Gold (GC=F) Deep Dive | 2026-10-05 — Selloff Exhausted at 20D Floor, Tactical Long | 2026-10-05

Call: Bullish gold on a tactical 1–5 day horizon, with invalidation on a settle below 4086.5 (S2). The prior session settled at 4162.3, down 3.68% over five sessions and 8.32% over twenty, yet that settle sits in the bottom 4.9% of the 20-day 4143.1–4537.8 channel — the weakest possible position inside a range that has already absorbed the entire managed-money reduction. Three supports underpin the call: (1) the 20-day floor at 4143.1 held on the last completed weekly bar (low 4143.1, close 4162.3), and early Asian trade on 2026-10-05 is already 0.33% above settle at 4176.1; (2) positioning has de-risked for four straight weeks, with net length down 14,654 lots from 134,972 to 120,318 while crowding remains 93rd percentile — the flush removed weak hands, not the structural bid; (3) GVZ at 23.23 sits in the 15th percentile of the past year against RV20 of 17%, so optionality is cheap relative to realized movement. The bear case is a clean break of 4086.5, which would open the 52-week low at 3823.7.

1. Bottom Line & Directional Bias

Call: Bullish gold (GC=F) on a tactical 1–5 day horizon. Invalidation: a daily settle below 4086.5 (S2).

Three reasons. First, price location: the 2026-10-02 settle of 4162.3 sits at the 5th percentile of the 20-day 4143.1–4537.8 channel, i.e. directly on the range floor, after a 5D decline of 3.68% and a 20D decline of 8.32%. Selling into the bottom of an established range with ATR14 at 89 (2.14% of price) is a poor risk-reward. Second, positioning: managed-money net length has fallen for four consecutive CFTC weeks, from 134,972 lots (2026-09-08) to 120,318 (2026-09-29), a reduction of 14,654 lots, while the crowding percentile stays at 92.51 — the market has shed momentum length without dismantling the structural long. Third, volatility: GVZ at 23.23 (15th percentile, 1Y) against RV20 of 17% gives an IV−RV spread of +6.2 vol points (IV/RV 1.37), so event protection is priced but not extreme, and the 20-session seasonal window (mean +0.99%, median +0.69%, up 9 of 15 years) leans positive.

Invalidation is a settle below 4086.5. That level is the second pivot support and sits roughly one ATR below the settle; a close beneath it would signal range failure rather than range noise and would shift the bias to neutral-to-short with the 52-week low at 3823.7 as the next reference.

2. Price Action & Technical Analysis

The prior session settled at 4162.3 (2026-10-02), down 0.95% on the day, down 3.68% over five sessions and down 8.32% over twenty. The 20-day channel runs 4143.1–4537.8, placing the settle at the 5th percentile — effectively the floor. The 52-week range is 3823.7–5586.2, so the market is trading in the lower third of its annual envelope but well above the annual low.

Pivots from the settle-based snapshot: P 4191.7, R1 4229.6, R2 4296.9, S1 4124.4, S2 4086.5. The settle is below P, which is consistent with the recent downtrend, but it is above S1 and comfortably above S2. The immediate task for bulls is a reclaim of P at 4191.7; a settle above R1 4229.6 would confirm the range-floor bounce and open R2 4296.9.

ATR14 is 89 points, equal to 2.14% of price — this is the full expected daily range, not a one-sided band. RV20 is 17% annualized. In early Asian trade on 2026-10-05 (07:00), gold last printed 4176.1, up 0.33% versus the prior settle, with an Asian session range of 4160.2–4176.6. That is a constructive start: the market is holding above the 20-day floor and above S1.

The last completed weekly bar (2026-09-28 to 2026-10-02) opened 4315, high 4315.6, low 4143.1 and closed 4162.3, down 3.68% week-on-week. The low of that completed week is 4143.1 — the same level as the 20-day channel floor — and the close at 4162.3 is 19 points above it. The current week has no settled bar yet, so no weekly conclusion can be drawn; the relevant fact is that the completed week closed near its low but did not break the 20-day floor.

View: tactical long bias while 4143.1 holds on a settle basis; first objective P 4191.7, then R1 4229.6.

