Gold (GC=F) Deep Dive | 2026-10-04 — Bearish, 4143.1 20-day floor in focus | 2026-10-04

Call: Bearish gold. The prior session settle of 4162.3 (2026-10-02) sits just 19 points above the 20-day low of 4143.1, with the 5D change at -3.68% and the 20D change at -8.32% (settle), so the path of least resistance remains lower. Three drivers underpin the call: (1) positioning is still crowded long — managed-money net is 120,318 contracts, netPct 29.6%, a 92.51 3-year percentile, and has been cut for four straight weeks while price fell, i.e. de-risking is not finished; (2) the macro impulse is hostile — the US 10-year yield at 5.28% and DXY at 101.93 keep the carry cost of a non-yielding asset elevated, and the curve is in contango (M1-M2 -15.2, roll yield -4.38%); (3) the tape is weak — price is in the bottom 4.9% of its 20-day channel, below pivot P 4191.7, with ATR14 at 89 (2.14% of price) and RV20 at 17%. Invalidation: a settle back above P 4191.7, and more decisively above R1 4229.6, would neutralize the bear case. The base case (55%) targets 4086.5 (S2) over the next 1-2 weeks.

1. Bottom Line & Directional Bias

Call: Bearish gold. The prior session settle was 4162.3 (2026-10-02), only 19.2 points above the 20-day low of 4143.1, and the trend structure is unambiguously down: -3.68% over 5D and -8.32% over 20D, both from the settle-based snapshot. The invalidation is a settle back above pivot P 4191.7; a settle above R1 4229.6 would kill the call outright.

Three reasons. First, positioning. Managed-money net length is 120,318 contracts with netPct at 29.6% and a 3-year crowding percentile of 92.51 — still a crowded long, and the net has been reduced for four consecutive weeks (134,972 → 133,116 → 127,389 → 120,318). Crowded longs in a falling market are fuel for further liquidation, not a contrarian buy signal. Second, macro transmission. The US 10-year yield at 5.28% and DXY at 101.93 raise the opportunity cost of holding gold; the curve is in contango (M1-M2 -15.2, roll yield -4.38%), which is a carry cost for longs, not a price cap, but it does mean passive length bleeds. Third, the tape. Price is in the bottom 4.9% of the 20-day channel (4143.1–4537.8), below P 4191.7, with ATR14 at 89 (2.14% of price) and RV20 at 17%. The last completed weekly bar (2026-09-21–2026-09-25) closed at 4321.2, -2.34% w/w, and the current week is unfinished — no weekly-close conclusions are available. Bias: SHORT, invalidation on a settle above 4191.7.

2. Price Action & Technical Analysis

The settle of 4162.3 (2026-10-02) is the reference for every level below. The 1D change was -0.95% (settle), the 5D change -3.68% (settle) and the 20D change -8.32% (settle). The 20-day channel runs 4143.1–4537.8, putting price at the 4.9% position — effectively at the floor. The 52-week range is 3823.7–5586.2, so the market is in the lower third of its annual envelope but not at an extreme.

Pivots from the snapshot: P 4191.7, R1 4229.6, R2 4296.9, S1 4124.4, S2 4086.5. Price is below P, which is the first structural confirmation of the bear case. The immediate battle is the 20-day low at 4143.1; a settle below it opens S1 4124.4 and then S2 4086.5. ATR14 is 89 points, or 2.14% of price — that is the full expected daily range, not a one-sided band, so a single session can travel from the 20-day low to S2 without being exceptional. RV20 is 17%, meaning realized movement has been moderate relative to the ATR; the market is trending, not chopping.

On the weekly frame, the last completed bar (2026-09-21–2026-09-25) opened 4413, high 4422.1, low 4278.3, closed 4321.2, -2.34% w/w. That is a completed weekly close below the prior week and below the 4400 handle. The current week (from 2026-09-28, five sessions) is not closed; the last print of 4162.3, -3.68%, is an unfinished-week figure and cannot be described as a weekly close. The early Asian snapshot on the report-date bar is not a settle and is not used for levels here.

View: bearish while below P 4191.7; the 4143.1 floor is the trigger for the next leg. A settle above 4191.7 would shift the tactical bias to neutral and put R1 4229.6 in play.

3. Supply-Demand Balance & Fundamental Drivers

COMEX registered gold stood at 15.09 Moz (469,352 kg) on 2026-10-01, unchanged d/d. The prior two prints were 15.09 Moz (469,352 kg) on 2026-09-30 and 15.14 Moz (470,907 kg) on 2026-09-29 — so registered stocks have drifted marginally lower over the three sessions, a modestly constructive physical signal that is not large enough to offset the macro and positioning drag. SHFE warrants were 116,028 kg on 2026-09-30, down 3 kg d/d, effectively flat. The absence of a visible inventory build means the sell-off is being driven by paper flows and macro, not by physical surplus; that matters for the shape of the decline (orderly rather than panic) but not for its direction.

