1. Price Action & Technical Analysis
Gold finished the latest completed session at 4321.2 (Fri Sep 25 close), a gain of 0.54% on the day but a loss of 2.36 over five sessions and 7.36 over twenty. The five-day sequence tells the story of a market that tried to stabilize and failed to reclaim anything meaningful: 4383.9 (Mon Sep 21 close, -0.93%), 4376.4 (Tue Sep 22 close, -0.17%), 4318.4 (Wed Sep 23 close, -1.33%), 4298.0 (Thu Sep 24 close, -0.47%), and 4321.2 (Fri Sep 25 close, +0.54%). The bounce on Friday was real but shallow, and it left the market below every close from the prior week.
The 20-day range is defined by the Sep 21 close at 4383.9 on the upper end and the Sep 24 close at 4298.0 on the lower end, a span of roughly 86 dollars, or less than one ATR14. That compression matters. ATR14 stands at 94.18, down from 108.73 on Sep 21, so realized range is contracting even as the market drifts lower. The latest close at 4321.2 sits in the upper-middle of that 20-day band, roughly 27% above the lower extreme and about 73% below the upper extreme. In plain terms, gold is not oversold on a range basis; it is mid-range with a downward slope.
Pivot structure is the cleanest guide here. For the latest session, the pivot P is 4320.4, with R1 at 4351.7 and S1 at 4289.3. The close at 4321.2 is effectively on the pivot, which is why the tape feels indecisive. The prior session's pivots were P 4303.3, R1 4321.7, S1 4268.7, and the session before that P 4344.9, R1 4376.5, S1 4282.3. The pivot has migrated lower from 4344.9 to 4303.3 to 4320.4, a choppy but net-lower path. R1 at 4351.7 is the first level where sellers have shown willingness to engage, and it lines up with the Sep 22 close at 4376.4 as the next shelf above.
Volume and open interest add a caution flag. Open interest on the latest session was 317,452 contracts, up from 315,425 and 313,214 in the two prior sessions. Rising OI into a flat-to-lower price is consistent with new short interest being added, not just long liquidation. Session volume was 121,016, below the 140,964 and 129,178 prints earlier in the week, so the Friday bounce came on lighter participation. The change-in-position metric printed 11.4% on the latest session versus 3.2% and 8.9% before, indicating more aggressive position adjustment on the bounce.
Risk metrics frame the downside. The 52-week drawdown is 25.06%, the 20-day drawdown is 5.33%, 20-day volatility is 19.38%, and the 30-day Sharpe is -0.29. A negative Sharpe with 19% vol means the recent risk-adjusted return has been poor, which is exactly the environment where crowded longs get trimmed. The 95% VaR at -2.88% is a useful sizing anchor: a two-standard-deviation bad day is roughly 124 dollars from the current close.
Net: the trend is lower, the range is compressed, the pivot is the battleground, and R1 at 4351.7 is the level where the desk wants to be a seller. A close back above 4376.4 would break the sequence of lower highs and force a rethink.
2. Fundamental Drivers
The rates picture is the primary headwind. The US ten-year yield (^TNX) printed 5.18% on Sep 25, up 0.43% on the day. A 5.18% nominal ten-year is a serious opportunity cost for a zero-coupon asset, and it keeps real-yield math unfavorable for gold unless inflation expectations are rising faster than nominal yields. The dollar index (DX-Y.NYB) at 100.97, down 0.32% on the day, is a modest offset, but a sub-101 DXY is still firm in absolute terms and has not broken down enough to give gold a durable currency tailwind. The combination of high nominal yields and a firm dollar is the classic bearish cocktail for bullion, and it explains why the five-day change is negative despite a positive Friday.
The physical side offers no squeeze. COMEX registered stocks stood at 471,529.06 kg on Sep 24, down slightly from 472,462.17 kg on Sep 23. SHFE warrants rose to 116,031 kg on Sep 24 from 114,831 kg on Sep 23, a 1,200 kg increase. Rising Shanghai warrants with flat-to-lower COMEX registered inventory is not a tightness signal; it is a market where metal is available and moving to where it is needed. The term structure confirms this: the market is in contango with M1-M2 at -17.4 (-0.4%) and a roll yield of -4.85%, with slope at 18.53. Contango means carry costs are real and holders of long futures pay to roll, which is a structural drag on passive long positioning.
