Twenty years of gold, silver & copper — and the ten-year theses that follow.
A backtest of the three metals ETFs in USD — and in ZAR where an observed exchange rate exists — read against the June 2026 macro, technology and financial landscape, distilled into four ten-year theses, each with stated refutation conditions.
Over the observed-FX window (7 Nov 2014 – 22 Jun 2026) the rand's depreciation — USD/ZAR R11.26 to R16.41, 3.3% a year — made rand-measured returns of the three dollar metals ETFs materially higher than their dollar returns. Whether it reduced drawdowns depended on timing, and the observed record is mixed: gold's worst rand drawdown was deeper than its dollar drawdown; copper's was not.
- Gold compounded fastest of the three over the long windows — 10.6%/yr in USD over 31 Dec 2004 – 22 Jun 2026 (silver led over 7 Nov 2014 – 22 Jun 2026: 12.4% vs 11.1%) — and had the highest USD return per unit of volatility in every window measured (USD Sharpe 0.59 from GLD's 18 Nov 2004 inception; 0.42 over 30 Dec 2011 – 22 Jun 2026; 0.70 over 7 Nov 2014 – 22 Jun 2026).
- The rand is a currency exposure, not a free addition to return — USD/ZAR rose 45.8% over 7 Nov 2014 – 22 Jun 2026, which lifted rand returns; from its R19.82 high on 25 May 2023 it fell to R16.41 by 22 Jun 2026 (−17.2%), so GLD's +113.4% in USD over that span was +76.7% in rand.
- Rand drawdowns were not reliably smaller — over 7 Nov 2014 – 22 Jun 2026, gold's worst rand drawdown was −25.2% (month-end, Jul 2020 – Mar 2021) against −20.5% in USD; silver's was similar in both currencies (−36.1% vs −36.8%); copper's was smaller in rand (−19.2% vs −34.2%).
- Silver is a volatility problem, not a buy-and-hold — 7.5% USD CAGR from SLV's 28 Apr 2006 inception, with 32.3% annualised volatility and a −72.2% month-end peak-to-trough fall (Apr 2011 – Mar 2020; −76.3% on daily closes).
- CPER's weak return mostly reflects copper's own path — CPER returned +58.4% in USD over 30 Dec 2011 – 22 Jun 2026 (3.2%/yr). On monthly averages from Dec 2011 to May 2026 (approximate), the copper price rose 4.1%/yr and CPER 3.2%/yr; the gap of 1.0 percentage points a year combines futures roll, fees and basis, net of collateral income.
Corrected 2026-10-01 (D-514). No observed USD/ZAR series was available when this note was first prepared, and its rand series was reconstructed from anchors. Its intermediate points did not match observed USD/ZAR — implied R20.34 / R24.86 / R29.86 / R22.98 at year-ends 2018/20/22/24 against observed R14.35 / R14.69 / R17.02 / R18.86 — and have been withdrawn, together with every rand drawdown and volatility figure derived from them and the statements built on those figures. Rand path statistics are now shown only where observed FX exists (7 Nov 2014 – 22 Jun 2026). All statistics were re-measured on 1 Oct 2026 from the same vendor's data; prices still end 22 Jun 2026.
Prices: daily adjusted closes for GLD, SLV and CPER via Alpha Vantage. Windows are stated with every figure: each ETF from its first date (GLD 18 Nov 2004, SLV 28 Apr 2006, CPER 15 Nov 2011); a common window 30 Dec 2011 – 22 Jun 2026; and the observed-FX window 7 Nov 2014 – 22 Jun 2026. Figures from different windows are not comparable.
Currency: Alpha Vantage FX_DAILY USD/ZAR, first observation 7 Nov 2014. A rand price is the dollar close times that day's USD/ZAR close; 1 ETF trading day without an FX observation (29 Dec 2014) carries the previous close. The Dec-2004 rate of R5.75 is unverified; rand figures that depend on it are labelled estimates and carry no path statistics.
Statistics: CAGR is endpoint-to-endpoint and is reported separately from path statistics. Volatility uses full calendar-month returns, annualised. Maximum drawdown is given on month-end points and, where noted, daily closes. USD Sharpe uses the 3-month US Treasury yield as the risk-free rate; a rand Sharpe ratio is withheld because no authorised South African risk-free series is available.
Two currencies, one observed window
Endpoint returns and path statistics answer different questions. Over 7 Nov 2014 – 22 Jun 2026 the rand's depreciation added roughly its own 3.3% a year to each ETF's rand CAGR. On monthly data, each ETF's USD return was negatively correlated with the change in USD/ZAR (GLD −0.35, SLV −0.44, CPER −0.50), which damped rand volatility for silver and copper but not for gold: gold's rand volatility (16.7%) exceeded its dollar volatility (15.1%).
