A doubling, a blow-off, and a lesson in why

Gold spent two years doing exactly what its believers said it would, and then taught them the harder half of the lesson. From about $2,062 an ounce at the end of 2023 it ran to a record of roughly $5,318 in late January 2026, a gain near 158 percent, before falling about 25 percent to near $3,990 by late June. It sits around $4,152 today. The people who held gold for the simplest reason, that fiat money is being debased and gold is the hedge, were handed a rally their framework did not really predict and a correction it did not protect them from. That is the paradox at the centre of this note: the price proved its owners right and their reasoning wrong. The move was real and large. The story attached to it was mostly wrong.

Where this departs from the easy read

Nothing in this note is a price forecast. The figures are the market's own and the official record: the price of gold, real and nominal yields, the dollar, the Treasury curve, the tonnage central banks report to the World Gold Council, published policy rates, and the inflation and fiscal data. What a research note can add is discipline, to separate what happened from the story told about it, and to say plainly which of gold's traditional drivers were present and which were not.

The case for gold, and why it is fragile

There is a school of investors, old and internally coherent, that holds gold not as one asset among many but as an article of faith: gold is real money, the dollar is a promise, and over time the promise is broken. In that worldview gold is the hedge against inflation and the insurance against the debasement of paper currency. Parts of it are true over long enough spans. The problem is the horizon.

Gold is an inflation hedge only over centuries

The most careful work on this is Erb and Harvey's The Golden Dilemma. Their finding is that gold does hold its purchasing power, but only over spans measured in centuries. A Roman legionary and a modern soldier could each be paid roughly an ounce of gold. Over the horizons real investors actually hold it, years to a decade, the real, inflation-adjusted price of gold is volatile and tends to revert. When gold is expensive relative to consumer prices, subsequent real returns have tended to be low. When it is cheap, they have tended to be high. That is close to the opposite of a dependable short-run inflation hedge.

The driver that actually shows up in the data is real rates

If not inflation, then what. The relationship with the longest pedigree is with real interest rates. Barsky and Summers, working through Gibson's Paradox, showed that the price level under a gold standard is the reciprocal of the real price of gold, and that gold's relative price is driven by the same real returns on capital that set real interest rates. In modern terms, gold pays no coupon, so the real yield on safe bonds is its opportunity cost. When real yields fall, gold tends to rise, and when they rise, gold tends to fall. This is the compass a market professional actually watches. For most of the last two years it pointed the wrong way, and gold ignored it. The rest of this note is about why.

How gold has actually behaved across the decades

Before reading the last two years it helps to read the last fifty. Gold is spoken of as though it only rises. Its actual record is a sequence of long cycles, with advances that run for years and declines that are just as long and just as deep. When the United States ended the dollar's convertibility into gold in 1971, the metal was fixed near 35 dollars an ounce. It rose through the inflation and oil shocks of that decade to a peak around 850 dollars in January 1980, then fell into a bear market that lasted two decades and bottomed near 255 dollars in 2001, where the daily series in the chart below begins. An investor who bought the 1980 top waited until 2008, twenty-eight years, merely to see that price again in nominal terms, and far longer after inflation. That is the first fact the debasement story omits: gold can spend a generation going nowhere.

From that low the modern record splits into two great advances separated by a punishing decline. Gold rose roughly sevenfold from 2001 to a peak near 1,889 dollars in 2011, carried by a weak dollar, the financial crisis and the same debasement conviction that surrounds it today. It then fell about 44 percent over the next four years, to near 1,051 dollars at the end of 2015, and did not reclaim its 2011 high for the better part of a decade, only passing it again in 2020. The second advance, from 2019, is the one that ran to the 2026 record.

The sterner test is in real terms, because inflation flatters every long nominal chart. Measured in constant 2026 dollars, the 2011 peak was worth about 2,790 dollars and the 2015 trough about 1,476, a real decline near 47 percent. On that measure gold set no new high at all between 2011 and early 2026. The January 2026 record, close to 5,300 dollars, was the first time this century that gold decisively exceeded its 2011 level in real, inflation-adjusted terms. This is the empirical shape of the golden dilemma named in the previous section: the real price of gold is volatile and slow to make new ground, and the investor who pays a high real price is, more often than not, buying weak subsequent returns.

