Opportunity cost, not inflation

Hold a 10-year Treasury earning 4% nominal with inflation at 3% and your real return is about 1%. Hold gold and your real return is zero minus storage and financing. That gap — the real yield — is what you are giving up.

This is why the inflation-hedge framing misleads. If inflation rises but nominal yields rise faster, real yields rise, and gold can fall while inflation is accelerating. What matters is not the level of inflation but what safe bonds pay after it.

Which real yield to use

10-year TIPS yield — the cleanest market-implied real rate. It is quoted directly, needs no assumption about inflation expectations, and is the series most professional desks anchor on.
Nominal minus breakeven — equivalent by construction, but lets you decompose the move: did real yields rise because nominals rose, or because inflation expectations fell? Those two have very different implications for gold.
Front-end real yields — matter more during aggressive policy cycles, when the market is repricing the near-term path rather than the long run.

Decomposition matters more than the level

A rise in real yields driven by nominal yields spiking is usually bad news for gold and comes with a stronger dollar and tighter financial conditions. A rise driven by falling inflation expectations is often accompanied by growth worries — and gold may hold up or even rise despite the higher real yield. Same number, opposite outcome. Always decompose.

When the relationship breaks

The real-yield model is a workhorse, not a law. It degrades in three situations:

Crisis episodes — safe-haven flows overwhelm the opportunity-cost channel for weeks at a time.
Official-sector regime shifts — when central banks step up buying, they are not responding to real yields at all.
Very low absolute levels — when real yields are already deeply negative, further declines have diminishing effect, and the relationship flattens.

How to check this yourself

Put a 10-year TIPS real-yield chart and a gold chart on the same date range. Do not look at the levels — look at the turns. Mark the stretches where real yields fell hard and see what gold did. Then find the stretches where the relationship inverted and note what else was going on. The point is not to prove the correlation is strong; it is to see that it breathes, so you stop treating it as a constant.