3. Supply-Demand Balance & Fundamental Drivers

COMEX registered gold stood at 15.09 Moz (469,352 kg) on 2026-10-01, unchanged day-on-day. The prior two prints were 15.14 Moz (470,907 kg) on 2026-09-29 and 15.09 Moz on 2026-09-30, so registered stocks have drifted marginally lower across the week — a modest, not dramatic, draw. SHFE warrants were 116,028 kg on 2026-09-30, down 3 kg day-on-day, effectively flat. The read-through: there is no visible physical accumulation or liquidation signal from the warehouse data; the recent price decline is a paper-market event, not a physical-market one.

The term structure is in contango: M1–M2 at −15.2 (−0.37%), with roll yield of −4.38% and a slope of 17.87. This is carry, reflecting the level of short rates (US 10-year at 5.277, up 0.76% on 2026-10-02), not a bearish signal. Contango does not cap the price; it is a roll cost for long holders, and it argues for using futures or short-dated exposure rather than a long-dated roll-heavy position.

The macro transmission channel is the rate complex. With the 10-year at 5.277 and DXY at 101.86 (down 0.07% on 2026-10-04), the dollar has not confirmed the gold selloff — a mild positive divergence. The week-ahead FOMC minutes (BJT 10-08 02:00 | ET 10-07 14:00) are the key event for the rates channel; a dovish read would lower real-rate pressure and support the range-floor bounce, while a hawkish read would test S1 4124.4.

View: physical data is neutral, contango is a carry cost not a signal, and the dollar's failure to rally with the gold decline is the most constructive fundamental input this week.

4. Positioning & Fund Flows

CFTC managed-money positioning has de-risked steadily. Net length fell from 134,972 lots on 2026-09-08 to 133,116 (2026-09-15), 127,389 (2026-09-22) and 120,318 (2026-09-29) — four consecutive weekly declines totalling 14,654 lots. Open interest over the same span eased from 411,227 to 406,456. The weekly changes were −1,799, −1,856, −5,727 and −7,071 lots: the reduction is accelerating, which is characteristic of capitulation rather than orderly distribution.

Crowding, measured as the three-year percentile of managed-money net as a share of open interest, remains elevated: netPct 29.6% on 2026-09-29 against a crowding percentile of 92.51. The CTA trend proxy is 62 and the hedge ratio 14.19%. This combination — high crowding percentile but four weeks of net reduction — means the trade is still structurally crowded, but the marginal seller has been active and is closer to exhaustion than to initiation. A crowded long that has already shed 14,654 lots is a materially different risk than a crowded long at its peak.

On volatility, GVZ at 23.23 (15th percentile, 1Y) versus RV20 of 17% gives IV−RV of +6.2 vol points and IV/RV of 1.37. Options are paying up modestly for event risk around the FOMC minutes, but at the 15th percentile of the year, implied vol is not expensive in absolute terms. For directional longs, this favors defined-risk structures over outright leverage.

View: positioning is a headwind that is fading, not a reason to sell; the flow data supports a tactical long into the range floor.

5. Cross-Asset Relative Value

The copper/gold ratio is 1.57 (ratio ×1000), at the 94th percentile of the past year and the 52nd percentile of the past three years. A high one-year percentile means copper has been strong relative to gold — a pro-growth signal — but the three-year percentile near the median shows this is a cyclical normalization, not a structural regime shift. For gold, the read is that the recent decline is gold-specific (positioning and rates) rather than a broad commodity-liquidity event.

The gold/silver ratio is 68.9, at the 71st percentile of the past year and the 24th percentile of the past three years. The one-year percentile is elevated, meaning silver has lagged gold over the past twelve months, while the three-year percentile is low, meaning silver has been structurally strong versus gold over the longer window. This is a mixed signal for gold: near-term, gold has been the relative winner; structurally, silver's strength suggests the precious complex retains a bid.

The oil/gold ratio is 0.0219, at the 90th percentile of the past year and 46th of the past three years — oil has been strong versus gold on a one-year view, consistent with the copper/gold message of cyclical reflation rather than deflation. None of these ratios argues for a gold breakdown; they argue that gold's decline is idiosyncratic and therefore more likely to mean-revert.

View: cross-asset ratios are neutral-to-supportive for a gold bounce; the gold-specific weakness is the opportunity.

