📅 2026-10-01 · COMEX金 · 銘柄別深度レポート

本レポート提供元 MacroTerms.com

データ基準日: 2026-09-30 · 決済基準: 取引所最終日次決済価格

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

Call: LONG gold from the 4,143–4,160 support shelf, with a settle below 4,132 (S2) invalidating the view. Three reasons: (1) price has already retraced to the bottom of its 20-day channel — settle 4,186.7 sits at the 11th percentile of the 4,143.1–4,558.5 range, with the 5D decline of 3.05% and 20D decline of 4.77% (settle) leaving the market stretched into support rather than extended; (2) the options market is not pricing panic — GVZ at 23.74 sits in the 20th percentile of its 1-year range and only 3.3 vol points above RV20 of 20.4%, so downside protection is cheap and the crowd is not hedging; (3) the 20-session seasonal window from this calendar date has been positive in 11 of the last 15 years, with a median gain of 1.48%. The offsetting risk is positioning: managed-money net is 127,389 contracts, a 92.46 percentile 3-year crowding reading, and has fallen for four consecutive weeks. That argues for a tactical, level-based long rather than a trend entry.

1. Bottom Line & Directional Bias

Call: LONG gold, invalidation on a settle below 4,132 (S2).

The prior session settled at 4,186.7 (CME Group final daily settlement, 2026-09-30), up 0.17% on the day but down 3.05% over five sessions and 4.77% over twenty. That places the settle at the 11th percentile of the 20-day 4,143.1–4,558.5 channel — the market is at the floor of its recent range, not mid-range.

Three supports underpin the long. First, the 4,143.1 20-day low and the 4,132.4 S2 pivot form a defined shelf roughly one ATR (93.9, or 2.24% of price) below the settle, giving a tight, testable invalidation. Second, implied volatility is not stressed: GVZ at 23.74 sits in the 20th percentile of its one-year range and only 3.3 vol points above realized RV20 of 20.4%, so the market is not paying up for downside. Third, the seasonal window beginning at this calendar date has delivered a positive 20-session return in 11 of the last 15 years, median +1.48%.

The principal counterweight is crowding: managed-money net length of 127,389 contracts is a 92.46 percentile reading on a three-year window and has contracted for four straight weeks. That caps upside velocity and argues for a level-based entry rather than a momentum chase. A settle below 4,132 kills the setup.

2. Price Action & Technical Analysis

The settle of 4,186.7 (2026-09-30) is the anchor for every level below. The 1D change of +0.17% is a stabilization print after a heavy run: 5D -3.05% and 20D -4.77% (both settle-based). The 20-day channel runs 4,143.1 to 4,558.5, and the settle sits at the 11th percentile of that band — the lower decile. The 52-week range is 3,793.4–5,586.2, so the market is roughly 25% below its 52-week high, consistent with the 25.06% 52-week maximum drawdown reading.

ATR14 is 93.9, equal to 2.24% of price as a full daily range — not a one-sided figure. RV20 is 20.4% annualized. The ratio of ATR to price tells us a normal session spans roughly 4,093 to 4,281 around the current settle, which is why the 4,132.4 S2 pivot is the correct invalidation rather than a tighter level: anything inside one ATR is noise.

Pivots from the settle: P 4,205.3, R1 4,232.5, R2 4,278.2, S1 4,159.6, S2 4,132.4. The settle at 4,186.7 sits below the pivot P, which is a mildly bearish intraday posture, but above S1 — the first support has held. The actionable structure is the 4,159.6–4,143.1 shelf (S1 to the 20-day low), with 4,132.4 as the hard floor.

In early Asian trade on 2026-10-01 (07:00), gold last printed 4,184.4, -0.05% versus the settle, with a session range of 4,183.9–4,192. That is a tight, low-conviction Asian range — no directional information, but it confirms the market is holding above S1 into the European open.

The last completed weekly bar (2026-09-21 to 2026-09-25) opened 4,413, high 4,422.1, low 4,278.3, closed 4,321.2, down 2.34% w/w. The current week beginning 2026-09-28 is unfinished — three sessions in, last 4,186.7, down 3.11% — and no weekly-close conclusion can be drawn from it. The completed weekly bar is unambiguously bearish, which is precisely why the tactical long is a counter-trend, support-based trade with a tight stop rather than a trend-following entry.

View: tactical long against 4,132.4, first objective the 4,205.3 pivot, then 4,232.5 R1.

3. Supply-Demand Balance & Fundamental Drivers

COMEX registered gold stood at 15.14 Moz (470,907 kg) on 2026-09-29, up 1,244 oz (+0.3%) day on day, having been 15.1 Moz (469,663 kg) on 2026-09-28. The two-day build is marginal in absolute terms but directionally notable: registered stocks are rising into a price decline, which is consistent with metal moving into the deliverable pool rather than being drawn out of it. SHFE warrants were 116,028 kg on 2026-09-30, down 3 kg day on day and essentially flat versus 116,031 kg on 2026-09-29. Chinese exchange inventory is not signalling physical tightness in either direction.