Central-bank flows remain the supportive leg of the fundamental story. The World Gold Council published a deep-dive on Sep 26 examining why central banks are shifting gold reserves, which keeps official-sector demand in the headlines. Official-sector buying is price-insensitive and slow-moving, so it provides a floor rather than a catalyst. It does not offset a 5.18% ten-year yield in the short run, but it does argue against a deep structural breakdown.
Geopolitics is the wildcard, and the oil complex is where it is expressing itself. Headlines over the weekend describe the Trump administration rejecting an Iranian seven-day proposal tied to a return to the memorandum of understanding, with talks stalling and geopolitical risk premium returning to oil. A separate report notes a US-China joint statement after a Trump-Xi meeting, with Beijing unusually aligning on Hormuz Strait language, suggesting the Middle East supply-channel risk premium may be repriced. Hormuz flows are reported at 13-14 million barrels per day, comparable to the July peak, but Iranian flows are at zero and all cargoes are fresh loadings, which is a genuine supply tightening. Higher oil on a supply shock is not automatically strong demand, and for gold the channel is indirect: energy-driven inflation can support the inflation-hedge bid, but it also keeps central banks hawkish, which supports real yields. On balance, the geopolitical tape is a two-sided input, not a clean bullish driver.
3. Positioning & Fund Flows
This is the most important section for the current view. Managed-money positioning from the COT report dated Sep 22 shows open interest of 412,800 contracts, longs at 135,699, shorts at 8,310, and net at 127,389, a weekly change of -5,727. The prior three weeks show net at 133,116 (Sep 15, change -1,856), 134,972 (Sep 8, change -1,799), and 136,771 (Sep 1, change -7,976). Net length has fallen for four consecutive weeks, from 136,771 to 127,389, a cumulative reduction of 9,382 contracts. That is a persistent, orderly unwind, not a panic, and it is the signature of a market where the marginal long is losing conviction.
Crowding remains extreme. Net as a share of open interest was 30.86% on Sep 22, with a crowding score of 92.46. The prior weeks printed 32.48% (crowd 92.58), 32.82% (crowd 92.2), and 32.94% (crowd 92.03). The net percentage is drifting lower while the crowding score stays above 92, which is the classic setup where positioning is still historically stretched even as it unwinds. CTA positioning is reported at 62 across all four weeks, unchanged, and hedge positioning sits at 15.07%, down from 15.94% four weeks ago. The static CTA number is important: it means trend-following exposure has not yet capitulated, so there is a latent supply of selling if the trend confirms lower.
The tension between crowding and percentile deserves a direct comment. A crowding score above 92 means the long side is crowded. A falling net percentage means it is becoming less crowded. These are not contradictory; they describe a market in the early-to-middle stage of a de-risking cycle. The risk is that the unwind is not finished. If CTA positioning at 62 begins to roll, the next leg of net-length reduction could be larger than the 5,727 seen last week.
Options and volatility complete the picture. The CBOE gold volatility index (^GVZ) printed 22.44, down 0.62%, while the VIX printed 14.87, down 5.11%. Gold-specific implied volatility at 22.44 against equity vol at 14.87 means the options market is pricing more uncertainty in gold than in equities. That is consistent with a market where participants are paying for downside protection or upside convexity around a crowded, unstable positioning structure. It is not consistent with complacency.
4. Cross-Asset Relative Value
The copper-gold ratio (HG_GC_RATIO) sits at 0.0016, at the 91st percentile of one year and the 47th percentile of three years. A high one-year percentile means copper has outperformed gold over the past twelve months, which is a pro-cyclical signal. The three-year percentile near the middle says this is not an extreme on a longer horizon. For gold, a strong copper-gold ratio is a relative-value headwind: industrial demand is winning the growth trade, and gold is the laggard.
The gold-silver ratio (GC_SI_RATIO) is 66.68, at the 54th percentile of one year and the 18th percentile of three years. The three-year percentile is the important one. At the 18th percentile, silver has substantially outperformed gold over three years, and mean reversion favors gold. This is a slow, structural tailwind rather than a tradeable trigger, but it does argue that gold is not expensive relative to its monetary cousin on a multi-year view.
The oil-gold ratio (CL_GC_RATIO) is 0.0214, at the 86th percentile of one year and the 44th percentile of three years. Oil has outperformed gold over the past year, consistent with the geopolitical supply premium in crude. The three-year percentile near the middle says the relationship is not stretched on a longer view. For gold, a high oil-gold ratio is a mixed signal: it reflects energy-driven inflation that can support the inflation hedge, but it also reflects a supply shock that keeps central banks cautious.
