What is the front end of the curve saying about the path of policy?

**Marked to market, 20 September — the test fired on the first branch.** This position asked whether Fed speakers or strong data would push the five-year back above its 4.86% decision-day close. Inside a session it did: 4.86% on 18 September, up six basis points, erasing the drift the last mark recorded. By the reasoning set out here, a five-year that reclaims its decision-day level says the market's single-increase reading was premature. The more striking figure is at the other end of the curve. The ten-year settled 5.00% — unchanged, to the basis point, from its close the session before the committee moved — while the thirty-year gave up three basis points. A quarter-point of tightening moved the five-year three basis points net and the ten-year none, which says the front end is pricing the path and the long end is pricing something else.

**Marked to market, 18 September — the test resolved on the second branch.** This position said a five-year that drifted back would mean the market had taken the committee at its word on a single further increase. It drifted back on the first full session: down six basis points to 4.80%, three below the 4.83% pre-decision close, with the ten-year at 4.95% and the thirty-year at 5.30%, a parallel move rather than further flattening. Equities rose above their pre-decision level at the same time, so financial conditions are now easier than before the hike. We read that as the market pricing the end of the cycle at its start. The counterweight is the committee's own reaction function — conditions that ease after a hike are the usual case for the next — and one large bank has moved its forecast of the next increase to October. The falsifier (the five-year back through 4.55%) has not fired.

**Marked to market, 17 September.** This position named the 48 hours after the decision as its next test: whether the five-year kept leading the thirty-year once the number was out. On the first session it did — the five-year up three basis points, the ten-year one to 5.01%, the thirty-year down one. That is a continued flattening, and on this position's own terms it says the market is pricing a path beyond this move rather than treating the move as terminal. The scale is modest: one further increase was signalled, and a report on the published projections reads that as the last move through 2027.

What the first session added is where the adjustment went. A long bond that rallies on a first hike is crediting the institution rather than charging it for inflation, so domestic borrowing costs tightened less than the decision implies and the dollar carried the rest. The front end is still the part of the curve that carries a surprise in the path; the long end has shown it will not.

Marked to market on 12 September, in the direction this position argued but well beyond its scale. Where version 4 read a front end refusing to price a cut, the question has since changed shape entirely — the front end is now pricing a rise. From 4 September to 11 September the five-year added 24 basis points to 4.79% while the thirty-year added 11 to 5.36%, so the curve flattened and the front end did more than twice the work of the long end. That asymmetry is the substantive update: policy-error risk has moved from the issuer to the reaction function. The position's reading that the path was being set by something other than the news the market was trading still holds; what set it was an energy shock arriving at a central bank with no room to look through it.

This position last recorded a front end that had taken a softer path on July producer prices before the whole curve cheapened into weak consumer data, finishing 4.36% at five years, 4.70% at ten and 5.26% at the long bond on 14 August.

The test it named — the next price releases — has now landed, and it landed alongside the largest piece of good energy news of the month. Brent closed lower three sessions running to $87.84, 6.94% below its 21 August settle. On 26 August the inflation gauge released that morning was reported as still well above the Federal Reserve's target, keeping hike expectations alive, and every maturity sold: the five-year up three basis points to 4.38%, the ten-year three to 4.67%, the thirty-year two to 5.19%. The dollar posted its largest advance in over a fortnight.

The weekly view is where the two ends separate. Across five sessions the five-year is three basis points higher and the thirty-year one lower, compressing the thirty-to-five spread from 85 to 81 basis points. An energy-led disinflation transmits to the front and belly first, because that is where the policy path is priced. This one transmitted nowhere.

Europe supplies the control. German import-price inflation is scheduled to accelerate to 7.2% year on year from 6.1%, euro-area selling-price expectations to 25.0 from 17.7 and euro-area headline inflation to 3.1% from 2.9% — all into the same falling barrel.

The 27 August session removes the last easy explanation. If the equity re-rating were a rates trade, the front end would have supplied it; instead the only maturity that moved went the wrong way for a cut, and it did so with crude 6.22% below its 21 August settle. The session before had already closed lower on an inflation print reported as hot, and CoinDesk reports desks now pricing a hike rather than a cut.