Research · Dataset
The gold/platinum ratio, 1960–2026
Gold is described as a monetary metal and platinum as an industrial one, and the ratio between them is quoted as a reason to buy platinum. This page computes the ratio from 802 monthly averages, tests the two claims against the industrial cycle, and then asks the only question that matters for a trade: when the ratio was at this extreme, what actually happened next?
Last updated: 2026-10 (data through 2026-10) · Source data updated November 03, 2026
Source and method
Both legs are monthly average prices from the same workbook, so the comparison carries no source or frequency effect. It is not a tradable instrument: no storage, financing, spread or roll cost is included in either leg. A monthly average also smooths intramonth moves, so a ratio built from daily closes would be noisier and would not reproduce these figures.
One regime caveat runs through the whole page. US gold was under price controls at $35/oz from 1960 until mid-1968, and the ratio's all-time low of 0.26 in 1969-12 sits in that era and the 1970s inflation that followed. Where a statistic depends on which years you start from, this page reports every regime rather than the flattering one.
Where the ratio is now
The ratio sits at the 98th percentile of its own history, above the 1.6 90th-percentile line. On the conventional reading that means platinum is historically cheap against gold. The rest of this page is about whether that reading survives contact with the forward-return data.
In 2026-10, gold is $4,105/oz and platinum $1,621/oz. Gold is +1.2% over twelve months and platinum +0.3%; the ratio was 1.76 sixty months ago.
Decade by decade
| Period | Ratio mean | Ratio range | Gold | Platinum | Gold vol | Platinum vol | Months platinum beat gold |
|---|---|---|---|---|---|---|---|
| 1960-1969 | 0.38 | 0.26–0.44 | +0.0% | +58.3% | 5.3% | 6.1% | 18 / 119 |
| 1970-1979 | 0.67 | 0.26–1.04 | +1,200.0% | +351.9% | 21.2% | 17.1% | 41 / 120 |
| 1980-1989 | 0.94 | 0.69–1.31 | -39.4% | -37.2% | 24.2% | 26.5% | 58 / 120 |
| 1990-1999 | 0.89 | 0.64–1.04 | -31.0% | -11.8% | 10.5% | 12.5% | 63 / 120 |
| 2000-2009 | 0.56 | 0.43–0.97 | +299.6% | +225.9% | 13.9% | 22.3% | 59 / 120 |
| 2010-2019 | 1.14 | 0.67–1.75 | +32.3% | -40.7% | 11.6% | 15.7% | 50 / 120 |
| 2020-2029 (to 2026-10) | 2.17 | 1.45–3.38 | +163.0% | +64.2% | 13.4% | 26.8% | 31 / 82 |
Price totals, not annualised, and gross of any cost. Volatility is the annualised standard deviation of monthly changes within the decade.
The ratio's level has risen structurally: it averaged 0.38 in the 1960s and 1.14 in the 2010s. Any fixed threshold used to define "high" or "low" changes meaning every decade — a ratio of 1.0 meant something very different in 1960-1969 than it does today.
Does platinum really follow the industrial cycle?
This is the test of the "industrial metal" story. We correlated each metal's monthly return against the monthly change in US industrial production (FRED series INDPRO, 800 months), with a control row for calibration.
| Series | Correlation with industrial production change |
|---|---|
| Platinum monthly return | 0.0663 |
| Gold monthly return | -0.0409 |
| Industrial production vs itself (control) | 1.0000 |
Platinum's monthly return correlates 0.0663 with the industrial cycle and gold's -0.0409. Both are close to zero. On monthly data, neither metal shows a measurable co-movement with industrial output. The industrial demand story for platinum is a real demand story — autocatalysts, jewellery, substitution from palladium — but it does not appear as a monthly correlation with production, and it does not make platinum a reliable proxy for the industrial cycle.
We also scanned lags of −6 to +6 months. The strongest reading is 0.1705 at lag -1, against 0.0663 at lag 0 — so even the best-lagged version is weak, and platinum does not reliably lead the cycle. Gold's best lagged reading is 0.0633.
The test: what happened after an extreme ratio
We took the 20 months with the highest ratio and the 20 with the lowest since 1981, and looked at what each metal did over the following 12 months. Restricting to 1981 matters: every low-ratio month before then belongs to the price-control and 1970s inflation eras, so it is not comparable to a low ratio today.