The term structure is in contango: M1–M2 at -15.5 (-0.37%), with roll yield of -4.44% and a slope reading of 22.17. This is carry, not a directional signal — it reflects the cost of financing and storage against the risk-free rate, and it does not cap the price. It does mean longs pay a roll cost, which is a modest drag on multi-week holds and a reason to prefer the front month, GCZ26 on COMEX, for tactical exposure.

The macro transmission channel runs through rates and the dollar. The US 10-year yield (^TNX) printed 5.293, up 0.72%, and DXY was 101.46, up 0.09% (both 2026-09-30). A 5.29% ten-year is a genuine headwind for a zero-coupon asset, and it explains the 20-day drawdown better than any physical-market story. The counter is that the dollar's move was marginal (+0.09%) while gold's five-day decline was 3.05% — the selloff is more about positioning and rate expectations than about a broad dollar squeeze.

Headline flow is mixed and should be weighted lightly. Morgan Stanley's Gower is reported as seeing $4,000 as a strong floor with three supports cited (ForexLive, 2026-09-30), which is directionally consistent with the 4,132–4,143 shelf but sits well below it. CME reported gold options volume flat in September as price ended about 9% below the August peak (ForexLive, 2026-10-01) — flat volume into a decline is not capitulation. Mine-level news (Gold Fields halting Windfall work on a Quebec permit lapse; Arizona Gold and Miata drill results) is single-asset equity news with no aggregate supply impact.

View: physical balances are neutral-to-soft, contango is a carry cost not a cap, and the binding constraint is the 5.29% ten-year. Long bias holds while rates stabilize.

4. Positioning & Fund Flows

CFTC managed-money positioning has deteriorated for four consecutive weeks. Net length fell from 136,771 on 2026-09-01 to 134,972 (09-08), 133,116 (09-15) and 127,389 (09-22). The weekly deltas were -7,976, -1,799, -1,856 and -5,727. Open interest over the same span moved 415,196 → 411,227 → 409,899 → 412,800, so the net reduction is driven by long liquidation (longs 149,721 → 135,699) rather than by fresh short selling (shorts 12,950 → 8,310, i.e. shorts also covered).

This is the key nuance. Longs are leaving and shorts are leaving — a two-sided de-risking, not a bearish repositioning. The crowding metric nonetheless reads 92.46 (3-year percentile of net length as a share of open interest) on 2026-09-22, down only marginally from 92.58 the prior week. Net length as a share of OI is 30.86%. By the desk's convention, a 92nd-percentile reading on a three-year window is crowded, and the report should say so plainly: this market is still long and crowded even after four weeks of trimming. CTA trend-following proxy sits at 62, unchanged across all four weeks, and hedge pressure is 15.07%.

The implication is asymmetric. Crowded longs that have already begun liquidating can extend the move lower on a support break, which is why the 4,132 stop is non-negotiable. But the same crowding means that if 4,143 holds, the marginal seller is exhausted and a short-covering plus re-allocation bid can be sharp.

On volatility, GVZ at 23.74 (1Y percentile 20%) versus RV20 of 20.4% gives IV−RV of +3.3 vol points, an IV/RV ratio of 1.16. Options are paying a modest premium to realized, but the absolute level of implied vol is in the bottom quintile of the year. That is an environment where long optionality is cheap relative to history, and it supports expressing the long via defined-risk structures. Silver implied vol (^VXSLV) at 37.87 remains far above gold's, and VIX at 16.34 (33rd percentile) shows no broad risk-off impulse.

View: crowded but de-risking; long bias retained, size reduced, stop hard.

5. Cross-Asset Relative Value

The copper/gold ratio stands at 1.58 (ratio ×1000), in the 96th percentile of its one-year range and the 53rd percentile of three years. A high and elevated copper/gold ratio is a pro-growth signal, and its one-year percentile near the top of the range says the market has spent the past year pricing industrial demand over defensive metal. That is a relative headwind for gold, though the three-year percentile at roughly the middle of the distribution shows this is a one-year phenomenon, not a structural regime.

The gold/silver ratio is 69.13, in the 73rd percentile of one year and the 24th percentile of three years. The correct reading: over three years silver has been structurally strong versus gold (low percentile of the ratio), while over the past year silver has lagged (high percentile). The ratio is not at an extreme in either window, so it offers no strong relative-value signal — but the one-year drift toward a higher ratio means silver has been the weaker leg recently, and a gold long is the cleaner expression of the metal complex.

The oil/gold ratio is 0.0216, in the 87th percentile of one year and 45th of three years. A high oil/gold ratio means oil has been relatively strong versus gold over the past year — again a pro-cyclical tilt and a mild relative negative for gold. WTI implied vol (^OVX) at 52.24, 52nd percentile, is unremarkable.

