I. Bottom Line & Directional Bias
Direction: bearish. Three core reasons: First, the price structure has broken down — SHFE Silver settled at 14906 (2026-09-29), with the close at 14848 breaking below the prior 20-day low of 14975, sitting at just the 5.5% percentile within the 20-day channel of 14745–16628, down -8% over 5 days and -8.87% over 20 days, with momentum pointing downward rather than consolidating sideways. Second, the positioning structure confirms bearish dominance — main-contract notional open interest of CNY 61.73bn, daily change +CNY 3.323bn, open interest +14920 lots while the settlement price fell -2.17%, constituting an open interest build on a price decline (shorts initiating), while the top-20 seats' net long of 44042 lots rose by only 758 lots, meaning longs did not step in. Third, physicals and basis weakened in tandem — spot at 14779 versus main contract 14848 gives a basis of -69.0 (-0.47%), weakening by 205.0 from the prior reading of +136.0, flipping from premium to discount; SHFE warehouse warrants at 1460361 kg, up 21070 kg day-on-day (+1.5%), with a cumulative +37045 kg over 5 periods, indicating rising supply-side pressure. Conditions that would invalidate the direction: settlement price closing above 15031.7 (pivot R1) for two consecutive trading days, or SHFE warehouse warrants and COMEX registered stocks simultaneously turning to consecutive draws with the basis returning to a premium.
II. Price Action & Technicals
On a settlement basis, SHFE Silver settled at 14906 on 2026-09-29, -2.17% versus the prior settlement, -8% over 5 days and -8.87% over 20 days, a one-way decline rather than a range. ATR14 is 499.5, equivalent to 3.36% of the settlement price (full daily range), and RV20 is 38.9%, with volatility at elevated levels, meaning any directional position's stop-loss must allow sufficient volatility room. The 20-day channel is 14745–16628, with the current settlement at the 5.5% percentile of the channel, near the lower bound; the close at 14848 has already broken below the prior 20-day low of 14975, which is where this issue's trigger condition originates. Pivot system: P 14888.3, R1 15031.7, S1 14704.7; the settlement of 14906 sits about 18 points above P, less than half an ATR (about 250 points) from P, so P does not constitute a valid trading reference; R1 is about 126 points from the settlement and S1 about 201 points away, both also within half an ATR, and are therefore not used as standalone levels, only as boundaries for the invalidation conditions. The last 5 settlement daily bars show a continuous step-down: 09-22 close 15949, 09-23 close 16165, 09-24 close 15661, 09-28 close 14985, 09-29 close 14848, with 09-28 already breaking below the 15000 round-number zone and 09-29 further printing a range low of 14745. On the weekly timeframe: the last completed weekly bar (2026-09-21 to 09-24) opened at 16250, high 16400, low 15580, close 15661, a weekly change of -3.94%, a bearish candle with a relatively long body; the current week (from 09-28) has not yet closed, currently printing 14848, so no weekly conclusion is drawn for this week. In China's night session, from 21:05 on 09-29 to 02:30 on 09-30, SHFE Silver's night session opened at 14846, high 15022, low 14840, close 14958, up +0.74% versus the day-session close and +0.35% versus the settlement, a weak bounce after the breakdown that failed to reclaim 15031.7 (R1). On the international comparison, SI=F settled at 61.14, -0.94% on the day, -8.1% over 5 days and -8.73% over 20 days, at the 1.8% percentile of the 20-day channel of 61–68.98, with ATR14 at 2.49 (4.07% of price) and RV20 at 36.2%; the international and Chinese markets are in sync, both showing weak consolidation near the lower bound of the channel.