The high tail is unambiguous: after the highest readings, platinum averaged +55.12% over the next year and gold +42.10%, with platinum up in 95% of those months. That is the mean-reversion story, and it is the one the ratio is usually sold on.
The low tail is the uncomfortable one. After the lowest readings, platinum averaged -18.30% over the next year and was down in 85% of those months, while gold averaged +7.54%. So "platinum is cheap versus gold" was not a buy signal in this sample — the cheapness persisted and then some. Gold rose in 90% of the low-ratio months.
Caveat on sample size: 20 months per tail drawn from a 45-year series. These are averages over a handful of very different environments — the low tail is dominated by 2001 and 2008, the high tail by 2025 — and they describe history, not a rule.
Conditioning the whole sample by ratio level
A cleaner version of the same question: split the months with a full 12-month forward window at the median ratio, and compare the two halves. The critical part is which months go into the sample.
| Sample | Months | High-ratio half: forward gold | Low-ratio half: forward gold | Difference | Non-overlapping p | Bootstrap 95% CI |
|---|---|---|---|---|---|---|
| 1960-2026 full sample | 790 | +3.58% | +16.48% | -12.91pp | 0.045 | -23.9 to -3.9 |
| 1971-2026 post price controls | 658 | +3.62% | +19.78% | -16.15pp | 0.051 | -27.9 to -5.6 |
| 1981-2026 modern regime | 538 | +4.69% | +8.41% | -3.72pp | 0.093 | -11.4 to 4.9 |
| 1991-2026 recent decades | 418 | +6.80% | +10.64% | -3.84pp | 0.211 | -13.2 to 5.2 |
Forward returns are 12-month changes in monthly average prices. The non-overlapping column takes every 12th month so no two windows share a month. The bootstrap is a moving-block bootstrap with 12-month blocks, 2000 resamples and a fixed seed.
Does the effect survive? Only in the old sample.
On the full 1960 sample the difference is -12.91 percentage points, and the bootstrap interval (-23.9 to -3.9) excludes zero. That is what the conventional claim is built on.
Start the sample in 1981 — the regime today's reading belongs to — and the difference falls to -3.72pp, the non-overlapping test gives p = 0.093, and the bootstrap interval (-11.4 to 4.9) includes zero. On the non-overlapping subsample of 45 windows the difference is -9.22pp.
The same test, on four different start dates, produces four different answers — and the two that look significant are the ones covering the 1970s. That is the signature of a regime effect, not a property of the ratio.
The naive t-test on the post-1981 sample gives t = -2.51, p = 0.0122, which looks convincing. It is inflated because 12-month forward windows overlap, so a bad month is counted in eleven different windows.
Does one metal move first?
Cross-correlating gold's monthly return against platinum's at lags −6 to +6 months, with gold correlated against itself as a control:
| Lag (months) | -6 | -5 | -4 | -3 | -2 | -1 | 0 | +1 | +2 | +3 | +4 | +5 | +6 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Platinum vs gold | -0.0761 | 0.0039 | 0.0716 | 0.0208 | -0.0708 | 0.1140 | 0.5695 | 0.1807 | -0.0114 | 0.0233 | 0.1408 | 0.0773 | -0.0071 |
| Gold vs itself (control) | -0.0300 | 0.0150 | 0.0747 | 0.0290 | -0.0359 | 0.2691 | 1.0000 | 0.2691 | -0.0359 | 0.0290 | 0.0747 | 0.0150 | -0.0300 |
The strongest reading is at lag 0 (0.5695), so the two metals move together, not one after the other. Every lagged value is smaller in absolute terms.
The control row is what makes the test meaningful. Correlating gold's return against itself shifted by one month produces 0.269 at the strongest non-zero lag (lag -1). That is the noise floor: any apparent lead-lag structure with a correlation near 0.27 is indistinguishable from an artefact of the same series against itself.
Correlation traps
- Ratio change vs gold return: 0.3019. The trap here is arithmetic, not statistics. Because the ratio is gold divided by platinum, its change is (to first order) the gold return minus the platinum return. So this correlation is a restatement of how often platinum moves more than gold — not a discovered relationship. A "signal" built on the ratio is largely a bet on which leg moves more.