Observed-FX window, 7 Nov 2014 – 22 Jun 2026: USD and ZAR
| GLD USD | GLD ZAR | SLV USD | SLV ZAR | CPER USD | CPER ZAR | |
|---|---|---|---|---|---|---|
| CAGR (endpoint) | 11.1% | 14.8% | 12.4% | 16.1% | 5.7% | 9.1% |
| Volatility (monthly, annualised) | 15.1% | 16.7% | 28.7% | 25.7% | 20.7% | 18.4% |
| Max drawdown, month-end | −20.5% | −25.2% | −36.8% | −36.1% | −34.2% | −19.2% |
| Max drawdown, daily | −24.5% | −30.8% | −45.4% | −42.6% | −38.4% | −27.5% |
| Best 12 months | +83.7% | +69.4% | +200.2% | +156.2% | +86.0% | +52.3% |
| Worst 12 months | −13.4% | −21.6% | −25.1% | −27.7% | −29.1% | −12.3% |
| USD Sharpe | 0.70 | withheld | 0.53 | withheld | 0.30 | withheld |
| Corr. of USD return with USD/ZAR change | −0.35 | — | −0.44 | — | −0.50 | — |
Alpha Vantage adjusted closes and FX_DAILY USD/ZAR, 7 Nov 2014 – 22 Jun 2026; 138 full calendar months (Dec 2014 – May 2026) for volatility, correlation and Sharpe. Rand Sharpe withheld: no authorised South African risk-free series.
USD only: each ETF from inception, and the common window
| GLD | SLV | CPER | |
|---|---|---|---|
| First date | 18 Nov 2004 | 28 Apr 2006 | 15 Nov 2011 |
| CAGR from first date | 10.5% | 7.5% | 3.0% |
| Volatility from first date | 17.0% | 32.3% | 19.8% |
| Max drawdown from first date, month-end | −42.9% | −72.2% | −50.0% |
| Max drawdown from first date, daily | −45.6% | −76.3% | −54.0% |
| USD Sharpe from first date | 0.59 | 0.35 | 0.17 |
| CAGR, 30 Dec 2011 – 22 Jun 2026 | 6.6% | 5.6% | 3.2% |
| Volatility, 30 Dec 2011 – 22 Jun 2026 | 15.6% | 29.0% | 19.8% |
| USD Sharpe, 30 Dec 2011 – 22 Jun 2026 | 0.42 | 0.31 | 0.18 |
Rand returns for these windows are withheld: each starts before the first observed USD/ZAR (7 Nov 2014). The one long-run rand figure on this page — gold at an estimated 16.2%/yr over 31 Dec 2004 – 22 Jun 2026 — rests on the unverified Dec-2004 rate of R5.75 and is an estimate.
1 · Gold had the highest USD return per unit of volatility of the three ETFs in each window measured (USD Sharpe 0.59 from inception, 0.42 over 30 Dec 2011 – 22 Jun 2026, 0.70 over 7 Nov 2014 – 22 Jun 2026).
2 · The rand was a return contribution over this window; drawdown effects depended on timing. USD/ZAR rose 3.3% a year over 7 Nov 2014 – 22 Jun 2026, but gold's worst rand drawdown (−25.2%, Jul 2020 – Mar 2021, month-end) came while the rand was strengthening.
3 · Silver is the highest-torque metal — in both directions. Best 12 months +200.2% (to Feb 2026) and worst −40.5% (to Nov 2013) in USD from SLV's inception; in the theses below it is a sized, tactical expression.
4 · Copper-the-metal ≠ copper-the-ETF, but the gap is modest. The copper price rose 4.1%/yr and CPER 3.2%/yr (monthly averages, Dec 2011 – May 2026, approximate); CPER's weak return mostly reflects copper's own path. Producers carry a different exposure — operating leverage, costs and jurisdiction risk — not the futures index.
5 · Diversification is real. GLD and CPER monthly USD returns correlated 0.32 over 30 Dec 2011 – 22 Jun 2026 (0.26 over 7 Nov 2014 – 22 Jun 2026): the monetary and industrial legs moved partly independently.
The theses
Each thesis follows the MCH macro-lens discipline: a single testable claim, the indicators that confirm or refute it, the mechanism and its counter, US-listed expression vehicles, review triggers, and a kill switch. Probabilities are explicit and sum to 100%. US-listed comparatives only.
Disclaimer. MCH Analysis is a general investment-research publication of MCH Advisory (Pty) Ltd. It is not personal financial advice and does not take your circumstances into account. MCH is not a licensed financial services provider or a registered investment adviser. Past performance, whether actual or simulated, does not predict future results.
Data & sources. Price history: Alpha Vantage (daily adjusted closes for GLD, SLV, CPER; FX_DAILY USD/ZAR from 7 Nov 2014; COPPER monthly). Risk-free rate: 3-month US Treasury yield, MCH research warehouse. Statistics re-measured 1 Oct 2026 by scripts/research/metals-fx-recompute.mjs. Macro: US Federal Reserve / FOMC, Trading Economics, CNBC. Commodities: Wood Mackenzie & IEA critical-minerals outlooks; JPMorgan, Goldman Sachs, Bank of America copper research; industry reporting on AI power. All third-party figures paraphrased.