1. Bottom Line & Directional Bias
Call: Bearish gold (GC=F, COMEX December GCZ26) over a 1–2 week horizon. Invalidation: a daily settle above 4383.07 (R2).
Three reasons drive the call. First, price structure has deteriorated: the 2026-09-25 settle of 4321.2 sits at the 11.6% position of the 20-day 4273.3–4688 channel, with the 5-day change at -2.34% and the 20-day change at -7.35%. Gold is trading on pivot P 4320.67, and the burden of proof now sits with the bulls to reclaim 4352.13 (R1) before 4383.07 (R2). Second, positioning is crowded the wrong way: managed-money net length of 120,318 contracts as of 2026-09-22 carries a 3-year crowding percentile of 92.46, and net length has fallen for four consecutive weeks (134,972 → 133,116 → 127,389 → 120,318) while price has fallen — that is liquidation, not accumulation. Third, the macro backdrop transmits negatively: ^TNX at 5.18% and DXY at 100.97 raise the carry cost of a zero-yield asset, and with GVZ at 22.44 (13th percentile) versus RV20 of 19.9%, downside optionality is cheap but not yet being bought.
The main risk to the call is seasonality: the next 20 sessions have averaged +1.59% (median +1.6%) over the last 15 years, up in 12 of 15. That argues for tactical rather than structural bearishness. A settle above 4383.07 (R2) — roughly 1.4% above the last settle — invalidates the view and would shift the bias to neutral-to-long.
2. Price Action & Technical Analysis
The last completed settlement is 4321.2 (2026-09-25), +0.54% on the day but -2.34% over five sessions and -7.35% over twenty. The 20-day channel runs 4273.3–4688, placing the settle at the 11.6% position — near the floor, not mid-range. The 52-week range is 3724.7–5586.2, so gold remains in the upper half of its annual range but has surrendered the entire September advance.
The last five settled bars tell the story of a failed rebound: 09-21 closed 4383.9, 09-22 4376.4, 09-23 4318.4, 09-24 4298.0, and 09-25 4321.2. The 09-25 bounce came from a 4289.2 low and stalled at a 4351.6 high — below the 09-21 close. Lower highs and lower lows define the sequence. ATR14 is 94.18, or 2.18% of price as a full daily range; RV20 is 19.9%. The 09-25 session's 62.4-point range was well inside one ATR, so the bounce is not a volatility event.
Pivots from the settle: P 4320.67, R1 4352.13, R2 4383.07, S1 4289.73, S2 4258.27. The settle is effectively on P. A sustained trade below S1 4289.73 opens S2 4258.27 and then the 20-day floor at 4273.3 — note that S2 sits below the channel floor, so a break of S1 would put the 20-day low in play quickly. Conversely, reclaiming R1 4352.13 is the first sign of stabilization, and R2 4383.07 is the invalidation line.
On the weekly frame, the last completed bar (2026-09-14 to 2026-09-18) opened 4375, ranged 4273.3–4439.8, and closed 4424.9, +0.36% w/w. The current week (from 2026-09-21, five sessions) is unfinished and last printed 4321.2, -2.34%; no weekly-close conclusion can be drawn from it. The relevant weekly observation is that the completed bar's low of 4273.3 coincides with the 20-day channel floor — that is the structural line the market is now testing.
In early Asian trade on the report date, gold is quoted around the 4321.2 area (Asia), essentially unchanged from the prior settle. The immediate bias is bearish while below 4352.13 (R1); the tactical target zone is 4258–4273 (S2 / 20-day floor).
3. Supply-Demand Balance & Fundamental Drivers
COMEX registered gold stood at 15.16 Moz (471,529 kg) as of 2026-09-25, unchanged d/d from 2026-09-24. Flat registered stocks mean no physical tightness signal from the US delivery system; the metal is available, and the futures curve reflects that. SHFE warrants rose to 116,031 kg as of 2026-09-24, +1,200 kg (+1.05%) d/d from 114,831 kg on 2026-09-23. The Chinese exchange build is small in absolute terms but directionally relevant: Asian physical demand is not absorbing metal fast enough to tighten the visible float.
The term structure is in contango: M1–M2 at -17.4 (-0.4%), roll yield -4.85%, slope 18.53. Contango here is a carry phenomenon — with the 10-year Treasury at 5.18%, the cost of financing gold exceeds its lease rate, so the curve prices that carry. It is not a directional signal and does not cap price; it is a roll cost for longs. That said, a -4.85% annualized roll yield is a meaningful drag for anyone holding long futures exposure, and it reinforces the case for tactical shorts over buy-and-hold longs at this moment.
