-22. The prior week had seen a large +17,416 build, so the latest week is a partial give-back rather than a reversal. Open interest is essentially flat: 301,201 on 2026-09-29 versus 301,657 on 2026-09-22. Longs fell to 92,094 from 96,421; shorts fell to 14,036 from 13,899 — the net reduction came from long liquidation, not fresh shorting.

Crowding: netPct 25.92% as of 2026-09-29, crowding 68.51 on the three-year percentile. That is elevated but not extreme — well short of the >90 zone that would flag a stretched book. The CTA trend-following proxy is +62, unchanged across the last four weeks, indicating systematic length is stable rather than adding. Hedge pressure is 54.82%, down from 56.61% the prior week.

Divergence check: price fell over the twenty-day window (-2.55%) while net length also fell w/w (-4,464) — that is positioning following price, not a divergence. Price up plus open interest up would be active buying; here open interest is flat and net length is modestly lower, which is consolidation, not distribution.

Implied versus realized: RV20 is 25.8%. The available implied-vol proxies are cross-asset — ^OVX 48.79 (43rd percentile), ^GVZ 22.97 (14th percentile), ^VXSLV 37.19, ^VIX 15.01 (12th percentile) [2026-10-06]. The broad message is that equity and gold optionality are cheap relative to their own histories, while energy vol is mid-range. There is no copper-specific IV in the feed, so the read is indirect: a low-VIX, low-GVZ regime is generally supportive for carry and trend strategies in industrial metals.

View: positioning is a tailwind, not a constraint. The 68.51 crowding percentile leaves room for further managed-money length without forcing a squeeze. Bias constructive.

5. Cross-Asset Relative Value

The key ratio in the feed is the copper/gold ratio at 1.6 (ratio × 1000), with a one-year percentile of 98.41% and a three-year percentile of 57.94%. The one-year reading says copper has been exceptionally strong versus gold over the past twelve months; the three-year reading says that strength is closer to the middle of the longer distribution. The correct interpretation is that the current configuration is pro-growth — copper outperforming the defensive metal — but the one-year percentile is high enough that the ratio itself is not a fresh buy signal; it is confirmation of the regime, not a timing tool.

Dollar and rates: DXY 101.85 (-0.32%) and ^TNX 5.269 (-0.79%) [2026-10-06]. A softer dollar is a mechanical tailwind for copper priced in USD, and the move lower in the 10-year yield reduces the discount-rate headwind for industrial demand expectations. Neither move is large enough to be a standalone driver, but both lean the same way as the copper call.

Term structure: the curve is in contango, M1-M2 at -0.026 (-0.39%), with a roll yield of -4.68% and slope 0.0316. Contango here reflects carry — the cost of holding and financing metal — not a bearish physical signal. For a long, it is a roll cost of roughly 4.7% annualised, which is a real drag on a multi-month hold and argues for keeping the horizon in the 1–3 week window rather than carrying the position indefinitely.

View: cross-asset configuration (soft dollar, lower yields, high copper/gold percentile) supports the long, but the contango roll cost caps the appropriate holding period. Bias constructive, horizon disciplined.

6. Historical & Seasonal Patterns

Seasonality for the same calendar start over the next twenty sessions, last fifteen years: mean +1.52%, median +1.76%, up in 9 of 15 years. The best instance was 2011 at +9.08%; the worst was 2012 at -7.62%. The distribution is positively skewed at the mean but with a fat left tail — the 2012 outcome shows the window can deliver a sharp drawdown.

Context: a 60% hit rate with a median gain of +1.76% over twenty sessions is a mild tailwind, not a dominant signal. Applied to the 6.65 settle, the median path implies roughly 6.77 over the window, which sits just above R1 6.69 and below R2 6.72 — consistent with the base case of grinding higher rather than a breakout. The small sample (15 years) means the seasonal should be a tiebreaker, not a primary driver.

View: seasonality leans mildly long and supports the base case of a grind toward 6.72–6.77, but it does not justify ignoring the 6.58 invalidation.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 55% — grind higher toward 6.72–6.77. Trigger: price holds above S1 6.61 and settles through R1 6.69 on a daily close. Target: R2 6.72 first, then the seasonal median path near 6.77. Action: maintain long exposure, add on a confirmed settle above 6.69, trail stops behind 6.58. This scenario is consistent with the section 1 call: the two-session recovery off 6.51 continues, inventories stay neutral-to-tight, and positioning adds modestly without crowding out.

Bull case — 25% — breakout toward 6.93. Trigger: a daily settle above 6.72 accompanied by a further LME draw (below 244,900 MT) or a fresh Escondida escalation. Target: the 20-day high at 6.93, the top of the 52-week range. Action: hold the core long, take partial profit into 6.9–6.93, and raise the stop to breakeven. The bull case requires a catalyst — either a supply headline or a demand-side surprise from the Chinese CPI/PPI prints on 2026-10-14.

Bear case — 20% — failure at 6.69 and a break of 6.58. Trigger: rejection at R1 6.69 followed by a daily settle below S2 6.58, which would put the 6.51 weekly low back in play. Target: 6.51, then the 20-day low at 6.35. Action: exit longs on the 6.58 settle, stand aside, and reassess at 6.51. The bear case is driven by a positioning give-back extending beyond the -4,464 w/w seen on 2026-09-29, or a macro shock that lifts the dollar and yields together.

Probabilities sum to 100%. The base case agrees with the section 1 call. The bear case is the invalidation path, not an alternative conclusion.

8. Trading Strategies & Risk Management

Strategy 1 — Core long (HGZ26.CMX reference). Entry 6.65 (at the settle / current Asia level), stop 6.57 (below S2 6.58 and roughly 0.6× ATR14 of 0.127 beyond the level), target 6.72 (R2), horizon 1–5 days, conviction 7. Size: half of intended full position, adding the second half only on a daily settle above 6.69. Risk per unit is 0.08, or about 1.2% of entry.

Strategy 2 — Continuation long on strength. Entry on a daily settle above 6.69, stop 6.58 (S2), target 6.93 (20-day high), horizon 1–3 weeks, conviction 6. Size: the remaining half of the intended position. This leg accepts a wider stop in exchange for the higher target.

Risk management: total exposure should be sized so that a move to the 6.58 invalidation costs no more than the portfolio's standard single-asset risk budget. The contango roll cost of -4.68% annualised argues against holding beyond the 1–3 week horizon. Do not add below 6.61; do not average down through 6.58. If the FOMC minutes on 2026-10-08 (BJT 02:00) produce a dollar and yield spike, reduce to the core half and let the 6.58 stop define the outcome.

9. This Week's Data Calendar

2026-10-07 — EIA Crude Oil Stocks Change (BJT 22:30 | ET 10:30), medium impact; EIA Gasoline Stocks Change (BJT 22:30 | ET 10:30), medium impact.
2026-10-08 — FOMC Meeting Minutes (BJT 02:00 | ET 2026-10-07 14:00), high impact; FOMC Member Waller Speaks (BJT 16:30 | ET 04:30), medium impact.
2026-10-14 — China CPI y/y and PPI y/y (BJT 09:30 | ET 2026-10-13 21:30), high impact for copper.

The FOMC minutes and the China inflation prints are the two events most likely to move the dollar and the demand narrative respectively.

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 2026-10-07

Gold (GC=F) Deep Dive | 2026-10-07 — Fading Crowded Longs Into Support, Tactical LONG | 2026-10-07

Copper (HG=F) Deep Dive | 2026-10-07 — Constructive Base, Long Bias Into 6.72/6.93 | 2026-10-07

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