FO Analysis: The High-Water Mark.

THE VERDICT

The Desk Wrote a Test. The Market Answered. This Is That Note.

CONFIRMED · On 3 August this desk published a sentence it knew might have to be eaten: "If September pricing slides back below half and stays there, last week's seven dissents will read in hindsight as the hawks' high-water mark, and the desk will write that note."

OBSERVED · It slid below half on 7 August, on the worst payrolls print of the cycle, recovered to roughly half early the following week as crude re-fired, and then spent every settlement after the inflation data below it: 42% after Wednesday's consumer prices, and roughly a third at both Thursday's and Friday's closes, computed from the September fed funds contract itself. The published condition was a slide below half that sticks. Three consecutive post-CPI settles is what sticking looks like, and the score stands. The seven hawkish dissents of late July were the high-water mark of hike pricing, and this is that note. No lawyering: the desk's test said the market would keep a September rise majority-priced through benign data, and the market did not.

FO INFERENCE · What the desk got wrong deserves one precise sentence, because the error is instructive: the desk believed the reaction function's asymmetry would hold the policy rate's pricing up through soft data, and instead the market found a cheaper instrument for the same conviction.

OBSERVED · Because the second half of the week is the other half of the story. While September pricing fell to a third, the market gave back its post-CPI rally into the ten-year auction deadline four and a half hours later, roughly five basis points of rebuilt concession on the desk's computation, then watched the thirty-year tail its deadline level the next day at 5.216%, the highest since 2001, with dealers left holding little over a tenth of it. Retail sales printed their worst month in over a year on the same Friday that households told the Michigan survey they expect 4.3% inflation over the next year, up on the month. The two-to-thirty-year curve ended at 108 basis points, the steepest of this cycle. And the desk's refining gauge re-widened to $66.72 through the largest crude inventory build since January 2023.

FO INFERENCE · That configuration has a name in this desk's framework. The ratchet was never a claim about the funds rate; it was a claim that supply shocks leave the policy stance tighter than they found it. For five months the stance expressed itself through hike pricing. Last week it stopped, and the same pressure re-emerged at the long end, priced not by a committee's intention but by the auction clock, at the highest thirty-year stop since 2001. The market took the pen from the Fed and wrote the same sentence in a different hand.

Data timing: every market level in this note is an exchange settlement or day-final official print through Friday 14 August 2026, with two exceptions dated and labelled as such: the Strait of Hormuz shipping halt and the dollar index's break of its August floor, both Monday 17 August, the first press-reported, the second an intraday level that is flagged where used and never scored. Treasury yields are the US Treasury's daily par curve at official closes. September probabilities are the desk's computation from fed funds futures settlements against the New York Fed's effective rate; press-cited ticks are quoted with their own timestamps and never averaged with settlements. The crack spread is the desk's per-contract computation from NYMEX settlements, with the August-to-September roll flagged in-series as contract basis. Positioning is the CFTC report as of 11 August, released 14 August: it cannot yet show the response to Friday's data or Monday's break. Spot FX and gold are New York day-final closes of 14 August. No unsettled session is scored anywhere in this note.

The data points above are the data spine of this Premium. Throughout the body the desk labels material as Confirmed (a sourced primary fact), Observed (a live market level), FO inference (the desk's read), or FO risk scenario (a forward path under specific triggers). When in doubt, follow the labels.

THESIS PERFORMANCE · SCORED AGAINST THE TAPE

The Trigger, Scored at Full Weight

A Premium thesis is only as good as its performance record. Subscribers should expect this scorecard in every Premium.

OBSERVED · The path down, for the record the desk keeps: 67% on 31 July, the post-decision high. 44.4% on 7 August, the payrolls shock. A brief recovery to roughly half by Tuesday the 11th as crude re-fired. Then 42% after Wednesday's inflation data, and roughly 34% at both Thursday's and Friday's settlements, computed from the September contract itself. The October contract adds barely two conditional points: the strip has not deferred the rise, it has removed it.

OBSERVED · Two honest annotations, and then the score stands. First, the slide began on labour, five days before the inflation prints the test was written around; the desk records that contamination here and does not lean on it, because the test's own words named "benign data" and benign data is what arrived, three releases in a row. Second, the week's prints were genuinely soft on the statistics the market prices: an unrounded core of 0.215%, a flat producer headline (the pipeline cut, core ex-food, energy and trade, ran +0.4 and sits in the fact box), a retail decline with breadth. The market did not misread the data. Where the desk parts company with the consensus is on what the market did next, which is the subject of the rest of this note.

The scoreboard consequence is recorded in the table below, alongside the calls that survived the same week. A record that cannot carry a miss at full weight cannot carry a hit either.

The full desk read continues below for Premium subscribers: the two auctions read against Friday's stagflationary tape, the retired band and the redrawn 5.05 to 5.45 operating range with its explicit invalidation, the third act of the refining margin with the Hormuz clock attached, the Regime Dashboard, where the crowd is standing in the positioning data, the six-instrument Cross-Asset Breakdown, the three dates that decide the fortnight, the probability-weighted Scenario Map with per-path invalidation, the Trader's Checklist, the glossary, and the Premium PDF.