一、 Price Action & Technical Analysis
SHFE Gold main contract closed at 927.66 CNY/t on September 24, down 1.33% on the single day, marking the largest single-day decline in the past 5 trading days; it closed at 939.5 CNY/t on September 23, down 0.08%; closed at 935.78 CNY/t on September 22, down 1.05%; closed at 942.9 CNY/t on September 21, down 0.35%; closed at 951.38 CNY/t on September 18, up 1.51%. Over the 5-day horizon, it has retreated from 951.38 CNY/t to 927.66 CNY/t, with a cumulative decline of approximately 2.49%; over the 20-day horizon, the 20-day change of the international counterpart GC=F is -7.36%, with China moving in sync.
In terms of range positioning, the 20-day high is 1003.7 CNY/t, the 20-day low is 923.3 CNY/t, and the latest closing price of 927.66 CNY/t is only 4.36 CNY/t away from the 20-day low, located near the 4th percentile of the lower end of the 20-day channel, belonging to the bottom area of the range. The moving average structure shows a bearish alignment: MA5 is 939.44 CNY/t, MA20 is 949.23 CNY/t, the closing price is below both, and MA5 is below MA20, with short- and medium-term moving averages diverging downward.
In terms of pivot system, the latest pivot P is 930.37 CNY/t, R1 is 933.33 CNY/t, S1 is 924.69 CNY/t, and the closing price of 927.66 CNY/t is below the pivot and above S1, weak intraday but not breaking S1. ATR14 is 14.84 CNY/t, based on the latest closing price, the upper bound of 1x ATR is approximately 942.5 CNY/t, and the lower bound is approximately 912.82 CNY/t, providing clear reference value for intraday trading range.
Key levels sorted out: Upper resistance levels are in sequence: 924.69 CNY/t (after S1 failed to turn into support), pivot 930.37 CNY/t, R1 933.33 CNY/t, MA5 939.44 CNY/t, MA20 949.23 CNY/t; lower support is the 20-day low of 923.3 CNY/t, if it breaks down effectively, there is no near-term reference below, and ATR extrapolation is required. In terms of open interest, OI on September 24 was 222,206 lots, an increase of 24,660 lots from 197,546 lots on September 18; price fell with open interest increasing, showing obvious characteristics of short positions building, forming a continuous increase with OI of 212,832 lots on September 23, 206,703 lots on September 22, and 201,225 lots on September 21. Volume on September 24 was 183,228 lots, significantly enlarged from 122,113 lots on the previous day; decline accompanied by volume amplification indicates that short-term selling pressure has not been fully released.
二、 Fundamental Drivers
On the macro rate front, the U.S. 10-year Treasury yield (^TNX) stood at 5.18% on September 25, up 0.43% on the single day, remaining in a high range. High nominal rates directly push up the opportunity cost of holding gold, which is the core variable suppressing gold prices. If yields remain above 5%, gold valuations will struggle to gain systemic support; if yields subsequently fall below 5%, real rate pressure will ease, and gold prices are expected to gain a repair window.
On the U.S. dollar front, the U.S. Dollar Index (DX-Y.NYB) stood at 100.97 on September 25, down 0.32% on the single day. A modest weakening of the U.S. dollar provides marginal support for gold priced in U.S. dollars, but the magnitude is limited and fails to reverse the decline of the China-side contract. If the U.S. Dollar Index continues to fall below the 100 integer mark, external pressure on gold will ease; if it rises back above 101, gold price pressure will intensify.
On the inflation and central bank gold purchase front, the top headline related to China in the past 48 hours is the World Gold Council’s in-depth interpretation of why central banks are shifting gold reserves (Sina_Futures, 2026-09-26 16:38:09), indicating that central bank reserve management behavior remains part of the medium- and long-term demand narrative, but this narrative has limited marginal driving force for short-term prices.
On the inventory front, SHFE Gold inventory (warrants) was 116,031 KG on September 24, with a weekly change of +1,200 KG; it was 114,831 KG on September 23, with a weekly change of 0. China’s inventory has accumulated slightly continuously, reflecting that the spot side is not tight. The international counterpart COMEX registered inventory was 471,529.1 KG on September 24, and 472,462.2 KG on September 23, a slight decline.
On the term structure front, the international counterpart is in CONTANGO, with M1-M2 at -17.4 (-0.4%), RY at -4.85%, and slope at 18.53; the near-term backwardation structure indicates that spot supply and demand are relatively loose, holding costs are positive, which is not conducive to long positions rolling over for gains.
On the geopolitical and policy front, China’s financial calendar is dense this week, with manufacturing and non-manufacturing PMI and RatingDog series PMI scheduled for release on September 30, and bank holidays from October 1 to 2, which will affect liquidity rhythm. If PMI data exceeds expectations, risk appetite may rise and divert gold’s safe-haven funds; if data falls short of expectations, safe-haven demand may provide support for gold prices.
