Everyone's blaming central banks for gold's slide. The number behind that blame was just slashed by 80%.

Gold did something in 2026 that its own fans rarely have to face: it set a record above $5,000 an ounce in January, then gave back roughly a fifth of it by late spring. Tamer inflation has steadied it since — it hangs near $4,400 — but the question that replaced "how much higher" is now "is it over?"

The bears have an answer, and for once it is uncomfortably specific. The engine is cooling. Central banks, the buyers credited with turning gold into a must-own reserve asset, stopped buying as much. The same institutions that were right about $5,000 now publish forecast resets and headlines about slashed purchases.

There is just one problem with that answer. The number it rests on was corrected by nearly 80% two months ago.

The number everyone blames was rewritten

Start with the story that made gold credible. Since 2022, when Western sanctions froze Russia's dollar assets, central banks became the decisive buyer. Over the past four years they accumulated about 1,000 tonnes a year — double the 500-tonne pace of the prior decade. That structural demand — call it de-dollarization — is the reason the bull market had a floor beneath it, and the reason analysts felt safe forecasting $5,000 and beyond.

In 2026 that story wobbled. For the first quarter, official purchases were reported at a still-impressive 244 tonnes. But when the World Gold Council published its second-quarter demand report, it revised that first-quarter figure down to about 57 tonnes — the weakest first quarter in more than fifteen years, and part of the lowest half-year of official buying since 2022.

So the engine that supposedly powered the rally was never turning as fast as the headlines said. Part of the discrepancy is bookkeeping: some gold previously counted as official buying was reclassified into an "OTC and other" bucket. Part reflects that, since the 2022 sanctions, many central banks stopped disclosing their purchases fully, so the World Gold Council reconstructs the numbers from market sources. In other words, the most-publicized proof of cooling was partly a data correction, not evidence of a strategic retreat.

That gives you two completely different readings. The bearish one: the bulls were right to worry, the engine really is dying. The other: the single most cited number in the bear case was simply wrong, and has since been fixed.

The next quarter contradicted the first

Then look at what happened right after. In the second quarter, central banks bought back 289 tonnes — sharply higher than the weak first quarter and right back at the elevated annual pace of recent years.

Within one calendar half-year, official buying swung from "worst first quarter in fifteen years" to "back at lofty levels." Both cannot be the signal; one of those quarters is the anomaly, and we simply do not yet know which.

Meanwhile, the set of buyers that really did retreat in 2026 were not central banks. They were Western exchange-traded funds, whose holdings now sit about 1.5% below where they started the year after shedding 45 tonnes (roughly $4 billion) in the second quarter. The brutal March stretch — gold's worst month since 2013, down 12% — was a liquidity-driven deleveraging as real interest rates jumped and the dollar firmed, not a collapse in official reserve strategy.

That split is the whole story. It separates the bull market into two engines most people treat as one.

• The cyclical engine: US real interest rates, the dollar, and what Western ETF investors do with them. This is the lever that actually broke in 2026. It is the ficklest buyer in the market.
• The structural engine: central bank reserve diversification and de-dollarization. This never reversed. Official buyers kept purchasing, and their own surveys still show nearly nine in ten expect global gold reserves to rise over the next year, while roughly three-quarters expect the dollar's share of reserves to keep falling.

The correction you are watching is a fight between those two engines. The cyclical one stalled; the structural one went nowhere.

The wrong ruler

When someone declares "the gold bull is over," they are almost always measuring the wrong denominator. They point at price distance from the January record, and they point at a single quarter of central-bank tonnes. The first is a backward-looking hump. The second, as we have seen, was just rewritten by 80%.

The number that actually governs the next move is the cyclical lever — real yields and the path of the Federal Reserve — because that is what turned sentiment in 2026. Gold pays no coupon, so when inflation-adjusted bond yields rise and hold, the metal's zero income looks worse by comparison. Early 2026 flipped from pricing Fed rate cuts to fearing a hike, and that shift, more than any tonne count, is what cracked the price. Tamer inflation readings in August steadied it again.

Here is the inversion hiding in the open. The bad news that killed gold is a rate story, and rates are the most reversible variable in the whole trade. The investor whose ledger was supposedly the problem — central banks — did not behave the way the correction implies, and one of their quarters turned out to be a measurement artifact.

What the price is asking you to believe

At roughly $4,400 an ounce, about a fifth below the record, you are not paying for perfection; the January peak already embedded it. You are paying for the cyclical leg to keep wobbling. That is a less heroic bar than it sounds. Analysts are genuinely split on it: one house has reset forecasts to the low-$4,000s, while another still targets $6,000 by the end of the year and $6,300 into 2027. The spread between those two views is the whole argument about whether the bull paused or ended.

What would actually end the bull — the disconfirming signal to watch — is not what most people are watching. It is three things:

1. Q2's 289-tonne central-bank rebound turning out to be the anomaly rather than the weak first quarter, so the next couple of quarters print small like Q1 did.
2. Official buyers flipping to net sellers, a shelf that is already flickering as some Middle East sovereign funds sell gold to cover losses and a few central banks went the other way this year.
3. A real Federal Reserve tightening regime that produces sustained Western ETF outflows, because elevated interest rates are the engine that actually turned the price down.

None of those is true today. The structural buyer is still in the room; the cyclical one retreated and is hesitating. Bull markets end when the incentive to hold is priced out and the marginal buyer becomes a seller. Neither has happened yet — which is worth remembering the next time someone answers "is the bull over?" with a number that was, quite literally, a correction.