Nominal yield minus inflation expectations: what the real rate is, where it stood in 2026, and what Apex measured about its effect on gold, Bitcoin and equities.
Real rate = nominal yield − expected inflation. It is the return a lender actually earns after prices have risen, and it is the cost of holding anything that pays nothing — gold, Bitcoin, a growth stock whose earnings are years away. When the real rate rises, that cost rises with it. Everything else about "rates and risk assets" follows from this one subtraction.
The cleanest measure is the 10-year Treasury inflation-protected yield (the market's own real rate), or the nominal 10-year minus the 10-year breakeven inflation rate. Apex uses the latter, from FRED, and expresses it as a percentile of its own history since 2003 so that "high" has a meaning.
Gold pays no interest. When the real interest rate — the nominal yield minus inflation — rises, holding gold costs more relative to holding a bond, and gold tends to fall. Everyone knows the story. The question Apex asked was whether it survives measurement.
Jobs reports move rates sharply and for a known reason, so the rate change is close to an exogenous shock rather than a response to something else. On ordinary days rates and gold both respond to the same news and the relationship blurs.
An early version of the test binned rate changes into buckets. The p-value went from 0.00005 to 0.09 — the finding nearly disappeared. Binning threw away the gradient that carried the signal. Apex now tests continuous relationships continuously.
The popular thesis is that Bitcoin is a hedge against monetary debasement and should rise when inflation does. Measured over the 120 days to September 2026, Bitcoin's correlation with the real rate was −0.28 — it fell when money got more expensive, like any other long-duration asset. The debasement criterion (Bitcoin rising with rates while equities fall) has not been met; Apex records the thesis as not confirmed so far.
There is one conditional exception, and it is instructive. When the real rate is above 2 % and equities fall and bonds fall on the same day — capital leaving both — Bitcoin was up the next day 66 % of the time against a 50 % baseline (n = 85). It passed the half-split test and fell only because the family-wise correction over 276 hypotheses demanded p < 0.00018 and it had 0.001. It is now traded in shadow to find out.
High real rates compress valuations by raising the discount rate on future earnings, and they do it unevenly: long-duration growth first, cash-generating cyclicals last. That is why the sector spread matters more than the index when rates move — see capital rotation. In 2026 the Nasdaq kept rising with real rates at the 99th percentile, which Apex reads as "rates are restrictive but not yet the brake" rather than as a contradiction.
Continuously, never binned; on event days where the rate move is closest to a shock; against the unconditional baseline; with a half-split by date; and against the whole family of hypotheses ever tested. The binning lesson above is the reason: a real relationship can be destroyed by the way it is measured. Full method in the methodology.
All numbers on this page were published in Apex’s own record at the time; they describe the past under stated conditions and promise nothing about the future. Research · Learn · The Apex day