三、 Positioning & Fund Flows
International counterpart COT data shows that net long positions have declined for four consecutive weeks: net long was 136,771 lots on September 1 (Δ-7,976), 134,972 lots on September 8 (Δ-1,799), 133,116 lots on September 15 (Δ-1,856), and 127,389 lots on September 22 (Δ-5,727). The latest weekly reduction of 5,727 lots is the largest in four weeks, with active long liquidation being the main marginal change. Total open interest (OI) decreased from 415,196 lots on September 1 to 412,800 lots on September 22; long positions (L) decreased from 149,721 lots to 135,699 lots, short positions (S) decreased from 12,950 lots to 8,310 lots; both long and short positions decreased, but long positions declined more, leading to a contraction in net long positions.
In terms of crowding degree, netPct was 30.86% on September 22, crowd was 92.46, CTA was 62, hedge was 15.07%. Compared with netPct of 32.94% and crowd of 92.03 on September 1, the proportion of net long positions has declined slightly, but the crowding degree remains in the extreme interval above 92. There is a contradiction between percentile and crowding degree here: the decline in net long proportion indicates marginal position reduction, but the absolute level remains high, meaning long positions have not been fully unwound; if prices continue to fall, there is a risk of further passive liquidation in a chain; conversely, if prices stabilize, short covering under high crowding degree may amplify rebound elasticity. CTA positions remained at 62 for four consecutive weeks, with trend funds showing no obvious shift; hedge proportion decreased slightly to 15.07%, and hedging pressure eased slightly.
On the options and volatility front, ^GVZ (CBOE Gold Volatility Index) stood at 22.44 on September 25, down 0.62% on the single day; implied volatility declined slightly but remains above the normal range; ^VIX stood at 14.87, down 5.11% on the single day; risk asset volatility declined, forming a divergence with gold volatility, indicating that gold’s own downward volatility stems more from commodity-internal fund behavior rather than systemic risk.
四、 Cross-Asset Relative Value
The gold-silver ratio (GC_SI_RATIO) is currently at 66.68, with a 1-year percentile of 54.37% and a 3-year percentile of 18.12%. The 1-year percentile is neutral, and the 3-year percentile is low, indicating that silver is relatively stronger than gold over a longer cycle, and the gold-silver ratio is in a historically low range. Based on mean reversion logic, a low percentile means a higher probability of the ratio moving upward, i.e., gold has a catch-up rally or anti-fall demand relative to silver; based on momentum logic, silver has stronger momentum during the ratio’s downward trend, so it is not advisable to go long the gold-silver ratio against the trend in the short term.
The copper-gold ratio (HG_GC_RATIO) is currently at 0.0016, with a 1-year percentile of 90.87% and a 3-year percentile of 47.49%. The 1-year percentile is extremely high, indicating that copper has significantly outperformed gold, and the ratio is at a 1-year high. Based on mean reversion logic, a high percentile means a higher probability of the ratio moving downward, i.e., gold has catch-up rally demand relative to copper; however, the 3-year percentile is only 47.49% (neutral), with no extreme in the long cycle, so the repair space needs to be evaluated cautiously.
The oil-gold ratio (CL_GC_RATIO) is currently at 0.0214, with a 1-year percentile of 85.71% and a 3-year percentile of 44.05%. Similar to the copper-gold ratio, the 1-year percentile is high and the 3-year percentile is neutral, indicating that crude oil is relatively stronger than gold over the 1-year horizon.
Overall, gold is relatively weaker than copper and oil over the 1-year horizon, and relatively stronger than silver; cross-asset relative value gives a neutral-slightly bullish signal for gold, but none of the ratios have reached extreme percentiles, which is insufficient to form an independent long driver. If the copper-gold ratio and oil-gold ratio fall from their highs, gold’s relative allocation value will improve; if they continue to rise, gold’s relative weakness will persist.
五、 Sentiment & Public Opinion Monitoring
There is only one top headline related to China in the past 48 hours: the World Gold Council’s in-depth interpretation of why central banks are shifting gold reserves (Sina_Futures, 2026-09-26 16:38:09), which focuses on central bank reserve management, a medium- and long-term structural narrative with limited guidance for short-term price direction, leaning neutral-bullish.
In terms of sentiment dimensions, combined with consecutive price declines, increasing open interest, declining COT net long positions, and GVZ remaining above 22, comprehensive sentiment is bearish. International counterpart risk indicators show: Sharpe30 is -0.286, with a negative 30-day risk-adjusted return; Vol20 is 19.38%, with a high 20-day annualized volatility; DD20d is 5.33%, with a significant 20-day drawdown; DD52w is 25.06%, with a deep drawdown from the 52-week high. VaR95 is -2.88%, with a single-day 95% confidence downside risk approaching 3%.
Public opinion tendency judgment: Currently, there is no clear bullish catalyst; the central bank gold purchase narrative is a slow variable, which is difficult to offset interest rate and crowding pressure in the short term, so sentiment remains bearish.
VI. Historical Statistical Patterns
Using international benchmark data to contextualize current prices: GC=F latest close at 4321.2, 20-day change -7.36%, 5-day change -2.36%, currently in a 20-day downtrend channel. DD52w stands at 25.06%, meaning the current price is approximately one-quarter off its 52-week high, placing it in the lower-middle range of the 1-year interval; DD20d is 5.33%, a moderate 20-day drawdown.
