Silver (SI=F) Deep Dive | 2026-09-27 — Long the 62.75–63.89 Retest, Invalidation 62.75 | 2026-09-27

Call: LONG silver (SI=F, front month SIZ26.CMX) on the pullback into 63.89–62.75, with the trade invalidated on a settle below 62.75. Three reasons: (1) price has retraced 7.74% over 20 sessions to the 22nd percentile of the 20-day channel (62.75–72.05) while the last completed weekly bar (2026-09-14–18) closed at 67.149, +3.01% w/w — the higher low structure is intact; (2) the options market is not paying up for downside, with ^VXSLV at 35.99 versus RV20 of 34.4% (IV−RV +1.6 vol pts, IV/RV 1.05) and gold implied vol at the 13th percentile of its 1-year range; (3) seasonality for the same calendar window over the last 15 years is mean +2.12%, median +2.56%, up 11 of 15 years. Positioning is not crowded (netPct 12.5%, 55th percentile of 3 years), so there is room for fresh length. The bear case is a hawkish core PCE (forecast 0.3% m/m) that pushes the 10-year yield (5.18%) and DXY (100.97) higher and forces a settle below 62.75.

1. Bottom Line & Directional Bias

Call: LONG silver (SI=F, front month SIZ26.CMX) on the retest of 63.89–62.75. Invalidation: a daily settle below 62.75.

Three reasons. First, the correction is mature but the structure is not broken: settle 64.801 [2026-09-25], 5D -3.5%, 20D -7.74%, sitting at the 22nd percentile of the 20-day channel (62.75–72.05), while the last completed weekly bar (2026-09-14–18) closed at 67.149, +3.01% w/w. The pullback is a retracement inside a higher-low sequence, not a trend reversal. Second, optionality is not pricing a crisis: ^VXSLV 35.99 versus RV20 34.4% gives IV−RV of just +1.6 vol pts (IV/RV 1.05), and ^GVZ at 22.44 sits in the 13th percentile of its 1-year range — the market is not paying up for a silver breakdown. Third, the seasonal window is supportive: same calendar start, next 20 sessions, last 15 years, mean +2.12%, median +2.56%, up 11 of 15 years.

Positioning supports adding length rather than chasing: managed-money net is 12.5% of open interest, only the 55th percentile of the 3-year window — not crowded. The invalidation is a settle below 62.75, which would break both the 20-day low and the higher-low structure and shift the call to neutral.

2. Price Action & Technical Analysis

Settle 64.801 [2026-09-25], +1.25% on the day (settle), but 5D -3.5% and 20D -7.74% (settle). The 20-day channel runs 62.75–72.05, placing price at the 22nd percentile — the lower quarter of the recent range. The 52-week range is 43.785–121.3, so the current level is roughly mid-range on a 12-month view but clearly at the bottom of the short-term distribution.

ATR14 is 2.502, i.e. 3.86% of price — the full expected daily range, not a one-sided band. RV20 is 34.4% annualized. With ATR at 3.86% of price, a stop must sit beyond a real level and at least roughly one ATR away; the 62.75 20-day low is the natural structural line.

Pivots from the settle-based snapshot: P 64.697, R1 65.609, S1 63.889, R2 66.417, S2 62.977. Price settled at 64.801, just above P — a mildly constructive posture, but the sequence of last five settled bars shows distribution: 09-21 C 66.415, 09-22 C 66.53, 09-23 C 64.964, 09-24 C 64.002, 09-25 C 64.801. The 09-25 settle bounced off the 63.505 low of 09-24, leaving 62.75 (20-day low) and 62.977 (S2) as the support shelf, with 65.609 (R1) and 66.417 (R2) as the first resistance band.

On the weekly frame, the last completed bar (2026-09-14–18) opened 64.785, high 67.895, low 62.75, closed 67.149, +3.01% w/w — a bullish weekly bar with a long lower wick. The current week (from 2026-09-21, five sessions) is unfinished and last printed 64.801, -3.5%; no weekly-close conclusion can be drawn from it. The report-date bar is an unfinished Globex/Asia bar and is not a close or settlement.

View: the 62.75–63.89 shelf is the decision zone. Holding it keeps the higher-low structure alive and targets 65.61 then 66.42; losing it on a settle opens 62.977 and then the low-60s.

3. Supply-Demand Balance & Fundamental Drivers

COMEX registered silver stood at 96.28 Moz (2,994,645 kg) as of 2026-09-25, unchanged d/d (+0%) from 2026-09-24. The flat registered stock is the key physical fact: no visible drain, but also no rebuild — the deliverable pool is stable at a level that leaves the front spread sensitive to any pickup in industrial or investment demand. SHFE warrants were 1,443,896 kg as of 2026-09-24, up 1,753 kg d/d (+0.12%) from 1,442,143 kg on 2026-09-23; the prior day's build was 4,494 kg. The Chinese exchange stock is drifting higher, a mild offset to the flat COMEX picture, but the absolute size is small relative to COMEX and does not change the global balance.