III. Supply-Demand Balance & Fundamental Drivers
Inventories and warehouse warrants are currently the most direct marginal variable. SHFE silver warehouse warrants stood at 1460361 kg (2026-09-29), up 21070 kg day-on-day (+1.5%), with a cumulative +37045 kg over 5 periods; registered warrants have risen for consecutive periods, indicating deliverable supply is concentrating toward the exchange, exerting physical pressure on near-month prices. On the international side, COMEX registered stood at 2994644 kg (2026-09-25), unchanged day-on-day, with no synchronized drawdown, so a “international squeeze” cannot explain Chinese prices; instead, it indicates no sign of tightening in global visible inventories. On basis, spot at 14779 versus main contract ag2612 at 14848 gives -69.0 (-0.47%), versus the prior +136.0, a change of -205.0, flipping from premium to discount; a deepening discount means the spot side is no longer tight, with traders' and downstream buying interest weaker than the screen, corroborating the rise in warehouse warrants. On processing and capacity, this report does not incorporate verifiable smelting margin and mine supply-demand balance data, so no quantitative judgment is made on the mine side or smelting margins, using only visible inventories and basis as fundamental proxies. On the macro transmission chain, the directly relevant factors for this instrument are the dollar and rates: the US 10-year Treasury yield at 5.255 (+0.29%) and the dollar index at 101.38 (+0.17%), with rates and the dollar strengthening in tandem, raising the opportunity cost of holding silver and pressuring the valuation of non-yielding assets; meanwhile, the gold-silver ratio at 68.02 (1-year percentile 65.87%, 3-year percentile 21.96%) shows silver on the weaker side relative to gold, with capital within precious metals favoring gold over silver. International news flow is also bearish: over the past 48 hours, international headlines point to silver still facing further downside risk under the macro narrative dominated by US-Iran negotiations, and mining equities have fallen more sharply during the price correction, reflecting equity markets pricing silver's price elasticity negatively. Overall, the three threads of inventory accumulation, basis flipping to a discount, and a stronger dollar and rates all point the same way, and fundamentals provide no basis for going long.
IV. Positioning & Fund Flows
China positioning: main-contract notional open interest of CNY 61.73bn, daily change +CNY 3.323bn, corresponding to open interest +14920 lots and settlement price -2.17%; the volume-price combination is an open interest build on a price decline, bearish initiation rather than long absorption. The top-20 seats total 286380 lots long (+2592) and 242338 lots short (+1834), for a net long of 44042 lots (+758); the net long rose slightly but by a magnitude far smaller than the price decline, indicating that at the seat level longs did not mount effective resistance, with short additions aligned with the price direction. On the international side, COT (as of 2026-09-22): non-commercial longs 19303, shorts 5994, net long 13309, weekly change +185; the prior three weeks' net longs were 13124 (-1262), 14386 (+1788), 12598 (-1475), with the net long broadly oscillating in the 12500–14400 range, while over the same period SI=F fell -8.1% over 5 days and -8.73% over 20 days, forming a divergence of “net long not clearly declining while prices fall sharply” — this typically means longs have not completed their clearing, leaving subsequent downside pressure from passive deleveraging. On crowding, netPct 12.5%, crowd 55.01, CTA 62, hedge 24.99%, down from netPct 13.93% and crowd 58.85 on 09-08; crowding is cooling but the absolute level remains elevated. Volatility structure: ^VXSLV at 39.13 (-0.32 points on the day), SI=F RV20 at 36.2%, IV−RV at +2.9 volatility points, IV/RV at 1.08; the options side still carries a small premium for event risk, but the premium is limited, indicating the market has not priced an extreme tail; versus ^GVZ 24.37 (1-year percentile 31%) and ^VIX 16.04 (1-year percentile 28%), silver's implied volatility is clearly higher than gold's and equities', making long-volatility trades of average value, with directional positions relying more on levels than on volatility expansion.
V. Cross-Asset & Onshore-Offshore Value
Onshore-offshore value: SHFE silver, after removing VAT and converting at USD/CNH, carries a premium/discount of -0.36% versus SI=F (2026-09-29), versus -0.13% five days ago, with the premium turning further negative, i.e., China is cheaper relative to the international market, tilting the import window toward closure and widening import losses. The direct implication of this structure for Chinese prices: if the international market continues to weaken, China lacks the buffer of “onshore-offshore spread repair” and is highly likely to follow lower; conversely, if the international market stabilizes, China's rebound elasticity is also capped by the negative premium. The gold-silver ratio is 68.02, with a 1-year percentile of 65.87% and a 3-year percentile of 21.96%; on a short-cycle basis silver is weak relative to gold, while on a long-cycle basis the ratio remains in the lower part of its 3-year range, meaning silver's valuation relative to gold is not extremely cheap, and a long-silver/short-gold relative-value strategy lacks a margin of safety. The term structure (international comparison) is in CONTANGO, with M1-M2 at -0.33 (-0.54%), RY at -6.49%, slope 0.3211; the deferred-month discount structure does not indicate near-end squeeze. On risk metrics, SI=F's 52-week drawdown is 51.43%, 20-day drawdown 10.93%, Sharpe30 -1.944, VaR95 -6.28%, with risk-adjusted returns negative, further supporting not chasing longs.
