US Treasury Yield Repricing — Pre-registration

The thesis

As of 2026-08-12, one falsifiable claim: the US 10-year Treasury yield is on a sustained trajectory toward and through 5.0%, while the 30-year is establishing a durable foothold above that level. Two separable parts, tracked independently. (a) Structural repricing: the term premium at the long end is re-rating higher, driven by fiscal supply, competition for capital from private-sector AI capex, and a Fed that has shifted from cutting to contemplating hikes. (b) Forced revaluation: that repricing transmits to equities and credit, forcing a broad revaluation of risk assets. (a) can confirm while (b) does not. Confidence at registration: medium on the ladder; no numeric probability, because no calibrated basis exists.

What would prove it wrong

Five conditions define when this is wrong; each is a measurement target, not a belief. Statuses below are observations as of today.

F1 and F2: twenty sessions closing below 4.30% on the 10-year, or 4.80% on the 30-year, by end-Q4 2026 would refute durability. F3 attacks the mechanism: breakevens above 3.0% alongside 50bp or more of Fed cuts would mean an inflation regime with an easing Fed — the wrong story. F4 attacks transmission: a 50bp-plus yield drop in ten sessions amid an equity selloff would make Treasuries the refuge again. F5 attacks the fiscal premise: enacted deficit reduction of 2% of GDP or more would undercut the supply driver. None of these lines being met does not confirm the thesis — it survives into the next window. Confirmation needs positive evidence; for part (b), revaluation visible in equities and credit, of which there is none.

Where things stand

The evidence behind part (a). The 30-year has closed at or above 5.0% for 25 consecutive trading sessions; the last close below the level was 4.99% on 2026-07-06 (FRED DGS30, through 2026-08-10). The confirming series is duration: 10-year 4.72%, 30-year 5.25% (FRED DGS10/DGS30, 2026-08-10), up from 4.38%/4.95% three months earlier. The composition is the crux: real yields at or near multi-year records — 10-year real 2.43%, 30-year real 3.00% (FRED DFII10/DFII30, 2026-08-10) — while the 10-year breakeven stays contained at 2.27% (FRED T10YIE, 2026-08-11). This is a term-premium repricing, not an inflation repricing; the steep +0.48% 10Y−2Y curve (FRED T10Y2Y, 2026-08-11) fits a market paying for duration.

Treasury’s quarterly refunding (2026-08-05) reads as revealed preference: the Q3 2026 borrowing estimate rose to $739B, up $68B on the quarter, yet coupon and FRN auction sizes stayed flat “for at least the next several quarters,” pushing marginal funding onto bills — consistent with a debt manager seeing the long end as demand-constrained. Fed policy reads accommodative, 3.63% against a 4.75% Taylor-rule estimate, and the debate is now hold-versus-hike rather than cut-versus-hold (centralbank.watch, mirroring CME FedWatch data, 2026-08-12). Because the marginal long-end buyer is often a foreign official, these flows matter: Japan’s finance minister said the government will steer the $1.8T GPIF toward domestic assets (Reuters, 2026-07-10); US and Japanese authorities executed a coordinated yen-buying intervention on 2026-07-31, confirmed by Japan’s Ministry of Finance (Reuters, 2026-08-03); and China’s PBoC bought gold a 17th straight month through March 2026, framed as reduced US-asset exposure (World Gold Council, 2026-04-14).

The case against, at full strength. Transmission is absent, so part (b) has zero confirming evidence: the S&P 500 closed at 7,728.2 on 2026-08-11, within 0.4% of its 2026-08-07 record (FRED SP500), and credit sits at or near cycle tights (HY OAS 2.72%, IG OAS 0.79%; FRED BAMLH0A0HYM2/BAMLC0A0CM, 2026-08-11). Inflation is softening, not accelerating — July CPI at 3.4% headline and 2.5% core (BLS, 2026-08-12) — which weakens the hike leg. The only mechanical transmission visible at baseline is mortgages: 6.69% on the 30-year fixed (Freddie Mac PMMS, 2026-08-06). The base rate looms: in October 2023 the 10-year touched ~5% briefly, then fell sharply as supply and demand rebalanced (public market history).

The strongest counter-read is the auction-demand pattern: buyers have absorbed long-end supply at the level without bidding the level down — the August 3-year cleared at 4.291% with a 2.71 bid-to-cover and a −0.5bp tail (Treasury auction results, 2026-08-11). That fits a regime floor — or yield-starved allocators near a mean-reversion top. Regime floor versus mean reversion is the central open question; the cleanest live discriminator is whether the next long-end auctions — the 10-year and 30-year sales through 2026-08-13, then refundings through Q4 — show widening tails or shrinking bid-to-cover.

Open questions: is there a yield where long-end demand fails outright (bid-to-cover, tails, indirect share; TIC, two-month lag); can bills absorb roughly $300B of marginal issuance per quarter without stress; is the soft jobs print noise or a labor turn? Named catalysts: July FOMC minutes (2026-08-19), July PCE (2026-08-28), the August jobs report (~2026-09-04), the September FOMC and SEP dots (2026-09-15/16), and the November refunding (2026-11-04), which resets the coupon path. As of this registration, the evidence tilts toward part (a) strengthening, at medium confidence; part (b) is unconfirmed.

How to read this series

Every future episode carries this full ledger forward — met conditions stay met — and any change to a condition, threshold, or window appears only as a dated revised entry. Episodes arrive at irregular cadence, keyed to the catalyst calendar rather than a clock. The thread closes with a sunset ledger: a terminal status for each condition and one unhedged verdict — confirmed, falsified, or unresolved. The thresholds are frozen before the outcomes; readers can score us against them.