The term structure is in contango: M1-M2 -0.27 (-0.42%), roll yield -5.02%, slope 0.2948. Contango here reflects carry — the cost of financing and storage against a 5.18% 10-year yield — not a signal that price is capped. For a long, it is a roll cost, and at roughly -0.42% per month it is a manageable drag that argues for a multi-week rather than multi-month hold unless the curve flattens.

Macro transmits through two channels. First, the dollar: DXY at 100.97, -0.32% [2026-09-25], a softer dollar is a tailwind for dollar-denominated silver. Second, real rates: ^TNX at 5.184, +0.43% [2026-09-25], still elevated; a further rise in nominal yields without a matching inflation print is the main fundamental risk to the long. The gold/silver ratio at 66.68 (P1Y 54.37%, P3Y 18.19%) shows silver has been structurally strong versus gold over the three-year window — the ratio sits in the bottom fifth of its 3-year distribution, meaning silver has outrun gold, not lagged it. That is a description of the current state, not a forecast of mean reversion.

View: flat COMEX registered stock plus a small SHFE build is a neutral-to-slightly-constructive physical backdrop; the driver for the next move is the rates/dollar channel, and a softer DXY with stable yields is the combination that lets the 62.75 shelf hold.

4. Positioning & Fund Flows

CFTC managed-money positioning (latest report as of 2026-09-22, 15 days old — not the current week): open interest 106,474, longs 19,303, shorts 5,994, net 13,309, change +185 w/w. The prior weeks: 2026-09-15 net 13,124 (Δ -1,262), 2026-09-08 net 14,386 (Δ +1,788), 2026-09-01 net 12,071. Net length has been broadly stable in the 12,000–14,400 range for a month.

Crowding: netPct 12.5% as of 2026-09-22, 55th percentile of the 3-year window; 2026-09-15 was 12.65% (55.04), 2026-09-08 13.93% (58.85), 2026-09-01 12.07% (53.27). The percentile is mid-range, so the long is not crowded — there is room for fresh managed-money length without a positioning squeeze. The CTA trend proxy reads 62 across all four weeks, i.e. trend followers remain net long, and hedge pressure is around 25%.

Note the sequencing: the latest COT snapshot (2026-09-22, net 13,309) predates the 09-23 to 09-25 decline to 64.801, so the reported net length does not yet reflect the pullback. The absence of a crowding extreme means the recent selling is more likely profit-taking than a positioning unwind. Implied vol confirms: ^VXSLV 35.99, -1.81 pts on the day, versus RV20 34.4% — IV/RV 1.05, so options are roughly fairly priced and there is no panic bid for downside protection.

View: positioning is a neutral-to-supportive input. Mid-range crowding plus a trend-following community still net long argues the 62.75 shelf can absorb selling; a break would more likely come from a macro shock than from a positioning flush.

5. Cross-Asset Relative Value

The relevant cross-market metric is the gold/silver ratio: 66.68, with a 1-year percentile of 54.37% and a 3-year percentile of 18.19%. Read correctly: a low 3-year percentile means the ratio has been falling over the three-year window, i.e. silver has been outperforming gold structurally. The 1-year percentile at 54.37% is mid-range, so the recent 20-day silver underperformance (-7.74%) has taken some froth out of the ratio without reversing the multi-year trend.

Against the dollar, silver's -7.74% 20-day move came while DXY was at 100.97, -0.32% [2026-09-25] — silver underperformed a softening dollar over the month, which is the kind of divergence that typically resolves in silver's favor when the macro impulse stabilizes. Against rates, ^TNX at 5.184, +0.43%, is the competing asset: at a 5.18% nominal yield, silver's carry disadvantage (roll yield -5.02%) is real, so the long case needs either a yield pause or continued dollar softness.

Cross-asset vol is informative: ^VIX 14.87, 1Y percentile 9%, and ^GVZ 22.44, 1Y percentile 13% — broad risk and gold vol are both cheap, while ^OVX at 55.09 sits at the 58th percentile. Silver's own ^VXSLV at 35.99 is the highest of the group, consistent with silver's higher beta. The configuration — cheap gold vol, cheap equity vol, mid-range oil vol — is not a risk-off regime, which historically favors silver over gold on a relative basis.

View: silver's structural strength versus gold (3-year ratio percentile 18.19%) is intact; the near-term relative-value trade is silver catching up to a softening dollar once the 62.75 shelf is confirmed.