Putting the three ratios together: copper and oil have both outperformed gold over one year, while silver has outperformed over three years. Gold is the relative laggard on the cyclical cross-assets and the relative laggard against silver on the multi-year view. That is a market that has been left behind by the reflation trade and is now dealing with a positioning unwind. The relative-value case for gold is a mean-reversion case, and mean reversion is patient. It does not help a short-term short, but it does cap how bearish one should be on a medium-term horizon.
5. Sentiment & News Monitor
The sentiment read is cautious-to-negative. The only gold-specific headline in the last 48 hours is the World Gold Council deep-dive on central-bank reserve shifts, which is structurally supportive but not a trading catalyst. The cross-asset news flow is dominated by oil and geopolitics: the Trump administration rejecting an Iranian proposal, stalled US-Iran talks, a US-China joint statement on Hormuz, and reports of Iranian flows at zero with Hormuz volumes at 13-14 million barrels per day. Copper headlines include a Zimbabwe lithium mining variable and ArcelorMittal halting Kryvyi Rih steel output after Russian strikes, both of which are supply-side industrial stories.
The bias of the 48-hour headline flow is mixed for gold. Geopolitical risk premium is rising in oil, which historically provides a modest safe-haven bid to bullion, but the same headlines keep inflation risk alive and therefore keep central banks hawkish. There is no gold-specific negative headline, but there is also no gold-specific positive catalyst. In a market with 92.46 crowding and a negative 30-day Sharpe, the absence of a catalyst is itself a reason for longs to trim.
6. Historical & Seasonal Patterns
Placing today's price in context: the latest close of 4321.2 is 25.06% below the 52-week high, which is the 52-week drawdown figure. That is a meaningful correction from the cycle peak, but it is not a crash. The 20-day drawdown is 5.33%, so the recent damage is contained relative to the annual drawdown. The 20-day volatility of 19.38% is elevated for gold, which typically runs in the mid-teens, and it explains why the ATR14 at 94.18 is large relative to the 86-dollar 20-day range.
The spread Z-scores embedded in the cross-asset data tell a consistent story. The copper-gold ratio at the 91st one-year percentile is a high Z-score for copper versus gold, meaning gold has underperformed. The oil-gold ratio at the 86th one-year percentile is similarly a high Z-score for oil versus gold. The gold-silver ratio at the 18th three-year percentile is a low Z-score for gold versus silver, meaning gold has underperformed silver over three years. All three relative-value signals point the same way: gold is the laggard. Mean reversion favors gold on a medium-term horizon, but momentum favors the laggard staying a laggard until positioning clears.
The 30-day Sharpe of -0.29 is the historical-pattern summary. A negative Sharpe means the recent return stream has not compensated for risk, and historically that regime persists until either price stabilizes or positioning resets. With crowding at 92.46 and net length still falling, the reset is in progress but incomplete.
7. Bull/Bear Scenario Analysis
Bull case:
- Central-bank demand remains price-insensitive and structurally supportive, with the World Gold Council keeping official-sector buying in focus.
- The gold-silver ratio at the 18th three-year percentile means gold is cheap relative to silver, and mean reversion could pull the ratio higher, which implies gold outperformance.
- A dollar index at 100.97 with a -0.32% daily move shows the dollar is not accelerating; a break below 100 would remove a key headwind.
- Geopolitical risk premium in oil, with Hormuz flows at 13-14 million barrels per day and Iranian flows at zero, could spill into a broader safe-haven bid if the situation escalates.
- The 52-week drawdown of 25.06% means a lot of froth has already been removed; value buyers may step in around the 4298-4300 shelf.
Bear case:
- Ten-year yields at 5.18% and rising keep the real-yield carry argument firmly against gold.
- Managed-money net length has fallen for four consecutive weeks, from 136,771 to 127,389, and the unwind is not finished.
- Crowding at 92.46 with CTA positioning static at 62 means trend-following supply is still latent.
- COMEX registered stocks at 471,529 kg and rising SHFE warrants at 116,031 kg show no physical tightness.
- Contango with M1-M2 at -17.4 and a roll yield of -4.85% penalizes passive long positioning.
- The copper-gold and oil-gold ratios at the 91st and 86th one-year percentiles confirm gold is the cyclical laggard.
- A negative 30-day Sharpe of -0.29 and 20-day volatility of 19.38% mean risk-adjusted returns are poor.
Near-term balance: the bear case has more immediate catalysts. The positioning unwind, the yield backdrop, and the contango structure are all live. The bull case is structural and slow. Near term, the path of least resistance is lower, with 4289.3 (S1) as the first support and 4210 as the medium-term target. Medium term, the central-bank bid and the silver mean-reversion signal argue against a sustained breakdown below 4200.