Taken together, the cross-asset matrix says the past year has rewarded cyclical exposure over defensive metal. That is a reason to keep gold position sizing moderate, not a reason to abandon the support-based long.

View: relative-value backdrop is neutral-to-negative for gold; keep the trade tactical.

6. Historical & Seasonal Patterns

Using the same calendar start date and a 20-session forward window over the last 15 years, gold has been positive in 11 of 15 years, with a mean return of +1.5% and a median of +1.48%. The best instance was 2023 at +9.08%; the worst was 2012 at -3.54%.

The distribution is informative in two ways. First, the mean and median are nearly identical (+1.5% versus +1.48%), which means the result is not driven by a single outlier year — the typical outcome and the average outcome agree. Second, the worst case at -3.54% is roughly 1.6 ATRs of downside over 20 sessions, which is a manageable tail relative to the 4,132 invalidation, itself only about 1.3% below the settle.

The sample is small — 15 observations — and seasonality should be treated as context, not as a driver. It does not override the crowded-positioning risk or the 5.29% ten-year yield. What it does is tilt the probability distribution modestly in favour of the long over a 20-session horizon, which is the horizon the trade is sized for.

View: seasonal tailwind is real but modest; it supports the long without justifying size above normal.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 50%: range repair, settle drifts to 4,205–4,233. The 4,143.1 20-day low holds on a closing basis, the 4,159.6 S1 pivot is defended, and gold grinds back to the 4,205.3 pivot and then 4,232.5 R1 over one to two weeks. Trigger: a daily settle back above 4,205.3 with GVZ holding below 25. Action: hold the long entered at 4,160–4,187, trail the stop to breakeven once 4,205 prints. This is the path consistent with the section 1 call.

Bull case — 30%: short-covering squeeze through R1. A soft Non-Farm Payrolls print (forecast 89K versus prior 162K, surprise threshold ±73K) on 2026-10-02, or a dovish FOMC minutes read on 2026-10-08, forces the crowded short-term book to cover. Price clears 4,232.5 R1 and extends to 4,278.2 R2, with a stretch objective at the 4,300 area. Trigger: a settle above 4,232.5 accompanied by a decline in the 10-year yield below 5.2%. Action: add on the R1 break, move the stop to 4,186, take partial profit at R2.

Bear case — 20%: support failure, liquidation extends. A hot NFP print or a hawkish Waller commentary (2026-10-01, 22:00 BJT) pushes the ten-year yield higher and triggers the fifth consecutive week of long liquidation. Price settles below 4,132.4 S2, opening 4,093 (one ATR below the settle) and then the 4,000 psychological level cited in press commentary. Trigger: a daily settle below 4,132.4. Action: exit the long on the settle, stand aside, and do not re-engage until price reclaims 4,159.6 S1.

Probabilities sum to 100%. The base case agrees with the section 1 long call. The bear case is the invalidation path and is fully defined by a single level.

8. Trading Strategies & Risk Management

Strategy 1 — Tactical long from the support shelf. Entry 4,160–4,187 (scale in across the S1 pivot and the current settle area), stop 4,128 (below the 4,132.4 S2 pivot and beyond one ATR of noise), target 4,232 (R1), extended target 4,278 (R2). Horizon 1–5 days. Conviction 6/10. Size at 60% of a normal gold position given the 92.46 crowding percentile. Instrument: COMEX December gold, GCZ26.

Strategy 2 — Defined-risk long via options. With GVZ at 23.74 in the 20th percentile of its one-year range and IV only 3.3 vol points above RV20, buy a 4,200-strike call financed partly by selling a 4,300-strike call, 2–3 week tenor. Maximum loss is the net premium; the structure removes the gap risk that a futures stop cannot fully control around the 2026-10-02 payroll print. Horizon 2–3 weeks. Conviction 5/10. Size at 40% of a normal position.

Risk management: the single hard rule is the 4,132.4 settle-based stop on the futures leg. Do not average down below S1. Reduce total exposure by half ahead of the 2026-10-02 20:30 BJT Non-Farm Employment Change release, given the ±73K surprise threshold. If the ten-year yield breaks above 5.4%, cut the futures leg regardless of the gold price level.

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, prior 54.6, surprise outside ±0.2) and ISM Manufacturing Employment (forecast 51.5, prior 51.2, ±0.3). BJT 10-02 20:30 | ET 10-02 08:30 — Non-Farm Employment Change (forecast 89K, prior 162K, ±73K), Average Hourly Earnings m/m (0.3% versus 0.3%, ±0.1%), Unemployment Rate (4.1% versus 4.1%, ±0.1%). BJT 10-05 22:00 | ET 10-05 10:00 — ISM Services PMI (forecast 54, prior 55.4, ±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.

銘柄別深度レポート深度分析

COMEX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

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

本日のリスクマップ

銘柄別深度レポート 深度分析

過去のレポート

⟠ 今月の戦略バックテスト

銘柄別深度レポート深度分析