- Ratio level vs gold level: 0.7775. Gold and platinum correlate 0.5695 in monthly returns over 801 months, and 0.7775 in price levels. The level figure is inflated by two trending series and by the fact that platinum's ratio to gold has risen structurally since the 1960s. Reading a level correlation as a signal is reading a trend artefact.
- Rolling 36-month return correlation: median 0.5720, but negative in 8.8% of 706 windows, reaching -0.5539 at its lowest. The two metals are related most of the time and occasionally behave like unrelated assets. Sizing them as separate risks is wrong on average and occasionally right — which is a hard way to be wrong.
Platinum's volatility premium
The trailing 60-month ratio of annualised volatility (platinum ÷ gold) currently sits at 1.89, with gold at 14.0% and platinum at 26.3% annualised.
The widest reading is an artefact of the 1960s: gold was pegged at $35/oz, so its measured volatility was tiny while platinum's was not. The tightest reading (1974-01) is the modern case where platinum was momentarily calmer than gold. Median across the whole sample: 1.33×.
Practically: platinum currently carries about 1.9× gold's volatility. Sizing a platinum position at the same notional as a gold position runs roughly 1.9 times the risk. The position size calculator exists for exactly this adjustment.
Which metal won each year
Over the 66 calendar years in the sample, platinum beat gold in 26 of them (39.4%). Gold won the 1970s, the 1980s, the 2000s and the 2010s; platinum's wins are concentrated in the inflation years of the 1970s and a few later periods.
| Year | Gold | Platinum | Ratio start | Ratio end |
|---|---|---|---|---|
| 2025 | +59.0% | +99.5% | 2.86 | 2.28 |
| 2024 | +30.2% | +1.3% | 2.20 | 2.82 |
| 2023 | +6.7% | -11.2% | 1.80 | 2.17 |
| 2022 | -1.0% | +1.7% | 1.83 | 1.78 |
| 2021 | -4.1% | -13.4% | 1.71 | 1.89 |
| 2020 | +19.0% | +4.2% | 1.58 | 1.81 |
| 2019 | +14.5% | +14.5% | 1.60 | 1.60 |
| 2018 | -6.1% | -20.1% | 1.34 | 1.58 |
| 2017 | +6.0% | -6.6% | 1.23 | 1.39 |
| 2016 | +5.4% | +7.4% | 1.28 | 1.26 |
| 2015 | -14.0% | -30.8% | 1.01 | 1.25 |
| 2014 | -3.5% | -14.5% | 0.88 | 0.99 |
| 2013 | -26.9% | -17.3% | 1.02 | 0.90 |
| 2012 | +1.9% | +5.5% | 1.10 | 1.07 |
| 2011 | +20.6% | -18.5% | 0.76 | 1.13 |
| 2010 | +24.4% | +9.8% | 0.72 | 0.81 |
| 2009 | +32.1% | +51.7% | 0.91 | 0.79 |
| 2008 | -8.3% | -46.9% | 0.56 | 0.97 |
| 2007 | +27.3% | +30.2% | 0.55 | 0.54 |
| 2006 | +14.5% | +9.3% | 0.54 | 0.56 |
| 2005 | +20.3% | +13.7% | 0.49 | 0.52 |
| 2004 | +6.8% | -0.2% | 0.49 | 0.52 |
| 2003 | +14.0% | +28.7% | 0.57 | 0.50 |
| 2002 | +17.7% | +26.1% | 0.60 | 0.56 |
Last 24 calendar years shown; the full series is in the downloadable CSV. Year figures are January-to-December changes in monthly average prices.
Does adding platinum to a gold position help?
A monthly-rebalanced, cost-free blend of the two, over the period since 1971 (the gold series starts at $35/oz before that, so the pre-1971 years are not a meaningful allocation period):
| Platinum weight | Final multiple | Annualised return | Max drawdown |
|---|---|---|---|
| 0% | 110.9× | 8.80% | -62.1% |
| 20% | 81.4× | 8.20% | -60.3% |
| 50% | 45.2× | 7.07% | -62.4% |
| 80% | 21.7× | 5.66% | -67.2% |
| 100% | 12.2× | 4.58% | -70.4% |
Adding platinum lowered the return at every weight tested, and the drawdown barely improved: the 50/50 blend returned 7.07% annualised with a maximum drawdown of -62.42%, against 8.80% and -62.07% for gold alone. The reason is mechanical: the two metals correlate 0.5695 in monthly returns, so platinum added little diversification while adding its higher volatility and lower return.
