The measured relationship between gold and the 10-year real rate: correlation, sample size, the controls it survived, and the one number that makes it useless as a trading signal.
When the 10-year real interest rate rises, gold falls, on the same day, with a correlation of −0.2764 across 1,243 trading days. It is the strongest macro relationship in the Apex register, and it survives correction for multiple testing. Silver behaves the same way but weaker (−0.1974), and the Nasdaq follows at −0.1647.
A separate measurement on payroll days found the effect is not just directional but proportional: roughly −0.95 % in gold per 10 basis points of real-rate move, monotonic across the range, holding in both halves of the sample. Measured 27–28 September 2026.
Gold pays no coupon. The real rate is what a risk-free asset pays after inflation, so it is the direct opportunity cost of holding an asset that pays nothing at all. When that cost rises, gold becomes relatively less attractive, and the price adjusts. This is one of the few market relationships with a mechanism simple enough to state in one sentence — which is part of why it holds up better than most.
Attribution and regime reading. When gold falls and the real rate rose, the move is explained and needs no story. When gold falls and the real rate did not move, something else is happening and it is worth finding out what. That is how Apex uses it: as a control on the explanation, not as an entry.
All numbers on this page were measured and published in Apex’s own record at the time stated. They describe the past under stated conditions and promise nothing about the future. All relationships · Research · Learn