📅 2026-10-03 · COMEX精铜 · 单品种早报旧覆盖策略

数据截止: 2026-10-02 · 结算口径: 交易所最终日结算价

Copper (HG=F) Deep Dive | 2026-10-03 — Long into the seasonal window, 6.46 invalidation | 2026-10-03

Call: LONG copper, with the trade invalidated on a settle below 6.46 (S2). Three reasons: (1) the pullback is a positioning flush, not a fundamental break — COMEX registered stocks were unchanged d/d at 471,400 short tons on 2026-09-30, LME warehouse stock fell 1,325 MT d/d to 248,075 MT on 2026-10-01, and SHFE warrants dropped 721 MT d/d to 10,011 MT on 2026-09-30, so visible metal is not building; (2) the 5D decline of 3.21% to a 6.55 settle has left price at the 35th percentile of the 20-day 6.35–6.93 channel with ATR14 at 0.13 (1.99% of price) and RV20 at 26.5%, a mean-reversion setup rather than a trend break; (3) the same 20-session calendar window has been up in 10 of the last 15 years, median +1%, mean +1.06%. Managed-money net length rose 17,416 lots to 82,522 in the 2026-09-22 CFTC report, with crowding at the 69th percentile — elevated but not extreme. The copper/gold ratio at 1.56 sits in the 90th percentile of its 1-year range, confirming copper is not the weak leg of the complex.

1. Bottom Line & Directional Bias

Call: LONG copper (HG=F), invalidation on a daily settle below 6.46.

Three reasons underpin the call. First, the physical ledger does not corroborate the sell-off: COMEX registered stocks were unchanged d/d at 471,400 short tons on 2026-09-30, LME warehouse stock fell 1,325 MT d/d to 248,075 MT on 2026-10-01, and SHFE warrants fell 721 MT d/d to 10,011 MT on 2026-09-30. A 3.21% five-day decline into flat-to-falling visible inventory is a positioning event, not a balance-sheet event. Second, the technical configuration is stretched but intact: the 6.55 settle sits at the 35th percentile of the 20-day 6.35–6.93 channel, with ATR14 at 0.13 (1.99% of price) and RV20 at 26.5% — the move lower has been orderly, not disorderly. Third, the seasonal window is supportive: the same calendar start over the next 20 sessions has been positive in 10 of the last 15 years, median +1%.

The invalidation is explicit and level-based: a daily settle below 6.46 (S2) would signal that the 20-day channel floor at 6.35 is the next magnet and that the positioning flush has become a trend. Until then, the asymmetry favours the long side from the lower third of the range.

2. Price Action & Technical Analysis

Copper settled at 6.55 on 2026-10-02, up 0.17% on the day (settle). The five-day change is -3.21% and the twenty-day change is -1.73% (settle), so the bulk of the damage was done in the most recent week. The 20-day channel runs 6.35–6.93, placing the settle at the 35th percentile — lower third, but not at the floor. The 52-week range is 4.83–6.93, so price remains in the upper half of the annual distribution despite the pullback.

Volatility is moderate. ATR14 is 0.13, equal to 1.99% of price as a full daily range. RV20 is 26.5% annualized. For context, ^VXSLV at 36.9 and ^OVX at 51 imply that copper's realized volatility is the lowest of the three industrial/monetary complexes, while ^GVZ at 23.23 (15th percentile of its 1-year range) and ^VIX at 15.31 (15th percentile) show that macro event pricing is subdued. Copper is not being priced as a crisis asset.

Pivots from the settle-based snapshot: P 6.57, R1 6.61, S1 6.5, R2 6.68, S2 6.46. The settle at 6.55 sits just below the pivot, which is the mechanical reason the tape feels heavy — but S1 at 6.5 is only five ticks away and has held. A reclaim of P 6.57 opens R1 6.61 and then R2 6.68; a loss of S1 6.5 puts S2 6.46 in play, which is the invalidation line.

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. That is the most recent closed weekly reference and it was constructive. The current week (from 2026-09-28, five sessions) is unfinished and shows 6.55, -3.21%; no weekly-close conclusion can be drawn from it. The early Asian snapshot on the report date is not a settlement and is not used for levels here.

View: the tape is corrective within an intact 20-day range. Bias stays long while 6.46 holds on a settle basis; the first confirmation is a settle back above P 6.57.