These portfolios are illustrative. They are monthly-rebalanced, cost free, and model no spread, financing, storage or transaction cost. Real platinum exposure costs more to hold than gold, so these figures flatter platinum.
Buy and hold since 1971
Gold has won on return, on drawdown and on volatility over this period. Platinum's 12.6× versus gold's 108.0× is the whole platinum-vs-gold debate in two numbers. These are historical figures on monthly average prices, gross of storage, financing and transaction costs.
What this dataset does and does not support
Supported by the data: the ratio's level has risen structurally since the 1960s, so any fixed threshold is era-dependent; platinum's volatility premium over gold currently runs about 1.9×; the extreme high tail of the ratio does show clear mean reversion in platinum itself; and on the full historical sample a high ratio has been associated with weaker forward gold returns.
Not supported: the "platinum is cheap, so buy platinum" reading. In the modern regime the forward-return difference between high- and low-ratio months is -3.72pp, with a non-overlapping p = 0.093 and a bootstrap interval of -11.4 to 4.9pp that includes zero. The much larger full-sample figure of -12.91pp comes from the 1970s and does not carry into the regime the current reading sits in. The industrial-cycle test finds no meaningful co-movement with output in either metal, and the lead-lag test finds nothing outside the noise floor set by the control series.
A ratio can be a useful descriptive label for how precious metals are priced against each other. That is not the same as it being a tradeable signal, and this dataset is offered as the evidence for the difference.
Frequently asked questions
What is the gold/platinum ratio and where is it now?
The gold/platinum ratio is how many ounces of platinum one ounce of gold buys. As of 2026-10 it stands at 2.5, against a median of 0.8 over the 802-month sample from 1960-01 to 2026-10. That places it at the 98.4th percentile of its own history, and above the 1.6 90th-percentile line. The sample range runs from 0.3 in 1969-12 to 3.4 in 2025-05.
Does a high gold/platinum ratio mean platinum is cheap and should I buy it?
The ratio is high, but the "platinum is cheap" reading does not survive a proper test. Split the post-1981 sample at its median ratio of 0.9 and the high-ratio half's 12-month forward gold return averages 4.7% against 8.4% in the low half, a difference of -3.72 pp. On the non-overlapping subsample of 45 windows the difference is -9.22 pp with p = 0.093, which is not significant, and the moving-block bootstrap interval runs from -11.4 to 4.9 percentage points and includes zero. On the full 1960 sample the same test gives a much larger -12.91 pp, but that number comes from the 1970s inflation era, not from the regime you are trading in today.
How much does the answer depend on which era you start from?
A lot, and that is the main finding. The identical test gives -12.91 pp on the full 1960 sample, -16.15 pp after 1971, -3.72 pp after 1981, and -3.84 pp after 1991. The non-overlapping p-values run 0.045, 0.051, 0.093 and 0.211. The effect is therefore mostly a property of the 1970s, when gold was repricing from a pegged price while platinum moved with the industrial cycle. In the modern regime it is not statistically distinguishable from zero.
Is platinum really an "industrial" metal and gold a "monetary" one?
Not according to this data, at least not in the way the story is usually told. Platinum's monthly return correlates 0.0663 with the change in US industrial production and gold's correlates -0.0409 — and for calibration, the industrial index correlates 1 with itself. Both metals are close to uncorrelated with the industrial cycle on monthly data. The industrial story is a real demand story, but it does not show up as a measurable monthly co-movement with output in either metal.
Does one of these metals move first?
No lead-lag structure survives the test. Cross-correlating gold's monthly return against platinum's at lags of −6 to +6 months, the strongest reading is 0.5695 at lag 0, with 0.5695 at lag 0. For calibration, correlating gold's return against its own one-month-shifted self produces 1 at the best lag, so anything in that neighbourhood is indistinguishable from noise.
