THE CARD
Call Our number Market now Grades Fed hikes 25bps in September 18% — revised UP from 12 30–31¢ (entry 56.5¢) Sept 16 Three dated tests, each with its bar printed in advance: Wednesday Aug 26 — July core PCE: at or above 3.3% year-over-year, or hotter than +0.3% for the month, and this call breaks. Same day: the deadline on a sitting governor's removal response. Friday Aug 28, 10am — Chair Warsh's Jackson Hole keynote, his first substantive remarks of the cycle: a declarative link between rate policy and stalled disinflation fires the falsifier; structural musings hold the firewall. Friday Sept 4 — August payrolls: above +150,000 with net upward revisions, and the call is wrong.
The pre-read, graded against itself
Tuesday's issue committed to two things about the minutes, and they graded in opposite directions. The signal condition fired above its own bar: we scanned for "a few other participants" flirting with tightening, and the document said "many participants assessed that policy tightening would likely be necessary if inflation did not decline" — the strongest quantifier the Fed's vocabulary ladder offers, explicitly wider than the three known dissenters. The market prediction failed: we printed that a fourth sympathetic voice would gap this market up, and the price sat frozen at 28.5¢ through the release and for four hours after, then drifted a point the other way. Both graded, both on the record.
How can the most hawkish sentence of the cycle move nothing? The minutes answer their own riddle: they record that in late July the market was "fully pricing in a 25 basis point hike by the September meeting." The hawkishness in that room was already in the price a month ago — then the August payrolls miss unwound it. The document that landed Wednesday described a market, and a mood, that no longer exists. The desk called this the time-capsule problem last week; the tape just confirmed it in the market's own handwriting.
Why the number went up anyway
Here is the part the desk would rather explain than hide. The operative sentence is a conditional: tightening likely "if inflation did not decline." August's consumer prices declined, so the condition has not fired — but it is armed, and it resolves this Wednesday when July's PCE prints against the staff's 3.3% June level. A live, armed conditional carries probability mass whether or not it ultimately fires. Holding 12% while pointing at the conditional as a defense was, in the desk's own steelman words, anchoring dressed as discipline — if there's roughly a one-in-four chance Wednesday's print sticks at 3.3%, the baseline number has to absorb that risk rather than wave at it. The call moves to 18% — the desk's first upward revision — still twelve points under a market that is pricing yesterday's fear, but no longer pretending the conditional is free.