The Ten-Year Broke 5%; We Correct Our Equity-Cushion Number and Withdraw the Hormuz Gas Near-Zero

The UST 10-year printed 5.016% on the eve of a near-certain hike, and the entire move was REAL — the breakeven and the 5y5y forward both FELL. Two corrections today, both ours. We published an equity risk premium of -0.19% on a record low; on the anchor our own valuation standard specifies it is +0.20%, the 8th percentile since 1982 — thin, not negative. And on Hormuz, an LNG carrier DID transit and gas DOES have a ship-to-ship workaround.

(1) The 10-year broke 5%, and the decomposition is the whole story. UST 10y 5.016%, through the round number for the first time this cycle (4.987% Monday, 4.973% on the 11th); 2y 4.673%, 30y 5.378%. Now take it apart. The 10-year REAL yield is 2.624% and making new highs, while the 10-year breakeven FELL to 2.379% and the 5y5y forward FELL to 2.306% from 2.321%. Nominal yields are rising because real yields are rising; inflation compensation is going the other way. This is not an inflation scare — it is the market believing the Fed. We have named a breakeven-LED leg as the regime-changing tail. It has not fired, and this week it receded. Counter-intuitively that is the harder outcome for equities: de-anchoring at least inflates nominal earnings, whereas a pure real-rate rise raises the discount rate against unchanged cash flows.

Global Macro · Rates

(2) CORRECTION: our equity-cushion number used an anchor our own standard rejects. Yesterday we published a CAPE-versus-real-yield gap of -0.19%, negative for the first time and the lowest of its history. Wrong on both counts. We discounted at the raw 10-year TIPS; DELPHIC_VALUATION_STANDARD §2 specifies the Cleveland Fed model real rate, inflation-risk-premium-stripped, and states the preference over TIPS explicitly. The 38bp between them is that premium and it flipped the sign. Corrected: +0.20% on CAPE, +1.59% on trailing earnings — the 8th percentile since 1982, against a true low of -1.14% in February 2000. The TIPS series begins in 2010; a sixteen-year window masqueraded as a record. The house specifies that anchor partly to get the history — which is exactly what the shortcut cost us.

Delphic · Method

(3) What survives: the compression is a discount-rate event, not a re-rating. Since end-2021 the gap compressed 232bp, of which 93% is the real rate and only 17bp the earnings yield. The cushion was crowded out, not spent. That changes the repair path: on cyclically-adjusted earnings equities are priced roughly where they were at end-2021, so an ordinary correction does not restore it — lower real yields or delivered earnings growth do. And a naming discipline we should have applied from the start: this is an aggregate yield gap, NOT the Delphic ERP. The house ERP is name-specific and measured by inversion (frameworks/erp_compression.md), aggregate anchor ~4.3%. Quoting a 0.2% gap and a 4.3% ERP in the same breath without that distinction is how a reader concludes we cannot add up.

Global Macro · Valuation

(4) WE WITHDRAW: an LNG carrier did transit Hormuz, and gas does have a workaround. Our SoH file has carried for two months that no LNG carrier had transited since 11 July, and that gas — unlike crude — could not be ship-to-ship transferred and so had no way around the Strait at all. Both are wrong. The Qatari carrier Al Maruna, loaded in early August, was in the Gulf of Oman on the morning of 8 September bound for Port Qasim; it cleared the Strait around 7-8 September. Three LNG cargoes moved by STS in August — Al Rekayyat to Tembek, GasLog Shanghai to GasLog Savannah, Mraweh to LNG Enugu — at over $1m and 30-35 hours apiece. Worse: our 13 September edition named 'an LNG carrier transits' as a condition for tipping back to Phase 3, and that condition had already fired five days before we wrote it. The failure was not the data. It was that we never re-tested a claim once it had become load-bearing.

Geopolitics · Hormuz

(5) What replaces it is better: the gas shock is two shocks with opposite half-lives. The shipping premium is removable; the capacity loss is not. Two of Qatar's fourteen liquefaction trains are damaged — 17% of capacity, 12.8 mtpa — with QatarEnergy guiding three to five years to repair. War-risk cover, escort scheduling and STS cost all decay on a diplomatic outcome. No ceasefire rebuilds a train. Roughly 36 mt of Qatari and Emirati supply is absent year-to-date, only ~20 mt offset by US and Canadian volumes; Asian spot is up ~150% and European ~100%. Anyone modelling this as a single removable war premium is mispricing the back of the curve. In our own framework's language, part of this cost shock has already converted from a premium into a level — the conversion we said to watch for, now evidenced rather than feared.