The macro transmission channel is straightforward. ^TNX at 5.18% (+0.43% on the day, 2026-09-25) and DXY at 100.97 (-0.32%) together define a high-real-rate, firm-dollar regime. Gold pays no coupon, so a 5.18% nominal 10-year — even with inflation running near the 3.7% PCE forecast — leaves a positive real yield that competes with gold as a store of value. The dollar's modest pullback on 09-25 did not prevent gold from closing only marginally higher, which suggests the metal is trading off rates and positioning rather than FX.
On the demand side, the absence of a fresh physical catalyst matters. COMEX registered stocks flat, SHFE warrants building modestly, and no supply disruption in the feed means the fundamental balance is neutral-to-soft. The bullish case rests almost entirely on macro hedging demand and seasonality, not on physical scarcity. Until registered stocks draw down or SHFE warrants fall, the physical market provides no floor beyond the technical levels.
4. Positioning & Fund Flows
CFTC managed-money positioning as of 2026-09-22 shows open interest of 406,456 contracts, longs 131,711, shorts 11,393, and net length 120,318 — a weekly change of -7,071 contracts. The trend is unambiguous: net length has declined for four consecutive reports, from 134,972 (09-08) to 133,116 (09-15) to 127,389 (09-22) to 120,318. Over the same window, price fell from the 4383.9 close on 09-21 to 4321.2 on 09-25. Price down and net length down is liquidation, not divergence — longs are exiting, and there is no evidence of fresh short conviction building (shorts at 11,393 are low in absolute terms).
The crowding metric is the key risk: netPct of 30.86% of open interest carries a 3-year crowding percentile of 92.46. That is a high percentile on the multi-year window, so the long base remains crowded even after four weeks of reduction. The CTA trend-following proxy sits at +62, meaning trend followers are still net long — a vulnerable configuration if the 20-day floor at 4273.3 gives way, because systematic selling would then accelerate. Hedge pressure is 15.07%, modest.
Note the as-of date: the latest CFTC report is 2026-09-22, 15 days old relative to the report date, and has not been updated. It should be treated as the most recent confirmed positioning snapshot, not as current-week data.
On the volatility side, GVZ at 22.44 (13th percentile, 1-day -0.14 points) versus RV20 of 19.9% gives an IV−RV spread of +2.5 vol points (IV/RV 1.13). Options are pricing slightly more than realized movement, but at the 13th percentile of the past year, implied vol is cheap in absolute terms. For a bearish view, that means downside puts are inexpensive relative to history — but it also means the market is not braced for a large move, so a break of support could produce an outsized realized move as vol reprices.
5. Cross-Asset Relative Value
The copper/gold ratio stands at 1.57, at the 95.24% percentile over one year and 52.06% over three years. A high and elevated copper/gold ratio is a pro-growth signal — industrial demand is strong relative to the defensive metal — and it argues against a gold-led safe-haven bid. This is a headwind for gold on a relative-value basis.
The gold/silver ratio is 66.68, at the 54.37% percentile over one year and 18.19% over three years. The three-year percentile is low, meaning silver has been structurally strong versus gold over that window; the ratio is mid-range on a one-year view. A low long-run percentile is a description of the current state, not a forecast of reversal — it does not by itself predict gold outperformance. For gold specifically, the message is neutral: silver is not signaling a precious-metals-wide breakdown, but it is also not leading a rally.
The oil/gold ratio is 0.0214, at the 85.71% one-year percentile and 44.22% three-year. Oil is relatively expensive versus gold on a one-year view, consistent with the elevated WTI implied vol (OVX 55.09, 58th percentile) and the energy-market tightness implied by that vol level. For gold, a high oil/gold ratio is a mild inflation-hedge argument, but with the 10-year at 5.18%, the rates channel dominates.
Cross-asset summary: pro-growth copper/gold, neutral gold/silver, and a firm-dollar/high-rate backdrop. None of these ratios provides a bullish gold catalyst at current levels; the relative-value configuration is neutral-to-negative for the metal.
6. Historical & Seasonal Patterns
The seasonality block covers the same calendar start, next 20 sessions, over the last 15 years: mean +1.59%, median +1.6%, up in 12 of 15 years. The best year in the sample was 2023 at +6.01%; the worst was 2016 at -4.61%. This is a genuinely bullish seasonal window and is the strongest argument against the bearish call.