6. Historical & Seasonal Patterns

Seasonality block only: same calendar start, next 20 sessions, last 15 years — mean +2.12%, median +2.56%, up 11 of 15 years. Best case 2015 +9.37%, worst case 2016 -7.87%. The sample is small (15 observations) and is context, not a standalone signal.

The distribution is positively skewed in the median but with a fat left tail: the worst year (-7.87%) is larger in magnitude than the best year (+9.37%) is above the mean, so the window rewards a defined-risk long rather than an oversized position. The 11-of-15 hit rate is the more robust statistic and aligns with the technical setup — a retest of the 20-day low inside a seasonally firm window.

View: seasonality tilts the odds toward a bounce over the next 20 sessions, but the -7.87% worst case is the reason the stop belongs at 62.75 rather than at a wider discretionary level.

7. Scenario Analysis (Base / Bull / Bear)

Base case — 55%: retest holds, grind higher. Trigger: price holds 63.89–62.75 on a closing basis and reclaims P 64.697. Path: 65.609 (R1) first, then 66.417 (R2), with the 20-day channel midpoint near 67.4 as the stretch target. Action: initiate or add to length in the 63.89–62.75 zone, stop below 62.75, scale at R1 and R2. This scenario agrees with the section 1 call.

Bull case — 25%: macro tailwind accelerates the move. Trigger: a soft core PCE (forecast 0.3% m/m, surprise threshold ±0.1%) or a weak ADP (forecast 73K, threshold ±35K) that pulls ^TNX below 5.18% and DXY below 100.97, combined with a Chinese PMI beat (Manufacturing forecast 50.1, threshold ±0.3). Path: a quick reclaim of 66.417 opens 67.895 (last completed weekly high, 2026-09-14–18) and then the 72.05 20-day high. Action: hold the core long, trail the stop to 63.89 after a settle above 66.417, and add only on a close above 67.895.

Bear case — 20%: macro shock breaks the shelf. Trigger: a hot core PCE (0.4%+ m/m) or a strong ADP (108K+) that pushes ^TNX above 5.3% and DXY back above 101.5, forcing a daily settle below 62.75. Path: 62.977 (S2) gives way quickly, opening the low-60s and a test of the 52-week structure well below the current channel. Action: exit the long on the settle below 62.75, stand aside, and re-engage only on a reclaim of 63.89 with a higher low. Note the 20-day drawdown is already 6.98% and VaR95 is -6.28%, so the bear path is a real, sized risk, not a tail curiosity.

8. Trading Strategies & Risk Management

Strategy 1 — Core long the retest (primary). Entry 63.9–62.9 (scale into the S1 63.889 to 20-day low 62.75 shelf), stop 61.9 (below the 62.75 structural low and beyond one ATR of 2.502 from entry), target 66.4 (R2 66.417), horizon 1–3 weeks, size 1.0x risk unit. Conviction 7/10. The stop sits beyond a real level; the target is the first resistance band, not the stretch case.

Strategy 2 — Momentum add on confirmation. Entry on a daily settle above 65.61 (R1 65.609), stop 63.5 (below the 63.505 low of 2026-09-24), target 67.9 (last completed weekly high 67.895), horizon 1–2 weeks, size 0.5x risk unit. Conviction 6/10. This adds only after the market proves the shelf held; it is not a second directional call.

Risk management: total silver exposure capped at 1.5x risk units; the roll cost of -5.02% annualized (M1-M2 -0.27) is a drag on holds beyond three weeks, so reassess the curve at each roll. The 52-week drawdown of 51.43% and VaR95 of -6.28% argue for the defined stops above rather than averaging down. Do not add to the core long on a settle below 62.75 under any circumstance.

9. This Week's Data Calendar

BJT 09-29 22:00 | ET 09-29 10:00 — USD JOLTS Job Openings (F 7.23M, surprise outside ±0.04M). BJT 09-30 09:30 | ET 09-29 21:30 — CNY Manufacturing PMI (F 50.1, ±0.3) and Non-Manufacturing PMI (F 49.2, ±0.2). BJT 09-30 20:15 | ET 09-30 08:15 — USD ADP (F 73K, ±35K). BJT 09-30 20:30 | ET 09-30 08:30 — USD Core PCE m/m (F 0.3%, ±0.1%), PCE YoY (F 3.7%, ±0.1%), Final GDP (F 1.5%, ±0.1%), Personal Spending (F 0.8%, ±0.6%). BJT 10-01 22:00 | ET 10-01 10:00 — USD ISM Manufacturing PMI (F 54.8, ±0.2).

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.

Silver (SI=F) Deep Dive | 2026-09-27 — Long the 62.75–63.89 Retest, Invalidation 62.75 | 2026-09-27

今日风险地图

单品种早报 深度分析

往期早报

⟠ 本月策略回测表现

单品种早报深度分析