The Print That Didn't Flip

Issue #18 — October 20th, 2026

Wednesday’s CPI print was hot enough that rate futures briefly priced an October hike near 54%. Prediction markets barely blinked. This afternoon the October hold still trades at about 87% — the Number of the Week.

The Bureau of Labor Statistics reported headline inflation at 3.7% year over year (+0.6% on the month), with gasoline up 9.1% and energy up 5.2%. Core rose to 2.6%. CME FedWatch swung hard; Polymarket (offshore, data only) and Kalshi stayed near the mid-80s on a hold and finished the week higher, not lower. The board’s biggest live move was elsewhere: Flávio Bolsonaro faded from about 87% to 77% as runoff polls tightened ahead of October 25. December’s quarter-point hike still sits near 78%. — The Editor

Seven headline prices, Friday afternoon. Arrows compare with the Friday, October 9 snapshot behind last week’s issue. Bars are colored by category; figures are bid/ask midpoints from Polymarket’s international platform (offshore; cited as data only, not open to US residents) unless noted, verified live before this draft.

Eighty-seven to seventy-seven. That is the week’s largest swing on the board.

Polymarket’s Brazil presidential market (offshore, data only) moved Flávio Bolsonaro from about 87% last Friday to 77% this afternoon; Lula rose from roughly 13.5% to 22.5%. The catalyst was runoff polling, not a new count. A Datafolha survey released Thursday put Flávio at 47% and Lula at 45% of total intentions — a technical tie within a two-point margin — and 51% to 49% on valid votes. Flávio was down two points from the prior Datafolha. Quaest on Wednesday had a similar narrow lead. First-round tallies still stand: Flávio 47.03%, Lula 45.16%. The runoff is October 25.

Fed lines moved less. After the hot CPI, October hold rose about two points to the mid-80s and finished near 87%; December’s 25-basis-point hike rose about three points to roughly 78%. Those are real moves. They are not a ten-point fade.

Headline inflation is the all-items consumer price index — every category the Bureau of Labor Statistics tracks, including food and energy.

It answers a simple question: how fast did the prices ordinary households face rise over the past month or year? Wednesday’s print put that figure at +0.6% month over month and 3.7% year over year. Gasoline alone accounted for about half the monthly gain. “Core” inflation, which removes food and energy, rose 0.3% and 2.6% — still above the Fed’s 2% goal, but a different story from the pump.

Why the distinction matters for a Fed market: policymakers often treat energy spikes as temporary and lean harder on core. Markets sometimes disagree. After this print, futures briefly priced a higher chance of an October hike even though prediction markets still treated a hold as the base case. Headline is what shows up on the receipt; core is what many rate-setters say they watch.

Nine days before Brazil votes again, the question on Polymarket (offshore, data only) is unchanged: who wins the presidency, including any runoff.

Today the book reads about 77% Flávio Bolsonaro and 23% Lula — down from 87 / 13.5 a week ago. The contract does not settle on polls. It settles on the October 25 result (or a certified consensus if the count is delayed). That is why Sunday’s first round moved the price without ending the market, and why this week’s surveys can move it again.

What would move it next? A late poll that reopens a clear lead either way. A turnout surprise in the Northeast or among women, where Lula has led in recent surveys. A campaign shock in the final debates. Or simply a quiet week in which the 77% price holds into settlement day. The market is not asking who won the first round. It is asking who wins the second.

Last week’s Radar listed six items. One short rung settled. One print arrived hot. The rest are still open — which is how a calibration log is supposed to look mid-month.

Green. The Iran ladder’s October 15 rung on Polymarket (offshore, data only) settled Yes (no qualifying US strike by that date), after trading near 89% last Friday. A high-priced Yes that paid is consistent with a well-calibrated price; it is still one settlement.

Amber — open, and shifting. Brazil runoff: Flávio 87% → about 77%. Still the favorite; the fade is the news. October FOMC hold: ~85% → about 87%. December hike: ~75% → about 78%. Google best AI end-October: 71% → about 73%. Iran October 31 rung: ~75% → about 78%. None has resolved.

Graded on the print, not the price. September CPI printed hot (3.7% YoY, gasoline +9.1%). Last week’s Radar flagged it as the sheet that could move December. December’s hike price did rise about three points. October’s hold did not flip. That split is the lesson: a hot print can reprice the path without rewriting the next meeting.

Carried. SCOTUS reply clock (November 9) — still ahead. No grade.

Red. None this week.

Gasoline rose 9.1% in September. That is not an abstract index line. It is the number on the pump, and it drove about half of the monthly jump in consumer prices.

Advance retail sales still climbed: $745.0 billion in September, +0.8% from August and +6.3% from a year earlier. Higher gas prices inflate the sales total even when the gallons do not. Friday’s import-price report added another layer — import prices rose 1.1% in September, with fuel imports up 4.8%, and export prices up 2.2% (the largest year-over-year export rise since mid-2022).

A household does not see FedWatch or a 87% hold price. It sees a fuller tank cost and a grocery cart that still feels expensive. Prediction markets say the Fed is likely to wait until December for the next quarter-point. The receipt does not wait.

After Wednesday’s CPI, CME FedWatch briefly showed roughly even odds of an October hike. Polymarket and Kalshi stayed near 85–87% for a hold. Same news. Different prices. Why?

They are not the same instrument. Fed funds futures (and tools built on them) imply a probability from a continuum of rate outcomes and positioning. A prediction-market contract is a yes/no claim with an explicit resolution rule — for October, “does the Fed leave the target range unchanged?” Traders, capital, and time horizons differ. Liquidity can be thinner on one venue than another. And a futures-implied odds spike can reverse within hours while a prediction-market book moves more slowly.

Useful habit: when two “odds of a hike” disagree, ask which question each number answers and how it is constructed. Divergence is information about instruments, not proof that one side is “wrong.”

Headline inflation hit 3.7%. Core beat forecasts. Should the Fed hike on October 28?

Yes. The print was broad, not only gasoline: shelter up 0.4%, and core at 2.6%. Waiting until December risks letting a supply shock seep into wages and price-setting. Futures briefly priced a hike near 54% for a reason — the data surprised to the upside with one meeting left before the blackout ends.

No. Energy shocks are often treated as temporary, and the Fed already hiked in September. Most policymakers have signaled patience into year-end. Prediction markets still price an October hold near 87% on both Polymarket and Kalshi, with December — not October — carrying the next quarter-point. One hot month does not rewrite a meeting twelve days away.

We take no side. The book is about 87% hold. The meeting is October 28.