The long end reprices: a 2002 high on the 10-year, and why it matters that the rise is real

On 5 October 2026 the market yield on the US 10-year Treasury closed at 5.31%. The last time it stood that high was 14 May 2002. It has since eased — to 5.22% on 8 October — but the level is the point: it is 113 basis points above where it ended 2025, and above the 4.98% reached on 19 October 2023, the peak of the previous cycle.

That 2023 peak matters for this market in particular. The US spot Bitcoin vehicles began trading in January 2024, three months after it. Until mid-September 2026 the institutional holder base that arrived through them had only known a long end below that mark: between January 2024 and August 2026 the 10-year never closed above 4.79%.

The rise came after, not before, the Federal Reserve's decision of 16 September to raise its target range by 25 basis points to 3.75%–4.00%. On that day the 10-year stood at 5.01%. By 5 October it had added another 30 basis points — more than the policy move itself. The whole curve lifted: the two-year yield, 3.47% at the end of 2025, was 4.84% on 5 October, and three-month bills paid 4.22%.

A nominal yield can be read as two parts: the real yield on inflation-protected Treasuries, and the breakeven — the inflation rate at which the two securities would pay the same. Both are published daily, and their movements in this episode are unusually one-sided.

Between 16 September and 5 October, the 10-year real yield rose from 2.68% to 2.95%, and the 10-year breakeven from 2.33% to 2.36%. Of the 30 basis points the nominal yield added, 27 came from the real yield. Over the year the picture is the same: since 31 December 2025 the real yield has risen 102 basis points, the breakeven 11.

The decomposition has limits — breakevens carry liquidity and risk premia as well as expectations — but the direction is not in doubt. This is not, on the market's own pricing, an inflation scare. It is a rise in the return investors require in real terms to lend to the US government for ten years. At 2.95%, that return is the highest since 24 November 2008.

A bond pays its holder a real return; bitcoin pays none. The real yield is therefore the most direct measure of what an investor forgoes by holding it instead — the opportunity cost, expressed in the same units as the asset's own uncertain return. On 11 January 2024, the day the US spot vehicles began trading, that cost was 1.76%. On 5 October 2026 it was 2.95%, the highest since 2008.

It would be easy to turn that into a prediction. The record does not support one over short windows. Between 16 September and 5 October — the same weeks in which the real yield climbed 27 basis points — bitcoin rose from about US$75,600 to about US$86,500, roughly 14%. It has since given back part of that, to about US$82,600 on 10 October, still well above its level of about US$60,200 at the end of June but below the US$88,400 at which it began the year.

Over weeks, flows, positioning and the market's own recovery from a drawdown can outweigh a change in the discount rate. Over longer horizons the opportunity cost is a constant pressure rather than a timing signal. Both statements are true at once, and holding them together is more useful than choosing one.

The questions this raises are not new, but the numbers attached to them are. Three are worth asking again of any digital asset allocation, and of any manager running one.

It does not say where yields go next, whether the Federal Reserve will move again, or what bitcoin will do in response. The figures are daily observations on published series, dated where they are given, and they will have changed by the time this is read.

What it does say is narrower and, we think, more durable: the rise in long-term US rates since September has been a rise in real rates; real rates are what a non-yielding asset competes with; and an allocation case written in a 2% real-rate world deserves to be re-read in a 3% one.