The caveat is sample size: 15 observations is small, and the distribution is wide — the worst outcome (-4.61%) is nearly three times the magnitude of the mean. A 12-of-15 hit rate is meaningful but not overwhelming, and seasonality is a context input, not a timing signal. The seasonal tailwind argues for tactical rather than structural bearishness, and it is the reason the strategy below uses a defined target rather than an open-ended short.
Historically, the combination of a crowded long base (92nd percentile) and price below the 20-day midpoint has resolved lower more often than not in the near term, but the seasonal window can produce sharp countertrend rallies. The practical implication: respect the seasonal bid, keep the stop tight relative to ATR, and take profit at the 20-day floor rather than pressing for a larger move.
7. Scenario Analysis (Base / Bull / Bear)
Base case — 50%: grind lower toward 4258–4273. Trigger: gold remains below R1 4352.13 and fails to reclaim pivot P 4320.67 on a closing basis. Path: continued managed-money liquidation (net length already down four weeks) and a test of S1 4289.73, then S2 4258.27 and the 20-day floor at 4273.3. Target zone 4258–4273. Action: hold tactical shorts, take partial profit at S1, exit the remainder at the 20-day floor. This scenario agrees with the section 1 call.
Bull case — 30%: seasonal bid reclaims 4383. Trigger: a daily settle above R2 4383.07, aided by the +1.59% mean seasonal return and a softer dollar or lower 10-year yield. Path: gold reclaims R1 4352.13, then R2 4383.07, and targets the 20-day midpoint near 4480. Action: stand aside on shorts; the invalidation level has been breached and the bias shifts to neutral-to-long. This scenario is the primary risk to the call.
Bear case — 20%: liquidation cascade through 4273. Trigger: a daily settle below S1 4289.73 with rising volume, which would put the 20-day floor at 4273.3 under immediate pressure. Path: CTA trend followers (proxy +62) flip from long to short, accelerating the move toward the 52-week range's lower half. Target 4200–4210. Action: add to shorts on the S1 break, trail stops to breakeven, and target the 4200 area. This scenario is an extension of the base case, not a separate directional view.
Probability-weighted, the distribution favors the downside: 70% of outcomes (base + bear) resolve lower or sideways-to-lower, versus 30% for the bull path. The asymmetry is driven by positioning and the rates backdrop, with seasonality the main offset.
8. Trading Strategies & Risk Management
Strategy 1 — Tactical short (primary). Direction: short GCZ26 (COMEX December gold). Entry: 4321–4330, on a failure to reclaim pivot P 4320.67. Stop: 4390, beyond R2 4383.07 and roughly 0.7 ATR above entry. Target: 4260, at S2 4258.27 and the 20-day floor at 4273.3. Horizon: 1–2 weeks. Size: half of normal risk budget, given the seasonal headwind and the 92nd-percentile crowding that can produce sharp squeezes. Conviction: 6/10.
Strategy 2 — Break-and-retest short (secondary). Direction: short on a daily settle below S1 4289.73. Entry: 4285–4290 on the retest. Stop: 4340, above R1 4352.13. Target: 4210. Horizon: 1–2 weeks. Size: quarter of normal risk budget, as this is a momentum-continuation trade that depends on CTA selling. Conviction: 5/10.
Risk management: both strategies are in the direction of the section 1 call. The invalidation for the entire view is a daily settle above 4383.07 (R2); if that occurs, both positions should be closed regardless of stop placement. Position sizing should account for ATR14 of 94.18 — a full daily range of 2.18% — so stops inside 4300–4340 would sit within normal noise. The seasonal window (mean +1.59%) argues for smaller-than-usual size and disciplined profit-taking at the 20-day floor rather than open-ended targets.
9. This Week's Data Calendar
Key events (BJT | ET): 09-29 22:00 | 09-29 10:00 USD JOLTS Job Openings (F 7.23M, P 7.27M). 09-30 20:15 | 09-30 08:15 USD ADP Non-Farm Employment (F 73K, P 38K). 09-30 20:30 | 09-30 08:30 USD Core PCE m/m (F 0.3%, P 0.2%), Final GDP q/q (F 1.5%), Personal Spending (F 0.8%). 10-01 22:00 | 10-01 10:00 USD ISM Manufacturing PMI (F 54.8, P 54.6). China PMIs 09-30 09:30 BJT | 09-29 21:30 ET (Manufacturing F 50.1, Non-Manufacturing F 49.2).
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.