Over five centuries, gold has not been one investment. It has been three, and they behave nothing like each other. For 214 years gold had no price at all: it was the ruler against which prices were measured. For roughly fifty years after 1971 it was a leveraged bet on real interest rates. Since 2022 it has traded increasingly as a reserve asset with no counterparty. Gold price history only becomes useful once you separate those three jobs, because almost every argument you will hear for or against owning gold is quietly using evidence from one era to make a claim about another.
At the time of writing, gold trades near US$4,350 an ounce. That is roughly 22% below the record of about US$5,589 set on 28 January 2026, and roughly 10% above the June low near US$3,940. Both of those facts are true. Which one you find more interesting says more about your framework than about gold.
Gold price history begins with a number that never moved
In 1717, Sir Isaac Newton, then Master of the Royal Mint, set the mint price of gold at £3 17s 10½d per standard ounce. He was not trying to create a gold standard. He was trying to stop silver leaking out of England to Asia, and he got the ratio slightly wrong in gold’s favour.
That number then held, with a suspension during the Napoleonic wars, until Britain left the gold standard in 1931. Two hundred and fourteen years. The United States did something similar, fixing gold at $20.67 an ounce from 1834 until 1933, then at $35 from 1934 until 1971.
This matters more than it sounds. For most of the period people invoke when they say “gold has held its value for centuries”, gold’s price was a legal fact rather than a market outcome. It could not go up. What moved was everything else.
This is why the most quoted statistic in the gold debate is close to meaningless without context. Jeremy Siegel’s long-run data show gold returning roughly 0.6% a year after inflation from 1802 to 2021, against about 6.7% for equities. Gold bears cite it constantly. But researchers reviewing that dataset for the CFA Institute pointed out the obvious problem: for the first 130 years of that sample, gold and the US dollar delivered essentially identical returns, because gold was the dollar. You are not measuring an investment. You are measuring a definition.
The three jobs of gold
A more honest way to read the record is to treat gold as having held three distinct jobs, each with its own logic, its own drivers and its own failure mode.
Job one: the ruler (roughly 1560 to 1971)
Here gold was the measuring stick. The serious work on this period is Roy Jastram’s The Golden Constant, which tracked gold against wholesale prices in England from 1560 and in the United States from 1800. His conclusion was that gold’s purchasing power fluctuated, sometimes violently, but oscillated around a broadly flat line across four centuries.
Jastram also found two things that most gold advocates skip past. Gold was a poor hedge against major inflation. And gold’s purchasing power rose during deflation. That is the opposite of the modern sales pitch, and it makes sense once you remember the price was pegged: when prices collapsed, a fixed-price ounce bought more.
Job two: the anti-real-rate trade (1971 to roughly 2022)
When Nixon closed the gold window in August 1971, gold became a freely traded financial asset for the first time in modern history. It then behaved with remarkable consistency for fifty years, driven by one variable above all others: the real interest rate, meaning the interest you can earn after inflation.
Gold pays no income. When cash and bonds pay you well after inflation, holding gold costs you something. When they pay you badly, it costs you nothing. That single mechanism explains the 1970s surge, the twenty-year bear market that followed, the 2001 to 2011 bull run and the post-2013 slump.
Job three: the counterparty-free reserve (2022 to now)
Something changed after Western governments froze Russia’s dollar reserves in 2022. Gold began rising even while real interest rates were positive and rising, which the old model said should not happen. What central banks were buying was not an inflation hedge. It was an asset that cannot be frozen, sanctioned or defaulted on by another government.
The scale of that shift is now measurable. According to the European Central Bank’s June 2026 report, gold accounted for 27% of global central bank reserve assets at the end of 2025, up from 20% a year earlier, while US Treasuries fell to 22% from 25%. Central banks collectively hold more than 36,000 tonnes, approaching the roughly 38,000 tonnes held during the Bretton Woods era. Dollar-denominated assets still dominate overall at 42% of reserves, so this is not the end of the dollar. It is the partial displacement of one specific instrument, the US Treasury bond, by an asset that answers to nobody.
Every gold peak, and what broke it
The pattern across the record is uncomfortably consistent. Gold peaks when a monetary or geopolitical fear becomes consensus, and it falls when that fear is either resolved or priced out by higher real rates.
| Peak | What drove it | What followed | Time to recover |
|---|---|---|---|
| 1717 to 1931 £3 17s 10½d |
Legal peg, not a market | No nominal change for 214 years | Not applicable |
| 1934 $20.67 to $35 |
Depression, Executive Order 6102 forced US citizens to surrender gold | Holders sold at $20.67, then watched it revalued to $35 nine months later, roughly 69% higher | Private ownership stayed illegal for 41 years |
| Jan 1980 $850 |
Double-digit inflation, Iranian revolution, Soviet invasion of Afghanistan, Hunt brothers’ silver squeeze | Fell roughly 70% to about $253 by 1999 as Volcker pushed rates to 20% | 28 years to regain the nominal high. Around 45 years in inflation-adjusted terms |
| Sep 2011 $1,921 |
Global financial crisis aftermath, quantitative easing, US debt-ceiling standoff | Fell about 45% to roughly $1,050 by December 2015 | Roughly 9 years |
| Aug 2020 $2,067 |
Pandemic, zero rates, emergency fiscal spending | Drifted lower into 2022 as rates rose | Roughly 3 years |
| Jan 2026 ~$5,589 |
2025 gain of about 64%, best year since 1979. Record central bank buying, ETF inflows, de-dollarisation | Fell roughly 28% to near $3,940 by June 2026 | Unresolved |
Two entries deserve more attention than they usually get.