None of this argues against owning gold. Over the twenty-five years the chart covers, gold compounded at roughly 11 percent a year before inflation and about 8 percent after it, a strong result for a real asset that pays no coupon. But the return was neither smooth nor evenly delivered. It arrived in two multi-year bursts around a drawdown deep enough and long enough to break most conviction. That uneven path is precisely why gold earns its place as a diversifier rather than a core holding: its low, and at times negative, correlation with stocks and bonds is valuable because its own course is so independent of theirs. The research on gold as a safe haven finds the same thing, that gold tends to cushion equity stress in developed markets, but the cushion is a portfolio effect and not a promise that gold itself will not fall, as it did by about 30 percent inside the 2008 crisis before recovering.

From two thousand dollars to five, and part of the way back

Start with the path itself, because the numbers are large enough to be worth stating plainly. Gold ended 2023 near $2,062 an ounce. It rose through 2024 and 2025 with few meaningful pauses, and peaked at about $5,318 in late January 2026, a gain of roughly 158 percent in two years. It then fell about 25 percent to near $3,990 by late June, and trades around $4,152 today, still about 101 percent above where it began.

This is the stretch of chart that produced the headlines and the certainty. A vertical advance invites a simple explanation, and the debasement story was the one already loaded. But a chart that goes straight up is not evidence for any particular reason. It is, more often, evidence that the move has outrun whatever started it. The next section asks what the price was actually reacting to, day by day.

A timeline of the news that moved gold

One way to test the debasement story is to line the big moves up against the news that produced them. If gold were tracking a slow erosion of paper money, its largest single days would cluster around inflation data. They did not. They clustered around the price of money and around politics: tariffs, threats to the independence of the Federal Reserve, the return of rate cuts, and episodes of war.

Read down the right-hand column and the debasement clock is nowhere. The metal traded on real and expected interest rates, on the credibility of the institution that sets them, and on geopolitical risk. Those are the drivers the next sections measure one at a time, starting with the crowd that is usually blamed for a blow-off, and was, this time, largely absent.

How far price ran ahead of its own trend

There is a trader's proverb that the bigger the base, the higher in space, invoked to justify almost any advance after the fact. It is not a measurement. A measurement is how far price has travelled above its own long-run trend, and by that gauge the January 2026 peak was not a normal high. It was a blow-off.

This is what a sentiment reset looks like before it happens. When an asset trades 90 percent above the average price of the last four years, the marginal buyer is not hedging inflation or diversifying reserves. The marginal buyer is buying because the price is rising. That is the definition of a momentum crowd. The interesting question, and the subject of the next section, is where that crowd actually sat, because it was not where a market professional would first look.

Positioning says the crowd was in funds, not futures

When gold goes vertical, the reflex is to blame leveraged speculators in the futures market. The data does not support it here. The Commitment of Traders report, which the regulator compiles every week, shows that large speculators held a middling net long position at the price peak, well below their 2020 extreme and below even their end-2023 level.

If not futures, then where. The answer is Western exchange traded funds. After three years of outflows, gold funds took in a record sum in 2025, by the World Gold Council's tally the strongest year on record, and their holdings reached an all-time high. North America alone was well over half of that. Then, as the price broke, those same investors reversed: March 2026 was the largest monthly outflow from North American gold funds on record, even as central banks and Asian buyers kept adding. A monthly fund-manager survey made the same point from another angle, showing long gold as the single most crowded trade in the market in January 2026 before it unwound.

Gold rose while its traditional headwinds did too

Return to the compass. If gold trades off real rates and the dollar, then the last two years should not have happened. They did, and the cleanest way to see the anomaly is to put gold next to the 10-year Treasury yield.