How related are gold and platinum really?
Their monthly returns correlate 0.5695 over 801 months, and their price levels correlate 0.7775. The level correlation is mostly a trend artefact — two series that both go up over 66 years will correlate in levels. The rolling 36-month return correlation has a median of 0.572 but drops negative in 8.8% of windows, reaching -0.5539 at its lowest. So the two metals are related most of the time and occasionally behave like unrelated assets.
Is the gold/platinum ratio a "signal" or just an arithmetic identity?
Mostly an identity. The change in the ratio is, to first order, gold's return minus platinum's return, so the correlation between the ratio change and the gold return (0.3019) is a restatement of which leg moved more, not a discovered relationship. What the ratio does add is a level, and levels are regime-dependent: the ratio averaged 0.4 in the 1960s and 0.6 in the 2000s, so any fixed threshold like "above 80 means platinum is cheap" changes meaning every decade.
What happened after the most extreme ratio readings in the modern era?
Taking the 20 months with the highest ratios since 1981 (2.3 to 3.4), gold averaged 42.1% over the next year and platinum 55.1%, with the ratio itself down 3.2%. At the low end (0.4 to 0.5), gold averaged 7.5% while platinum averaged -18.3% and the ratio rose 36.9%. Platinum rose in only 15% of the low-ratio months, against 95% of the high-ratio months. The mechanical mean reversion of the ratio is real; whether it pays depends entirely on which leg you were holding.
Has platinum outperformed gold more or less often?
Less often. Over the 66 calendar years in the sample, platinum beat gold in 26 of them, or 39.4%.
Which metal has the better long-run return, and how risky is it?
Since 1971, gold returned 8.75% annualised with a maximum drawdown of -62.07% and an annualised volatility of 17.1%. Platinum returned 4.64% annualised with a maximum drawdown of -70.41% and volatility of 20.7%. Gold has won on return, on drawdown and on risk-adjusted return over this period. Those are historical figures on monthly average prices, gross of storage, financing and transaction costs, and they are not a forecast.
How volatile is platinum relative to gold right now?
On a trailing 60-month basis, platinum's annualised volatility is currently 1.89 times gold's, with gold at 14% and platinum at 26.3%. The median across the sample is 1.33 times. Sizing a platinum position at the same notional as a gold position therefore runs roughly 1.9 times the risk, which is what the position size calculator on the tools page is for.
Does adding platinum to a gold position help?
On these numbers, not really. A monthly-rebalanced, cost-free blend at 50% gold and 50% platinum returned 7.07% annualised since 1971, against 8.8% for all gold and 4.58% for all platinum, with a maximum drawdown of -62.42% against -62.07% for gold alone. Adding platinum lowered the return in every weight tested, because the two metals correlated 0.5695 and platinum was the higher-volatility leg with the lower return. These are illustrative portfolios with no transaction, spread, financing or storage costs modelled.
Where are gold and platinum right now?
In 2026-10, gold is at $4105/oz and platinum at $1621/oz, giving a ratio of 2.5 — the 98.4th percentile of the 802-month sample. Over the last twelve months gold is 1.2% and platinum 0.3%. The ratio was 1.8 sixty months ago. A high ratio is a description of relative pricing, not a reason to expect platinum to catch up.
Sources
- World Bank Commodity Price Data (The Pink Sheet),
CMO-Historical-Data-Monthly.xlsx— Monthly Prices sheet, columns 70 (Gold) and 71 (Platinum), updated November 03, 2026. https://www.worldbank.org/en/research/commodity-markets - FRED series INDPRO, US industrial production index (2017 = 100) — monthly US industrial production index, used as the industrial-cycle proxy.
-
Analysis script:
scripts/gold_platinum_analysis.py(in this project's source tree). Independent recomputation with a separate numpy implementation:scripts/verify_gold_platinum.py. - The other cross-metal comparisons: Gold versus silver (the other precious metal) and Gold versus Bitcoin (neither precious metal nor diversifier).
- For the correlation and level-correlation caveats discussed here, see Gold vs. real yields and the dollar and Gold price history.
Data accessed 2026-11-13. Prices are nominal US dollars and include no carry, financing or storage cost. Nothing on this page is investment advice or a forecast.