1934 is the risk nobody prices. American holders did not lose money because the market fell. They lost because ownership was made illegal, they were compelled to sell at the old official price, and the government then revalued gold upward. Gold protected the sovereign. It did not protect the citizen. Any argument that gold is beyond the reach of governments has to explain that year.
1980 is the risk everyone underestimates. Someone who bought at the January 1980 top waited nearly three decades to see their money back in nominal terms, and around four and a half decades to break even after inflation. That is not a drawdown. That is a career.
So why does gold have value at all?
Strip away the mysticism and there are four reasons, none of which involve gold being intrinsically special.
- It is chemically inert and effectively indestructible. Almost every ounce ever mined still exists. Gold does not rust, tarnish or decay, so a coin buried for two thousand years comes out unchanged.
- Its supply is close to fixed. Mine output adds only a small percentage to the existing above-ground stock each year, and recycling responded weakly even to a 67% price rise in 2025. Nobody can decide to create more of it.
- It is nobody’s liability. A bond is a promise. A deposit is a promise. Gold is not a promise, which is precisely why sanctioned or nervous central banks want it.
- Enough people agree it has value. This is circular, and that is the honest answer. Gold’s value rests on an unusually durable and geographically universal social convention, tested over roughly five thousand years. Durability is not the same as permanence, but it is not nothing either.
Notice what is absent from that list: cash flow. Gold produces nothing. Its entire return depends on someone paying more for it later. That is not a moral failing, but it does mean gold cannot be valued the way a business or a bond can. There is no fair value to anchor to, which is why forecasts for it disperse so wildly.
What the 2026 drawdown actually told us
This year has been an unusually clean natural experiment, and the result is not what either camp wanted.
Gold peaked near $5,589 in late January 2026 after a 2025 in which it gained roughly 64%, set 53 record highs and outperformed every major equity index. It then fell 9.8% in a single day on 30 January, its worst day since 2013, as markets repriced the expected direction of US monetary policy under incoming Federal Reserve Chair Kevin Warsh. The decline extended through a hawkish first half. Counter-intuitively, the escalation of the Iran conflict made things worse for gold rather than better: higher oil prices lifted inflation expectations, which pushed rate-cut expectations out and rate-hike odds up.
Here is the part worth sitting with. If job three had genuinely replaced job two, gold should have shrugged that off. Reserve managers buying sanction-proof assets do not care about the Fed’s next move. Instead gold fell 28%. The real-rate sensitivity of the old regime is alive and well.
This is a Tigris interpretation rather than an established fact: gold is currently running on two engines at once, a structural reserve bid and a cyclical rate trade, and those engines are not correlated. That combination produces a higher floor than the 1980s and 1990s allowed, and also more violence around it. Investors who bought the reserve-asset story and received a rate-trade drawdown were not wrong about the thesis. They were wrong about which engine was driving on any given day.
The floor that is not a floor
The most common structural argument for gold today is that central banks provide a permanent bid. It is the weakest part of the bull case, and the ECB’s own data shows why.
Turkey accumulated roughly 220 tonnes of gold after Russia’s 2022 invasion of Ukraine. Then, in early 2026, following the outbreak of the Iran war, Ankara sold or loaned approximately 130 tonnes. That was one of the largest reserve drawdowns in recent years, and it arrived at precisely the moment gold was already falling.
Call it the insurance paradox. Central banks buy gold as insurance against a crisis. When the crisis actually arrives, some of them need liquidity, so they sell the insurance. Official-sector demand is therefore a shock absorber, not a floor, and it is thinnest exactly when you are relying on it most. Net purchases have already eased from a record 1,082 tonnes in 2022 to 850 tonnes in 2025.
Worth noting as a marker of how strange this cycle has become: the ECB reported that stablecoin issuer Tether was the single largest gold buyer of 2025, acquiring more than 100 tonnes. Even the ECB, documenting gold’s rise, listed its drawbacks plainly: volatility, no interest, storage costs and inelastic supply.
So should you invest in gold?
That question cannot be answered in the abstract, and anyone who answers it confidently is selling something. What the historical record does support is a better question: which job are you hiring gold to do, and is it currently priced to do that job?
The framework we find useful:
- If you want insurance against monetary or geopolitical breakdown, gold has a genuine five-century record. But insurance has a premium, and after a 64% year the premium is not cheap. You are buying protection at an elevated price, which is when protection historically works least well.
- If you want an inflation hedge, the evidence is weaker than the marketing. Jastram found gold a poor hedge against major inflation across four centuries. It failed badly through the 1980s and 1990s while inflation was still positive. It works against regime change, not against CPI.