Geopolitics · Energy

(6) The growth axis has crossed below zero. The Radar's growth axis prints -0.05, its first negative reading of this cycle, with copper momentum the deepest component at -0.70. Monetary/Liquidity prints a cycle high +0.64; Inflation eased to +0.48 as the 5y5y de-anchor component came off. Read together: the tightening is intensifying while growth rolls over and inflation expectations recede. A real-rate squeeze, not an inflation scare. Equities took it accordingly on Monday — S&P -0.48%, Nasdaq 100 -0.82%, Europe -1.0%, EM -1.4%. Dollar real rates set the global cost of capital, and the damage ranked accordingly.

Global Macro · Regime

(7) A caught data fault, because we would rather publish it than bury it. Today's first Radar build printed GROWTH UP +0.64 — a nonsense reading on a day copper was flat and equities fell. The cause: our local copper series quoted dollars per pound and stopped updating on the 11th, while the vendor sync carried the 14th and 15th in cents. The join produced a +9,567% one-day move in copper, which saturated two of the three growth components at their caps. The deny-list that was supposed to catch this was hand-maintained, and a hand-maintained list needs a human to notice first. It is now an automatic scale check on every dual-source series, with a regression test. The corrected reading is the -0.05 above — the opposite sign to what the faulty build produced.

Delphic · Method

(8) The hike is not the question; what a synchronised DM hiking cycle does to a cushionless market is. The FOMC decides tomorrow with a 25bp hike ~88% priced and no cut priced anywhere. The external read now frames this as a broad DM hiking cycle rather than a Fed story: eight of nine DM central banks expected to hike by year-end, the three largest twice, a balanced Taylor rule guiding roughly 100bp higher, and a third consecutive year in which immaculate disinflation has failed to arrive. The ECB has taken its 2027 core forecast to 2.6% from 1.9% last December. The information is in the dots, not the decision.

Global Macro · Policy

What We're Watching

• Wed 16 Sep — FOMC. 25bp ~88% priced; the SEP and the dots carry the information. Watch for a 2027 core PCE revision toward 2.6%.
• Wed 16 Sep — BoJ; Thu 17 Sep — BoE. The broad DM hiking cycle in practice; the BoE expected to hold against a 5.37% gilt.
• The 5-5.5% band on the 10-year. Where the equity-yield correlation is at risk of flipping. We entered it this week.
• Breakevens. A breakeven-LED yield move remains the regime-changing tail. It receded this week — that is the benign read.
• Qatari LNG loadings. Four or more laden cargoes clearing inside a fortnight would be a genuine restart. One hull was not.
• Copper at $6.32/lb. The deepest negative in the growth axis; a stabilisation there matters more than another equity bounce.

The Bottom Line

The Fed question was settled by the August core and tomorrow is about the dots. The live question is what a synchronised DM hiking cycle does to a market whose cushion sits in the 8th percentile — and our answer is that it does not crash it, but it removes the market's capacity to absorb rate moves, which is what a growth axis crossing below zero on a small real move is telling us. Own quality, low leverage, real earnings and energy; keep duration light. Two corrections today and both are ours: the cushion is thin rather than negative, and the Hormuz gas near-zero is withdrawn in favour of a better distinction — the shipping premium is removable, the 12.8 mtpa of damaged liquefaction is not.

Synthesised from: Delphic SoH Monitor (15 Sep 2026), Daily Market Commentary (15 Sep 2026), Regime Radar (15 Sep 2026) · Delphic ERP module (_common/data/erp.py) — own construction from Shiller P/E, CAPE and TIPS 10y real · Bloomberg-sourced house store: UST curve, breakevens, 5y5y forward, equity indices, commodities · Vessel-tracking and trade press for Hormuz transits, STS transfers and Qatari liquefaction capacity · External macro and strategy reading, distilled — see macro/global/_parsed/2026-09-15 · S&P Global PMI press releases; NY Fed Global Supply Chain Pressure Index