The term structure is in contango: M1-M2 at -15.2 (-0.37%), roll yield -4.38%, slope 17.87. Contango in gold reflects the cost of carry — elevated US rates — and is a roll cost for long positions, not a ceiling on price. With the 10-year at 5.28%, the carry math is straightforward: a long futures position pays to roll, and the opportunity cost of a zero-coupon asset is high. That is a headwind for passive length and a reason why ETF and managed-money exposure tends to shrink when the trend turns, as it has.

The macro transmission channel is the dollar and rates. DXY at 101.93 (-0.17% on the day) is not surging, but it is at a level that keeps gold's dollar-denominated appeal muted; the 10-year at 5.28% (+0.76%) is the more important variable, because it sets the real-cost anchor. The FOMC Minutes on 2026-10-07 (ET 14:00 / BJT 10-08 02:00) is the key event risk for this channel. ISM Services PMI on 2026-10-05 (ET 10:00 / BJT 22:00) with a forecast of 54 versus prior 55.4 and a surprise threshold of ±1.4 is the first macro test of the week; a strong print would reinforce the higher-for-longer rate narrative and pressure gold, while a weak print would be the main upside risk to this bearish call.

View: the physical balance is neutral-to-mildly-supportive, but the carry and macro channel dominates and points lower. Watch the 10-year and the FOMC Minutes as the transmission events.

4. Positioning & Fund Flows

The CFTC data as of 2026-09-29 shows open interest of 406,456, longs 131,711, shorts 11,393 and a managed-money net of 120,318, down 7,071 w/w. The prior three weeks were net 127,389 (Δ -5,727), 133,116 (Δ -1,856) and 134,972 (Δ -1,799). The pattern is unambiguous: four consecutive weeks of net-length reduction, accelerating in the last two, while price fell from the 4400s to 4162.3. This is de-risking, not capitulation — the net is still 120,318 contracts and netPct is 29.6%.

The crowding percentile is 92.51 on a 3-year window, down only marginally from 92.46 and 92.58 in the prior two weeks. By the desk convention, a net-length percentile this high on the multi-year window qualifies the long as crowded. The CTA proxy is 62, unchanged across all four weeks, which suggests trend-following exposure has not yet flipped short in size; if the 4143.1 floor breaks, that cohort is the most likely source of the next wave of selling. Hedge pressure is 14.19%, down from 15.6% four weeks ago, consistent with producers trimming protection into weakness rather than adding it.

On volatility, GVZ is 23.23 (1Y percentile 15%), down 0.09 points on the day, while GC=F RV20 is 17%. That puts IV minus RV at +6.2 vol points and IV/RV at 1.37 — options are paying up relative to realized movement, which is typical when the market is pricing event risk (FOMC Minutes, ISM) into a fragile technical setup. For a bearish directional view, that argues for expressing the trade in futures or defined-risk structures rather than outright long puts, where the premium is rich.

View: positioning remains the single largest downside risk to price because the long base is still crowded and the reduction is incomplete. A break of 4143.1 is the trigger for the next leg of liquidation.

5. Cross-Asset Relative Value

The copper/gold ratio is 1.57 (ratio ×1000) at a 1-year percentile of 93.65% and a 3-year percentile of 51.98%. A high 1-year percentile means copper has been strong relative to gold over the past year — a pro-growth signal that is inconsistent with a safe-haven bid in gold. On the 3-year window the ratio is mid-range, so this is a cyclical tilt rather than a structural extreme.

The gold/silver ratio is 68.9, at a 1-year percentile of 71.43% and a 3-year percentile of 23.81%. A high 1-year percentile means silver has been lagging gold over the past year; the low 3-year percentile means that over three years silver has been structurally strong versus gold. The two readings together say the recent regime favors gold over silver, but the multi-year backdrop does not. For a bearish gold view, silver's relative weakness is a confirmation of the industrial-demand-over-safe-haven tilt, not a reason to prefer silver outright.

The oil/gold ratio is 0.0219, at a 1-year percentile of 90.08% and a 3-year percentile of 46.3%. Oil has been strong versus gold on a 1-year view, again consistent with a pro-cyclical, reflationary backdrop that is hostile to gold. The 3-year percentile is mid-range, so this is not an extreme. Cross-asset, the message is consistent: the market is pricing growth and carry over defensive stores of value, which supports the bearish gold call.

View: relative value confirms the bearish tilt — copper/gold and oil/gold both sit at high 1-year percentiles, and gold/silver at a high 1-year percentile means gold is the stronger of the two precious metals, so the short should be expressed in gold rather than silver.

6. Historical & Seasonal Patterns

The seasonality block covers the same calendar start, next 20 sessions, over the last 15 years: mean +0.87%, median +0.26%, up in 10 of 15 years, best 2023 +8.81%, worst 2012 -6.24%. The sample is small and the block itself flags it as context only.