6. Historical & Seasonal Patterns

The seasonality block covers the same calendar start, next 20 sessions, over the last 15 years: mean +0.99%, median +0.69%, with the market up in 9 of 15 years. The best instance was 2023 at +9.3%; the worst was 2022 at −3.84%. This is a modest positive edge — a hit rate of 60% with a positive mean and median — and it is context only given the small sample.

The distribution matters more than the average: the worst case (−3.84%) is roughly four times smaller in magnitude than the best case (+9.3%), so the seasonal window is positively skewed. Combined with the current position at the 5th percentile of the 20-day channel, the seasonal backdrop adds a mild tailwind to the tactical long, but it is not the primary driver. The primary drivers remain price location, positioning exhaustion and cheap implied volatility.

View: seasonality is a secondary, mildly supportive input; it does not override the 4086.5 invalidation.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 50%: range-floor bounce. Trigger: gold holds above the 20-day floor at 4143.1 on a settle basis and reclaims the pivot at 4191.7. Target: R1 4229.6, with R2 4296.9 as the stretch. Action: tactical long from the current area, stop below S2 4086.5, scale out into R1/R2. This scenario is consistent with the section 1 call and is supported by the Asian session print of 4176.1 (+0.33% vs settle), the four-week positioning flush, and the 15th-percentile GVZ.

Bull case — 25%: breakout extension. Trigger: a dovish FOMC minutes read (BJT 10-08 02:00 | ET 10-07 14:00) plus a soft ISM Services print (BJT 10-05 22:00 | ET 10-05 10:00, forecast 54, surprise if outside 54 ± 1.4) that pushes the dollar lower and real rates down. Target: a settle above R2 4296.9 opens the mid-range 4400s and re-tests the upper half of the 20-day channel toward 4537.8. Action: add on a confirmed settle above R1 4229.6, trail stops to the pivot.

Bear case — 25%: range failure. Trigger: a hawkish FOMC minutes read or a hot ISM Services print that lifts the 10-year above 5.277 and the dollar through 101.86, combined with a settle below S2 4086.5. Target: S1 4124.4 gives way first, then the 52-week low at 3823.7. Action: exit longs on the S2 break, stand aside, and re-engage only on a reclaim of 4143.1. The bear case is the invalidation path, not an alternative call.

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

8. Trading Strategies & Risk Management

Strategy 1 — Tactical long (primary). Entry 4162–4176 (current area, settle 4162.3 / Asia 4176.1). Stop 4086.5 (below S2, roughly one ATR14 of 89 points from entry). Target 4229.6 (R1), with a secondary objective at 4296.9 (R2). Horizon 1–5 days. Size: half of normal tactical risk, given the 92.51 crowding percentile; add the second half only on a settle above 4191.7 (P). Conviction 7/10.

Strategy 2 — Defined-risk long via options (secondary). With GVZ at 23.23 (15th percentile, 1Y) and IV/RV at 1.37, buy a 1–2 week call spread struck around 4200/4300 rather than outright futures. Maximum loss is limited to premium; target the same R1/R2 zone. Horizon 1–2 weeks, sized at no more than the premium risk budget. This structure avoids the contango roll cost of −4.38% annualized on the futures leg.

Risk management: the single hard rule is the 4086.5 settle-based stop. Do not average down below S1 4124.4. Reduce size into the FOMC minutes (BJT 10-08 02:00 | ET 10-07 14:00) if the position is profitable, and re-establish after the print. The 52-week low at 3823.7 is the disaster reference if the stop is gapped through.

9. This Week's Data Calendar

BJT 10-05 22:00 | ET 10-05 10:00 — ISM Services PMI (SEP), forecast 54, previous 55.4; surprise if outside 54 ± 1.4. Impacts GC, SI, DXY. BJT 10-08 02:00 | ET 10-07 14:00 — FOMC Meeting Minutes; high impact on GC, SI, DXY. Also BJT 10-07 22:30 | ET 10-07 10:30 — EIA crude and gasoline stocks (CL, BZ), a second-order input for the oil/gold ratio.

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 2026-10-05

Gold (GC=F) Deep Dive | 2026-10-05 — Selloff Exhausted at 20D Floor, Tactical Long | 2026-10-05

Gold (GC=F) Deep Dive | 2026-10-05 — Selloff Exhausted at 20D Floor, Tactical Long | 2026-10-05

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