Loading…
A statistical projection of where silver could trade over the next seven trading sessions, shown as a range rather than a single number — and, below it, how accurate that projection has actually been.
Prices last updated
Shaded bands are the 80% and 95% ranges. The dashed line is the central projection — the least reliable thing on this chart.
Across 180 historical starting points, this projection was no more accurate than simply assuming the price would not change — which is what a week-ahead metals forecast usually is. The useful part of this page is the range, not the line.
| Days ahead | Model error | No-change error | Skill | 80% range held |
|---|---|---|---|---|
| 1 | 3.10% | 3.11% | 1.000 | 84% |
| 2 | 4.43% | 4.44% | 1.000 | 82% |
| 3 | 5.61% | 5.62% | 0.998 | 81% |
| 4 | 6.54% | 6.53% | 1.001 | 80% |
| 5 | 7.37% | 7.35% | 1.003 | 78% |
| 6 | 8.08% | 8.10% | 0.998 | 77% |
| 7 | 8.59% | 8.60% | 0.998 | 77% |
Measured by walk-forward testing: the model is refitted at each of 180historical starting points using only the data available then, and compared with what actually happened. Skill below 1 means it beat the no-change benchmark. “80% range held” is how often the real price landed inside the 80% band — a figure above 80% means the band is wider than it strictly needs to be.
| Date | Central | 80% range | 95% range |
|---|---|---|---|
| 2026-09-14 | $65 | $63 – $68 | $62 – $69 |
| 2026-09-15 | $65 | $62 – $69 | $60 – $70 |
| 2026-09-16 | $65 | $61 – $69 | $59 – $72 |
| 2026-09-17 | $65 | $61 – $70 | $58 – $73 |
| 2026-09-18 | $65 | $60 – $71 | $58 – $74 |
| 2026-09-21 | $65 | $60 – $71 | $57 – $75 |
| 2026-09-22 | $65 | $59 – $72 | $56 – $75 |
Ranges widen with the square root of the horizon, from silver's own recent volatility — currently about 44.6% annualised. Anchored to the 2026-09-11 close of $65.
Starting from $65 on 2026-09-11, the central projection seven trading days out is $65, with an 80% range of $59 to $72. The range is the meaningful part: the central figure is close to the current price because that is what the data supports.
Across 180 historical starting points, this projection was no more accurate than simply assuming the price would not change — which is what a week-ahead metals forecast usually is. The useful part of this page is the range, not the line.
Because a week-ahead metals price is close to a random walk. Fitting this model across an out-of-sample window found that every version with a trend component performed worse than one without — a trend line drawn across the next seven days would be inventing information the price series does not contain. What can be estimated honestly is how far the price is likely to travel, which is the shaded range.
No. This is a statistical projection from past closes, published for context rather than as a recommendation. It knows nothing about interest rates, central bank buying, geopolitics or anything else that actually moves silver, and no part of this site is financial advice.
Nothing on this page is financial advice or a recommendation to buy or sell. Prices are indicative reference figures and may be delayed. See the terms.