The honest read is that the seasonal window is mildly positive on average, with a hit rate of roughly two-thirds. That is a headwind to a bearish call, and it is why the base case is not an aggressive downside target. But the median of +0.26% is close to flat, and the distribution is wide — the worst year was -6.24%, which is larger in magnitude than the best year is positive relative to the median. In a year where the 20D change is -8.32% and positioning is crowded long, the seasonal tailwind is not sufficient to override the trend and flow signals. Seasonality is a reason to size the short modestly and to take profits at S2 rather than to press for a deeper target.

View: seasonality is a mild counterweight, not a reason to abandon the bearish call. It argues for disciplined targets and no pyramiding below S2.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 55% — bearish continuation. Trigger: a settle below the 20-day low of 4143.1. Target: S1 4124.4 first, then S2 4086.5. Action: hold or add to short exposure on the break, with stops above P 4191.7. The base case agrees with the section 1 call. The logic is that the crowded long base (netPct 29.6%, crowding 92.51) continues to unwind, the 10-year at 5.28% keeps carry negative, and the tape is already at the 4.9% position of the 20-day channel. ATR14 of 89 points means the distance from 4143.1 to 4086.5 is roughly 0.6 ATR — achievable within one to two sessions if the floor gives way.

Bull case — 25% — squeeze back to the pivot. Trigger: a settle back above P 4191.7, most likely on a weak ISM Services print (forecast 54, prior 55.4, surprise threshold ±1.4) or a dovish FOMC Minutes. Target: R1 4229.6, with R2 4296.9 as the stretch. Action: stand aside on shorts, do not chase; a settle above 4229.6 invalidates the bearish call and shifts the bias to neutral. The bull case is not the base case because the trend, the positioning overhang and the carry all still point lower; a bounce would be a countertrend move into resistance.

Bear case — 20% — acceleration lower. Trigger: a settle below S1 4124.4 accompanied by a CTA flip (the proxy is 62 and unchanged for four weeks, so trend-followers have room to sell). Target: S2 4086.5 and, if momentum builds, the 52-week low region at 3823.7 becomes the medium-term reference. Action: trail stops to the prior day's high and let the position run, but recognize that a move of that magnitude would require a macro catalyst beyond the current calendar — the FOMC Minutes is the only high-impact event in the window. The bear case is sized below the base case because the 20-day low has not yet been breached on a settle basis and seasonality is mildly positive.

Probabilities sum to 100%. The base case is the only path consistent with the section 1 call; the bull case is the invalidation path, and the bear case is the tail.

8. Trading Strategies & Risk Management

Strategy 1 — Short GCZ26 on a settle below 4143.1. Entry 4140, stop 4230 (above R1 4229.6, roughly one ATR from entry), target 4086.5 (S2), horizon 1-5 days, size 0.5x normal risk budget given the crowded positioning and the event calendar. Conviction 7/10. If the market settles below 4143.1 but then reclaims 4191.7, exit early rather than waiting for the stop.

Strategy 2 — Short GCZ26 on a rally into P 4191.7. Entry 4190, stop 4230, target 4124.4 (S1), horizon 1-5 days, size 0.5x normal risk budget. Conviction 6/10. This is the higher-quality entry if the market bounces first, because it sells into resistance rather than into the floor. Do not initiate this leg if the market has already settled above 4229.6.

Risk management: the FOMC Minutes on 2026-10-07 (ET 14:00 / BJT 10-08 02:00) is the dominant event risk; reduce size into it or accept gap risk. ISM Services on 2026-10-05 (ET 10:00 / BJT 22:00) is the first test. Both strategies are in the direction of the call; no long exposure is recommended while price is below P 4191.7. If the market settles above 4229.6, both strategies are void and the bias moves to neutral.

9. This Week's Data Calendar

BJT 10-05 22:00 | ET 10-05 10:00 — ISM Services PMI SEP, forecast 54 vs prior 55.4, surprise if outside 54 ± 1.4. Affects GC, SI, DXY.
BJT 10-07 04:30 | ET 10-06 16:30 — API Crude Oil Stock Change OCT/02. Affects CL, BZ.
BJT 10-07 22:30 | ET 10-07 10:30 — EIA Crude Oil and Gasoline Stocks Change OCT/02. Affects CL, BZ.
BJT 10-08 02:00 | ET 10-07 14:00 — FOMC Minutes. Affects GC, SI, DXY. The highest-impact event for gold this week.

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 2026-10-04

Gold (GC=F) Deep Dive | 2026-10-04 — Bearish, 4143.1 20-day floor in focus | 2026-10-04

Gold (GC=F) Deep Dive | 2026-10-04 — Bearish, 4143.1 20-day floor in focus | 2026-10-04

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