8. Trading Strategies & Risk Management
Primary plan: SHORT gold on a retest of 4351.7 (R1 from the latest session). Entry 4351.7, stop 4446, target 4210. The stop is 94.3 dollars from entry, which is just above one ATR14 of 94.18, and the target is 141.7 dollars from entry, giving a reward-to-risk ratio of approximately 1.5. Horizon is 1-5 days. Conviction is 7 out of 10.
The rationale: R1 at 4351.7 is the first level where sellers have engaged, and it sits below the Sep 22 close at 4376.4, so a retest of R1 does not break the sequence of lower highs. The stop at 4446 is above the Sep 21 close at 4383.9 and above the Sep 22 close at 4376.4, so it is only triggered if the market reclaims the prior week's range, which would invalidate the lower-highs thesis. The target at 4210 is below the Sep 24 close at 4298.0 and below S1 at 4289.3, and it is consistent with a continuation of the four-week positioning unwind.
Invalidation: a daily close above 4376.4 would break the lower-high sequence and put the short thesis on hold. In that scenario, the market would likely be testing the Sep 21 close at 4383.9 and the 20-day range high, and the desk would stand aside rather than fight the reclaim.
Risk management: size the position so that a stop-out at 4446 costs no more than the standard single-trade risk budget. The 95% VaR of -2.88% implies a two-standard-deviation adverse day of roughly 124 dollars, which is larger than the stop distance, so position size should be conservative. Do not add to the short below 4289.3 without a fresh lower-high structure; the first target zone is where partial profit-taking is appropriate.
9. This Week's Data Calendar
| Time | Event | Importance |
|---|
| Sun 2026-09-27 19:50 ET / Mon 2026-09-28 07:50 Beijing | BoJ Monetary Policy Meeting Minutes | MEDIUM |
| Sun 2026-09-27 21:30 ET / Mon 2026-09-28 09:30 Beijing | China Industrial Profits ytd/y | MEDIUM |
| Mon 2026-09-28 09:30 ET / Mon 2026-09-28 21:30 Beijing | ECB President Lagarde Speaks | MEDIUM |
| Tue 2026-09-29 00:30 ET / Tue 2026-09-29 12:30 Beijing | RBA Rate Statement and Cash Rate (forecast 4.6%, previous 4.35%) | HIGH |
| Tue 2026-09-29 21:30 ET / Wed 2026-09-30 09:30 Beijing | China Manufacturing PMI (forecast 50.1, previous 49.8) | HIGH |
| Tue 2026-09-29 21:30 ET / Wed 2026-09-30 09:30 Beijing | China Non-Manufacturing PMI (forecast 49.3, previous 49.0) | HIGH |
| Tue 2026-09-29 21:30 ET / Wed 2026-09-30 09:30 Beijing | Australia CPI m/m (forecast 0.5%, previous 1.0%) and CPI y/y (forecast 4.1%, previous 3.5%) | HIGH |
| Tue 2026-09-29 21:45 ET / Wed 2026-09-30 09:45 Beijing | China RatingDog Services PMI (forecast 51.2) and Manufacturing PMI (forecast 51.7) | HIGH |
| Wed 2026-09-30 08:30 ET / Wed 2026-09-30 20:30 Beijing | US Final GDP q/q (forecast 1.5%), Personal Spending MoM (forecast 0.8%), Personal Income MoM (forecast 0.4%), Core PCE Price Index m/m (forecast 0.3%) | HIGH |
| Wed 2026-09-30 19:50 ET / Thu 2026-10-01 07:50 Beijing | Japan Tankan Large Manufacturers Index Q3 (forecast 25, previous 22) | HIGH |
| Fri 2026-10-02 08:30 ET / Fri 2026-10-02 20:30 Beijing | US Average Hourly Earnings m/m (forecast 0.3%), Unemployment Rate (forecast 4.1%), Non-Farm Employment Change (forecast 98K, previous 162K) | HIGH |
The week's risk events are back-loaded. The Core PCE print on Wed Sep 30 and the Non-Farm Payrolls report on Fri Oct 2 are the two events most likely to move the rates complex and therefore gold. A soft Core PCE or a weak NFP would lower the ten-year yield and challenge the short thesis; a firm print would reinforce it. The China PMI cluster on Tue Sep 29 is relevant for the copper-gold and silver complex and could indirectly affect gold through the reflation channel.
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.