From a percentile perspective: the 3-year percentile of the gold-silver ratio is 18.12%, the lowest among the three cross-asset spreads; the 1-year percentile of the copper-gold ratio is 90.87%, the highest; the 1-year percentile of the oil-gold ratio is 85.71%, the second highest. In terms of spread Z-scores, the COT net long position share has declined from 32.94% to 30.86%, falling for four consecutive weeks, with the Z-score trending downward, indicating that long position crowding is slowly easing from extreme levels but remains elevated in absolute terms.
Historical patterns show that when COT crowding exceeds 92 and net long positions decline consecutively, prices are typically in the mid-to-late stage of a downtrend rather than at the start, requiring time to digest open interest cut pressure. If the net long share continues to fall below 28%, crowding risk will be substantially relieved, increasing the probability of price stabilization; if the net long share stops declining near 30%, a temporary bottom structure may form.
For the onshore China market: the 20-day low of 923.3 CNY/t is a key support level over the past month. Historically, initial tests of the lower end of the range often involve volatility; a break with high volume would confirm the trend continuation, while stabilization with low volume would create a rebound window.
VII. Long-Short Game Deduction
Bullish arguments: 1. The 3-year percentile of the gold-silver ratio at 18.12% is low, implying a mean reversion logic that gold has a need to recover relative to silver; 2. The 1-year percentiles of the copper-gold ratio (90.87%) and oil-gold ratio (85.71%) are both high, meaning gold has room to catch up relative to industrial metals and energy; 3. The U.S. Dollar Index at 100.97 has declined by 0.32%, marginally easing external pricing pressure; 4. SHFE inventory weekly change of +1200 KG, while accumulated, has a limited absolute level of 116031 KG, with no extreme looseness observed in the spot market; 5. GVZ has declined slightly by 0.62% from its high, with no intensification of panic selling; 6. Central bank gold purchase narratives provide medium-to-long-term bottom support.
Bearish arguments: 1. Prices closed below MA5 (939.44 CNY/t) and MA20 (949.23 CNY/t), forming a bearish moving average arrangement; 2. On September 24, prices fell by 1.33% alongside open interest rising to 222206 lots, confirming a short position build; 3. COT net long positions have declined for four consecutive weeks, with a single-week reduction of 5727 lots in the latest week, the largest single-week decline; 4. Crowding at 92.46 remains extreme, with long position liquidation incomplete; 5. U.S. 10-year Treasury yields at 5.18% are elevated, exerting holding cost pressure; 6. The term structure is in CONTANGO, with M1-M2 at -17.4, resulting in negative roll yield; 7. Sharpe30 is -0.286, with negative risk-adjusted returns.
Force balance: Short-term (1-5 days): short positions dominate, with prices near the 20-day low of 923.3 CNY/t, posing greater breakdown risk than rebound momentum; Medium-term (1-3 months): long-short interplay, with cross-asset spreads and central bank gold purchases providing bottom support, but interest rate and crowding pressures require time to digest. Overall, the short-to-medium-term bearish pattern remains unchanged, with rebounds to be treated as short-selling opportunities.
VIII. Trading Strategies & Risk Management
No new trades: No entry points with appropriate stop-loss and take-profit levels have been identified near current prices; wait for prices to return to key levels in Sections VI and VII before reassessment.
IX. This Week's Data Calendar
- Beijing 2026-09-28 Mon 09:30 / New York 2026-09-27 Sun 21:30 [CNY/MEDIUM] Industrial Profits ytd/y
- Beijing 2026-09-30 Wed 09:30 / New York 2026-09-29 Tue 21:30 [CNY/HIGH] Manufacturing PMI Expected 50.1 Prior 49.8
- Beijing 2026-09-30 Wed 09:30 / New York 2026-09-29 Tue 21:30 [CNY/HIGH] Non-Manufacturing PMI Expected 49.3 Prior 49.0
- Beijing 2026-09-30 Wed 09:45 / New York 2026-09-29 Tue 21:45 [CNY/HIGH] RatingDog Manufacturing PMI Expected 51.7 Prior 51.5
- Beijing 2026-09-30 Wed 09:45 / New York 2026-09-29 Tue 21:45 [CNY/HIGH] RatingDog Services PMI Expected 51.2 Prior 51.4
- Beijing 2026-09-30 Wed 21:45 / New York 2026-09-30 Wed 09:45 [USD/MEDIUM] Chicago PMI SEP Expected 51.3 Prior 47.1
- Beijing 2026-10-01 Thu 07:01 / New York 2026-09-30 Wed 19:01 [CNY/LOW] Bank Holiday
- Beijing 2026-10-01 Thu 22:00 / New York 2026-10-01 Thu 10:00 [USD/MEDIUM] ISM Manufacturing PMI Expected 55.0 Prior 54.6
- Beijing 2026-10-02 Fri 07:01 / New York 2026-10-01 Thu 19:01 [CNY/LOW] Bank Holiday
This report is generated automatically from public quantitative and macro data for research and market tracking only. It does not constitute any investment advice.