The stronger version of the point uses real yields, since those, not nominal, are gold's true opportunity cost. They make the anomaly sharper, not softer. The inflation-protected 10-year yield was about 1.8 percent at the end of 2023 and, by the reading of asset managers who track it, stayed near 2 percent right through gold's record, having been deeply negative during the 2020 rally. A real yield near 2 percent is historically a serious headwind for a zero-coupon asset. Gold doubled against it. As Janus Henderson put it, real yields still set direction over short horizons but no longer set the floor.

The dollar tells the same story with a small qualification. It softened during the run, from about 101 on the dollar index at the end of 2023 to near 96 at the gold peak, which helps a little. But it has since recovered to about 101, back to where it started, while gold has held most of its gains. A five percent move in the dollar cannot explain a rise of this size in gold. And the usual inverse links loosened: the correlation of weekly gold and dollar moves eased from about -0.49 before 2023 to about -0.32 since, and the link with changes in the 10-year yield faded toward zero, from about -0.17 to about -0.07. The compass did not reverse. It stopped pointing. Something outside the standard model was setting the price.

Three central banks cut, and the opportunity cost fell

Part of that something is the most conventional force there is: monetary policy. Gold pays no coupon, so the return available on cash and short bonds is what an investor gives up to hold it. From 2024 the three central banks that matter most for global liquidity all lowered that return at once.

This matters twice. On the way up, falling policy rates were a genuine, measurable tailwind that has nothing to do with debasement: the cost of holding gold fell, and the metal rose, exactly as the real-rate framework predicts. On the way down, the same lever reversed. The single sharpest fall of the cycle came when a new Fed chair was nominated in late January 2026 and the market began to price a slower, more political path for rates. By mid-2026 the Fed was on hold, and the European Central Bank had actually raised rates again, its first hike in three years, as an energy-driven inflation scare returned. Gold's correction is, in large part, the market taking back the rate cuts it had assumed.

The one thing its owners were sure of was falling

Now the central claim, and the one most easily checked. Gold's traditional owners hold the metal because paper money is losing value. For that story to explain this rally, inflation should have been rising while gold rose. It was doing the opposite.

The sequence is fatal to the simple story. Gold more than doubled while measured inflation was cooling toward target, in the United States and in the euro area, where it fell from 10.6 percent to 2 percent, and while China flirted with outright deflation. Then, when an energy shock from the Middle East finally pushed headline inflation back up in the spring of 2026, gold was already in a 25 percent correction. If inflation were the hand on the tiller, these moves would run the other way. The metal was responding to something inflation does not capture.

The official sector, China, and diversification

That something begins with the official sector. Central banks have always held gold, but for most of the post-war era they were net sellers or marginal buyers. That changed after 2022, when the freezing of Russian dollar reserves gave every reserve manager a reason to hold an asset that no other government can freeze. The result is the missing variable in the retail debasement model: a large, strategic and price-insensitive buyer.

This buyer behaves nothing like the retail hard-money owner. It does not chase momentum and it does not sell on a pullback. It buys tonnage against a reserve-management mandate, often more when prices dip, and it reports with a lag, if at all. The World Gold Council estimates that well over half of 2025 official demand went unreported. China is the clearest case: its reported reserves rose for twenty consecutive months to about 2,346 tonnes by mid-2026, the longest run on record, and some analysts believe the true pace of its buying is a multiple of the figure it discloses. Poland was the largest single buyer of the period, taking gold past a quarter of its reserves. Its governor put the logic plainly: gold does not represent anyone else's liability.

A shift you can see in Chinese households, too

The same preference shows up below the central bank. In China in 2025, investment demand for bars and coins overtook jewellery for the first time in the World Gold Council's data. Jewellery, which is price-sensitive, was destroyed by the high price and fell sharply. Investment, which is not, rose to a record. Households were buying gold as a store of value, not as adornment, and they did not flinch at the level. That is the behaviour of insurance buyers, and it is the opposite of the momentum crowd in the Western funds.