3. Supply-Demand Balance & Fundamental Drivers

Visible inventory is the cleanest signal in the current tape, and it is not bearish. COMEX registered stocks stood at 471,400 short tons on 2026-09-30, unchanged d/d. LME warehouse stock was 248,075 MT on 2026-10-01, down 1,325 MT d/d (-0.5%), though the 20-report change is +13,425 MT — so LME has rebuilt over the past month even as the daily trend has turned down. SHFE warrants fell 721 MT d/d (-6.7%) to 10,011 MT on 2026-09-30, a very low absolute level that leaves the Shanghai deliverable pool thin.

The read-through: the +13,425 MT LME rebuild is the bearish fact in the block, and it is the reason a 3.21% five-day decline is not obviously wrong. But the marginal direction over the last two sessions is draws across LME and SHFE, with COMEX flat. A market that is drawing metal at the margin while price falls 3.21% is a market where the selling is financial, not physical.

Supply-side headlines over the last 48 hours are consistent with a structurally tight mine pipeline rather than a near-term supply surge. Ghana's draft bill proposes a state special share and shorter mining leases (SMM_EN, 2026-10-02), a jurisdiction-risk headline that raises the cost of future supply. Cascadia Intersected 75.55 m at 1.19% Cu at the Carmacks Copper-Gold Project (SMM_EN, 2026-10-02) is an exploration result, years from production. The World Bank unlocking $1.4B for Chile's copper giant (Mining_Copper, 2026-10-02) is a financing headline that supports future capacity but does nothing for 2026–27 tonnage.

Macro transmission is through the dollar and rates. DXY at 101.92, -0.17% on 2026-10-02, and ^TNX at 5.277, +0.76%, describe a market where the dollar is soft but the long end is backing up. A soft dollar is a mild tailwind for dollar-denominated copper; a 5.28% ten-year is a headwind for the whole complex via discount rates and inventory financing costs. The term structure shows CONTANGO M1-M2 at -0.026 (-0.39%), roll yield -4.73%, slope 0.0388 — a shallow contango that is a carry cost for longs, not a signal of surplus. It does not cap price.

View: the fundamental ledger is neutral-to-constructive. The LME 20-report rebuild is the risk; the SHFE draw and flat COMEX are the support. Net, the balance does not justify a sustained break below the 20-day floor.

4. Positioning & Fund Flows

The CFTC data is as of 2026-09-22 and is 11 days old — it is a lagging input, not a current-week read. In that report, open interest was 301,657, longs 96,421, shorts 13,899, net 82,522, a weekly change of +17,416. The prior week (2026-09-15) showed net 65,106, a change of -17,048. So the managed-money book swung from a 17k reduction to a 17k addition in one week, with open interest rising from 289,463 to 301,657.

Crowding metrics: netPct 27.36%, crowding 69.05 (3-year percentile), CTA 98, hedge 56.61%. The 69th percentile is elevated but well short of the extreme zone — this is not a crowded long by the multi-year standard, and the CTA reading of 98 shows trend-following proxies are already positioned long. That is a double-edged fact: it confirms the trend, but it also means CTAs are a source of supply if the 20-day floor gives way.

The divergence to note is between the 2026-09-22 positioning report (net length rising, +17,416) and the subsequent price action (5D -3.21%). The most likely explanation is that the long build into 2026-09-22 was followed by a partial flush in the last week of September, which is exactly the setup that creates mean-reversion opportunities when the physical ledger does not deteriorate.

Implied versus realized: copper's RV20 is 26.5%, while ^VXSLV is 36.9 and ^OVX is 51. Copper optionality is not being priced for a crisis. With ^VIX at 15.31 (15th percentile) and ^GVZ at 23.23 (15th percentile), the macro vol complex is cheap, which argues against a sustained copper breakdown on macro fear alone.

View: positioning is a headwind only if 6.46 breaks. Above it, the 2026-09-22 long build is a tailwind, not a warning.

5. Cross-Asset Relative Value

The copper/gold ratio is 1.56 (ratio ×1000), at the 90th percentile of its 1-year range and the 49th percentile of its 3-year range. The one-year reading is the important one: copper has been outperforming gold over the past twelve months, and the ratio sits near the top of that window. A rising copper/gold ratio is a pro-growth signal, and the current level says the market has not priced a growth scare into the industrial complex even as copper sold off 3.21% over five sessions.

The 3-year percentile at 48.68% is the caution flag: relative to the longer window, copper/gold is mid-range, so the 1-year strength is a recovery, not a structural breakout. That argues for a tactical long rather than a strategic overweight.

Dollar and rates complete the picture. DXY at 101.92, -0.17%, is a mild tailwind. ^TNX at 5.277, +0.76%, is a headwind for all real assets. The combination — soft dollar, firm long end — historically favours copper over gold on a relative basis, because copper's demand signal is more sensitive to nominal growth than to real rates.

