The near meeting got repriced. The path did not close.
Last week's useful move for automated traders was not "hawkish or dovish." It was a collapse in the October binary while December stayed live.
This post is educational. It is not financial advice and it does not promise returns.
Speak and PCE started the unwind
On 29 September, New York Fed President John Williams told the University at Buffalo that after the September hike there was "no need for urgency," and that the Committee had "time to gather more information." In the same remarks he said that if the economy evolved roughly with his forecast, "one further upward adjustment" of the federal funds target range "may be appropriate late this year" — not that October was locked.
That is path language with a calendar preference: one more move possible, not a commitment to the next sitting.
On 30 September the Bureau of Economic Analysis released August personal income and outlays. The PCE price index rose 0.3% month over month and 3.4% year over year. Core PCE rose 0.2% on the month and 3.0% over the year. The release also incorporated the annual update of the national accounts. Market coverage treated the print as cooler than consensus on core, and fed funds futures started to fade a back-to-back October hike.
On 1 October, Vice Chair Philip Jefferson spoke at UVA. He supported the September increase to 3-3/4 to 4 percent, kept upside risks to inflation in view, and said future adjustments should be determined by trends in the data, the outlook, and the balance of risks — language that buys time without taking December off the table.
By midweek, the October meeting was no longer the market's base case. The path question had already shifted forward.
Payrolls finished the October binary — not the year
Friday's Employment Situation (BLS release USDL-26-1549, 2 October) was the hard stop for near-term hike pricing. Nonfarm payrolls rose 29,000. The unemployment rate printed 4.2%. Average hourly earnings were $37.81, up 0.1% on the month and 3.0% over the year. July and August payrolls were revised down by a combined 60,000.
That is a soft labor print into a Committee that had just hiked and was still debating how soon to go again.
CNBC's same-day FedWatch read put the probability of a 25-basis-point October increase near 17%, down from about 36% a week earlier. Kalshi's October hike odds were reported near 18%, down from almost 70% a week prior. December was a different story: FedWatch still showed better than 75% odds of a hike later in the year on that coverage.
So the week did not deliver "no more tightening." It delivered "probably not on 28 October," while leaving a live late-year path and a data calendar still full — September CPI is next on the tape (14 October on the congressional joint-economic schedule).
For an algo book, that split matters more than the headline tone. Many stacks treat FOMC week as a special state, then treat a crushed near-meeting probability as permission to restore normal size. If the model of risk was only "will they hike next time," those permissions come back while the continuous path — CPI, another payrolls print, energy, and December — is still rewriting discount rates.
Durable lesson: odds can migrate without risk leaving
Treat policy as two prices.
The meeting price is discrete. It is the futures-implied probability that the next sitting delivers a move. That price can fall from a majority to a low teens print in a few sessions when speak, PCE, and payrolls line up.
The path price is continuous. It is how many more moves, how soon, and through which data. When October odds collapse and December stays heavy, path risk did not die. It moved.
For operators running automated risk, the practical questions are mechanical:
- After a data day that kills the near meeting, which strategies regain full size, and which stay on path-aware limits until the next CPI and the December strip stop moving violently?
- Does your risk layer treat "October hold priced" as green light while December still prices another hike above 75%?
- When payrolls gap the front end, do rates, FX, equities, and any mirrored books hit independent caps, or do they all lean on the same stale "post-event" flag from September's FOMC?
- Is 14 October CPI already an elevated state in the control plane, or only a calendar note next to 27–28 October?
The failure mode this week is familiar: systems that key off the nearest binary, then go quiet when that binary dies, while correlated books keep trading a path that never left the screen.
Soft next step
If a rates-path repricing changed how much risk you want mirrored across accounts, the free copy ratio and lot size calculators can help with the arithmetic. They do not replace broker specs, exchange rules, or a kill switch.