- If you want returns, gold has no cash flow, no fair value and a two-century real return that trails equities by an enormous margin. It has beaten equities over some decades and lost catastrophically over others.
- If you cannot articulate which of the three you are buying, that is the finding, not a gap in the analysis.
Position sizing is investor-specific and depends on liabilities, horizon and what else you own, which is why we do not publish a number. What we would insist on is that gold is sized as a portfolio function rather than a view, because a view can be wrong for 28 years and still be right eventually. Most investors cannot wait that long.
Fact, interpretation and forecast
| Type | Statement |
|---|---|
| Verified fact | Gold peaked near US$5,589 on 28 January 2026, fell roughly 28% to near US$3,940 by June, and trades near US$4,350 in early August 2026. |
| Verified fact | ECB data show gold at 27% of global central bank reserve assets at end-2025, ahead of US Treasuries at 22%. Dollar assets remain 42% overall. |
| Verified fact | Gold’s sterling price was fixed at £3 17s 10½d from 1717 to 1931, and the US price at $35 from 1934 to 1971. |
| Tigris interpretation | Gold is currently priced by two uncorrelated engines, a structural reserve bid and a cyclical real-rate trade. This raises the long-run floor and increases short-run volatility simultaneously. |
| Tigris interpretation | Central bank demand is a shock absorber rather than a floor. Turkey’s 2026 disposal of roughly 130 tonnes during a crisis is the template, not the exception. |
| Falsifiable forecast | If the reserve-asset regime is genuinely structural, the next sharp rise in real yields should produce a materially smaller drawdown than the 28% seen in the first half of 2026. If it produces an equal or larger fall, the reserve story is a narrative layered on an old-fashioned rate trade rather than a replacement for it. |
| Disputed | Year-end 2026 sell-side targets have ranged from roughly US$4,500 to US$6,300 across major houses and forecast vintages. The dispersion itself is the signal: gold has no cash flow to anchor a valuation. |
Frequently asked questions
Is gold a good hedge against inflation?
Less reliably than commonly assumed. Roy Jastram’s four-century study found gold to be a poor hedge against major inflation, and gold lost value in real terms through the 1980s and 1990s while inflation remained positive. Gold performs best when real interest rates fall or when confidence in the monetary system itself is questioned, which is a hedge against regime change rather than against rising prices.
What is the highest gold has ever been?
In nominal terms, about US$5,589 an ounce on 28 January 2026. In inflation-adjusted terms the January 1980 peak of $850 stood as the record for roughly 45 years, equivalent to somewhere between US$3,300 and US$3,600 in today’s money depending on the price index used.
Why did gold fall in 2026 if central banks are still buying?
Because official-sector buying and the interest-rate cycle are separate forces. A hawkish shift in expected US monetary policy raised the opportunity cost of holding a non-yielding asset, and higher oil prices from the Iran conflict lifted inflation expectations, which pushed rate cuts further out. Turkey also sold or loaned roughly 130 tonnes of reserves during the same period.
Can governments confiscate gold again?
It has happened. Executive Order 6102 in 1933 required Americans to surrender gold at $20.67 an ounce, after which the official price was reset to $35, and private ownership remained illegal for 41 years. Modern legal frameworks differ, but the historical precedent is real and rarely appears in gold marketing material.
How much of a portfolio should be in gold?
There is no universal answer, and any specific figure should be treated with suspicion. The allocation depends on your liabilities, time horizon, currency exposure and what the rest of the portfolio already does. The more useful discipline is to define in advance which job gold is doing for you and what evidence would tell you it has stopped doing it.
The question worth holding
Gold’s five-century record does not tell you to buy it, and it does not tell you to avoid it. It tells you something less convenient. Gold is a claim on the failure of other arrangements, and it is priced by how likely people currently think that failure is. That price has been wrong in both directions, for decades at a time.
The market currently believes something structural has changed in how the world stores official wealth. The ECB’s data suggests it is right. The first half of 2026 suggests the market is still capable of repricing that belief by 28% in five months when a central banker changes tone. Both of those can be true at once, and holding both is harder than picking a side.
For more on how we think about macro regime shifts and asset positioning, see the Tigris Insights hub.
Sources: European Central Bank reserve report (June 2026); World Gold Council, Gold Demand Trends full-year 2025; Roy Jastram and Jill Leyland, The Golden Constant (1560 to 2007); Jeremy Siegel, Stocks for the Long Run; McQuarrie and McCaffrey, “Stocks for the Long Run? New Evidence, Old Debates” (CFA Institute Research Foundation); Royal Mint and Bank of England historical records; US Federal Reserve historical analysis of the Gold Reserve Act 1934; LBMA Alchemist; J.P. Morgan Private Bank.
This material is for informational purposes only and does not constitute an offer or solicitation to buy or sell any investment product. Past performance is not indicative of future results. Investment products are available only to accredited and institutional investors as defined under the Securities and Futures Act 2001 of Singapore. Tigris Asset Management Pte Ltd holds Capital Markets Services Licence CMS101520 issued by the Monetary Authority of Singapore.