When the risk-free asset stops looking risk-free

There is one more structural force, and it connects the others. Gold competes with the government bond as the reserve asset of last resort. Over this period the case for the bond weakened in ways that are a matter of record, not opinion.

The numbers behind that long end are large. In May 2025 Moody's cut the United States from its top rating for the first time since 1917, the last of the three major agencies to do so. Net interest on the federal debt passed a trillion dollars for the first time, roughly three times its 2020 level, and now runs ahead of the defence budget. Federal debt held by the public approached 100 percent of output, a level not seen since just after the Second World War, with deficits near 6 percent of output in an economy that was not in recession. China, long a mainstay buyer, has cut its Treasury holdings by roughly a third over a decade.

None of this is a forecast of crisis. Foreign demand for Treasuries overall is still at record levels, and the dollar remains the reserve currency by a wide margin. The point is narrower and sufficient: the asset that gold competes with, the long-dated claim on the United States government, now carries a visible fiscal question mark and a positive term premium to match. When the risk-free asset looks a little less risk-free, a scarce asset with no coupon and no counterparty earns a structural bid. That, and not a debasement clock, is what the world's reserve managers have been acting on.

Sharp, but well within gold's history

The fall from the peak has been called violent, and from a late buyer's seat it is. A drop of about 25 percent in five months is real money. But read against gold's own record it is ordinary. Gold is a volatile asset that has repeatedly given back a quarter or more of its value, and the current decline is not yet the largest of the century.

The mechanics of this correction fit everything above. It began not with an inflation surprise but with a Fed-chair nomination and the repricing of rate cuts, deepened as the dollar firmed and an energy shock pushed real yields up, and was cushioned because the structural buyer never left. Western funds sold, the official sector did not. The US Treasury Secretary himself described the episode as a classical speculative blow-off, pointing to disorderly trading in China and to tightened margin requirements, which is a fair description of a positioning unwind and a poor fit for a debasement story.

The relevant precedent is 2011. Gold made a blow-off high near 1,900 dollars that autumn on a wave of debasement conviction after the US credit-rating downgrade, then fell more than 40 percent and did not see that level again for nine years. The investors most hurt were not the sceptics. They were the believers who bought the top because the story was loudest there. That is how gold disappoints its own believers: the narrative is most persuasive exactly when the price is most extended.

The commentary that prompted this note reads the same correction as a bear trap before a later push toward much higher targets, a reset of sentiment after what its author calls a media-mania blow-off. We take no view on the targets. Forecasting the next move is a different exercise from understanding the last one, and this note has tried to do the second. Whether the advance resumes depends mostly on whether the official-sector bid persists and whether real yields finally fall, and neither of those is settled by a price chart.

The golden dilemma, restated

Gold did not fail. It did roughly what a scarce, monetary, zero-yield reserve asset does when the largest and least price-sensitive buyers in the world decide they want more of it, and when the bond it competes with loses a little of its shine. What failed was the story its retail owners told about why. They were long for inflation and debasement. They were rewarded by central-bank reserve policy, by the return of monetary easing, by an eroding fiscal anchor, and by a Western fund crowd, forces their framework does not contain. When those forces paused, the framework offered no exit, and the believers held the top.

A story that cannot be wrong cannot manage risk.

The lesson is not to avoid gold. Gold earns a place in a serious portfolio for what it measurably is: a real asset with low correlation to stocks and bonds, no credit risk, and a genuine structural bid from the official sector. The lesson is to hold it for those reasons, at a size chosen in advance, and to resist the temptation to treat a rising price as confirmation of a prophecy. Own gold as an asset. Do not worship it as a verdict.

Selected references and further reading

The reading behind the argument, on gold as an inflation hedge, the role of real rates, official-sector demand, reserve composition, and the fiscal backdrop.

Data as of 7 July 2026. Price and yield series from public market data; positioning from the CFTC; tonnage from the World Gold Council; policy rates from the Federal Reserve, ECB and PBOC; inflation from the US Bureau of Labor Statistics; fiscal figures from the US Treasury, CBO and Moody's. Figures are rounded for readability.