VI. Historical Statistics & Seasonality
Seasonality statistics (20 trading days from the same date over the past 10 years): average +0.84%, median +2.73%, up in 7 out of 10 years, best 2022 +4.52%, worst 2014 -9.51%. This sample shows that silver seasonality from early October to month-end is mildly positive, with the median gain exceeding the mean, indicating a right-skewed distribution with moderate gains in most years. Two caveats apply: first, the sample covers only 10 years, so statistical power is limited and it can only serve as background rather than a trading basis; second, the current price has already broken below the 20-day low and sits at the 5.5th percentile of the 20-day channel, accompanied by a decline on open interest build and warrant accumulation, so the historically bullish seasonal pattern requires the price structure to stabilize first before it can materialize. This report therefore treats seasonality as “a supplementary加分项 if a stabilization signal emerges,” rather than a reason to go long against the trend.
VII. Scenario Analysis
Base case (55% probability, bearish bias): Trigger conditions are weak consolidation within the 14745–15031.7 range, with rebounds capped below R1 15031.7, continued modest warrant accumulation, and the basis remaining at a discount. Target prices: downside toward below S1 14704.7, with further reference to the extension space after an effective break below the 20-day channel lower bound at 14745; based on 1 ATR14 (499.5), the next reference band below 14745 is approximately 14250–14300. Corresponding action: hold shorts or sell on rebounds, with a stop-loss set about 1 ATR above 15031.7.
Bullish scenario (20% probability): Trigger conditions are the settlement price reclaiming above 15031.7 (R1) for two consecutive days, alongside SHFE warrants turning to destocking, the basis recovering from -69.0 to a premium, and the dollar index falling below 101. Target prices: upside toward the mid-point of the 20-day channel, referencing 15600–15700 (corresponding to the prior closed weekly close around 15661). Corresponding action: exit shorts and switch to light long probes at the lower bound of the range, with a stop-loss set below 14745.
Bearish scenario (25% probability): Trigger conditions are a break below 14745 (the 20-day channel lower bound) accompanied by continued increases in main-contract positioning and the basis discount widening beyond -0.5%, or US employment and core PCE data beating expectations and pushing up the dollar and Treasury yields. Target prices: based on 1 ATR14 extension, downside toward around 14250, with the extreme case referencing lower ranges against the backdrop of the 52-week drawdown. Corresponding action: add shorts with the trend, with a stop-loss set above 15031.7.
The three scenarios sum to 100%, and the base case direction is consistent with the bearish conclusion in Section I.
VIII. Trading Strategies & Risk Management
Strategy One (trend-following short, aligned with direction): Entry range 14900–14960 (referencing the settlement at 14906 and the night session close at 14958), stop-loss about 1 ATR above 15031.7, i.e. approximately 15530, target below 14704.7, with further downside at 14250–14300, horizon 1–5 trading days, position size no more than 15% of total capital, with single-trade risk controlled within 1% of equity. Strategy Two (sell on rebound, conservative version): If the price rebounds and stalls near 15031.7, enter at 15000–15030, stop-loss 15530, target 14704.7, horizon 1–3 trading days, position size no more than 10%. Both strategies are short in direction, consistent with the conclusion in Section I. Risk note: ATR14 is 499.5 and RV20 is 38.9%, so volatility is elevated; stop-losses must allow sufficient room to avoid being stopped out by full intraday ranges; if the settlement price closes above 15031.7 for two consecutive days, exit immediately and move to the sidelines.
IX. This Week's Data Calendar
Beijing time 09-30 09:30 China non-manufacturing PMI (expected 49.2, previous 49.0) and manufacturing PMI (expected 50.1, previous 49.8); 09:45 RatingDog services PMI (expected 51.3) and manufacturing PMI (expected 51.7). 09-30 20:30 US personal income MoM (expected 0.4%), personal spending MoM (expected 0.8%), final GDP (expected 1.5%), core PCE MoM (expected 0.3%). 10-02 20:30 US average hourly earnings MoM (expected 0.3%), nonfarm payrolls (expected 90K, previous 162K), unemployment rate (expected 4.1%). 10-05 22:00 ISM services PMI (expected 54). 10-08 02:00 FOMC meeting minutes. The above events affect SHFE Silver through the dollar and interest rate channels.
This report is generated automatically from public quantitative and macro data for research and market tracking only. It does not constitute any investment advice.