View: cross-asset confirms the long bias. Copper is not the weak leg; the ratio's 90th-percentile 1-year position says the market still believes in industrial demand.

6. Historical & Seasonal Patterns

Using the same calendar start and the next 20 sessions, the last 15 years show a mean return of +1.06%, a median of +1%, and an up-year hit rate of 10 out of 15. The best year in the sample was 2011 at +15.57%; the worst was 2012 at -8.2%.

The distribution is positively skewed: the best year is nearly twice the magnitude of the worst year, and the hit rate is 67%. That is a favourable seasonal backdrop for a long positioned in the lower third of the 20-day range.

The caveat is sample size. Fifteen observations is a small sample, and the 2012 worst case of -8.2% is a reminder that the seasonal window can fail violently when macro conditions dominate. The seasonality block is context, not a standalone reason for the trade — it is the third leg of the argument, behind the inventory ledger and the technical position.

View: seasonality adds a modest positive tilt to the long. It does not override the 6.46 invalidation.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 55% — grind higher from the lower third of the range. Trigger: a settle back above P 6.57, confirming that S1 6.5 has held. Target: R1 6.61 first, then R2 6.68. Action: hold the long, add on a 6.57 settle, trail the stop behind 6.5. This is the path consistent with the section 1 call: the physical ledger does not deteriorate, the positioning flush exhausts, and the seasonal window delivers a median +1% over 20 sessions.

Bull case — 25% — range breakout. Trigger: a settle above R2 6.68 with LME warehouse stock continuing to draw and SHFE warrants staying near 10,011 MT. Target: a retest of the 20-day high at 6.93, the top of the 52-week range. Action: add to the long on the breakout close, move the stop to 6.57 (the pivot), and let the position run into the 6.93 retest. The catalyst would be a soft ISM Services print on 2026-10-05 or a dovish FOMC Minutes read on 2026-10-08.

Bear case — 20% — channel floor failure. Trigger: a daily settle below S2 6.46, with LME's 20-report change of +13,425 MT extending and COMEX registered stocks at 471,400 short tons beginning to build. Target: the 20-day floor at 6.35, then the 52-week midpoint zone. Action: exit the long on the 6.46 settle, stand aside, and re-engage only on a reclaim of 6.5. The bear case is the minority path because the daily inventory trend is draws, not builds, and because copper/gold at the 90th percentile of its 1-year range argues against a growth scare.

Probabilities sum to 100%. The base case agrees with the section 1 call.

8. Trading Strategies & Risk Management

Strategy 1 — tactical long from the lower third of the range. Entry 6.55 (current settle), stop 6.42 (below S2 6.46, beyond the invalidation line and roughly one ATR14 of 0.13 away), target 6.68 (R2), horizon 1–5 days, conviction 7. Size at half normal risk budget given the 5D -3.21% momentum against the position. The trade works if S1 6.5 holds on a closing basis and P 6.57 is reclaimed.

Strategy 2 — add on confirmation. Entry 6.58 on a settle above P 6.57, stop 6.45, target 6.93 (20-day and 52-week high), horizon 5–15 days, conviction 6. This leg is sized at the remaining half of the risk budget and is only triggered if the base case confirms. The wider target reflects the seasonal window's median +1% over 20 sessions and the 2011 best case of +15.57%.

Risk management: total exposure across both legs should not exceed a normal single-asset risk budget. The invalidation is a daily settle below 6.46 — if that prints, both legs are closed regardless of the stop level. The 2026-09-22 CFTC report is 11 days old and should not be used to size up; crowding at the 69th percentile leaves room, but the CTA reading of 98 means trend followers are already long and can accelerate a downside move if the floor breaks.

9. This Week's Data Calendar

ISM Services PMI for September on 2026-10-05 at 22:00 BJT / 10:00 ET, forecast 54 versus prior 55.4, a surprise if outside 54 ± 1.4 — relevant to copper via DXY. FOMC Minutes on 2026-10-08 at 02:00 BJT / 2026-10-07 14:00 ET, the week's highest-impact event for the dollar and the metals complex. API and EIA crude stock changes on 2026-10-07 at 04:30 and 22:30 BJT respectively are energy-specific and second-order for copper.

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.

单品种早报深度分析

COMEX黄金GC=F 2026-10-03

Gold (GC=F) Deep Dive | 2026-10-03: Bearish While Below 4191.7 Pivot | 2026-10-03

Copper (HG=F) Deep Dive | 2026-10-03 — Long into the seasonal window, 6.46 invalidation | 2026-10-03

今日风险地图

单品种早报 深度分析

往期早报

⟠ 本月策略回测表现

单品种早报深度分析