Thirty Points in a Weekend

Issue #17 — October 13th, 2026

Last Friday, Brazil’s presidential market priced Flávio Bolsonaro at 57%. By Monday morning it was 84%. This afternoon it is 87% — the Number of the Week.

Sunday’s first round handed the senator 47.03% of valid votes to President Lula’s 45.16%. Neither cleared half, so the two meet again on October 25. Final polls had shown Lula narrowly ahead; the count did not. A market that spent September near a coin flip now treats the runoff as Flávio’s to lose — a price, not a result.

Brazil was not the only weekend jolt. On Thursday President Trump pledged no US strikes on Iran before the US midterms, and the near-dated ceasefire contract climbed 19 points. At home, Friday’s consumer survey slipped again. The Fed board barely moved: October hold about 85%. — The Editor

Seven headline prices, Friday afternoon. Arrows compare with the Friday, October 2 snapshot behind last week’s issue. Bars are colored by category; figures are Polymarket bid/ask midpoints unless noted, verified live before this draft.

Fifty-seven to eighty-seven. No other line on the board came close.

Polymarket’s Brazil presidential market moved Flávio Bolsonaro from 57% last Friday to about 84% by Monday and 87% this afternoon; Lula fell from 44% to roughly 13.5%. The catalyst was the count itself. In the last surveys before the vote, Quaest had Lula at 46% of valid votes to Flávio’s 45%, and Datafolha 45% to 42%, per O Globo. The electoral court’s full tally put Flávio first by about 2.2 million votes.

Note what did not happen: the contract did not settle. Its rules include any second round, so Sunday was information, not resolution. Elsewhere, the Iran October 31 contract rose from 56% to about 75%, Google’s end-of-October “best AI” price from 62% to 71%, and “another Fed hike in 2026” from 72% to about 78%.

Myth: “inflation expectations” are an economist’s forecast. Most often, they are a survey of ordinary people.

The best-known gauge is the University of Michigan’s monthly question: how much do you expect prices to rise over the next year, and over the next five to ten? Friday’s preliminary October reading put year-ahead expectations at 4.7% (from 4.6%) and long-run at 3.5% (from 3.4%) — both up a second straight month, both the highest since May. Michigan notes year-ahead stood at 3.4% in February, before the Iran conflict began.

Why a Fed market cares: central banks watch whether expectations stay “anchored.” If people expect higher prices, they may ask for higher wages and accept higher price tags, which can make inflation stickier. Rising expectations are part of the case hawks make for more hikes. December’s 25-basis-point hike trades near 75%; September CPI arrives October 14.

Read the Iran contract left to right and the price steps down a staircase.

Polymarket’s “US–Iran ceasefire continues through…?” ladder asks one narrow thing: does a US air or missile strike hit Iranian territory before each date? Threats and announcements do not count. Today the rungs read about 89% through October 15, 75% through October 31 (56% a week ago), 56% through November 15, 41% through November 30, and 34% through year-end.

The shape has a source. On Thursday, President Trump said the US would not attack Iran before the November 3 midterms, while keeping the blockade of Iranian ports in place. That pledge covers October’s rung and stops short of November’s — which is roughly where the staircase drops.

What would reshape it? Tehran says its reply to the latest US proposal will travel through mediators within days, and diplomats describe channels via Qatar, Pakistan and China as open but deadlocked on guarantees, enrichment and the Strait of Hormuz. A deal that outlasts Election Day would lift the November rungs. A broken pledge would sink October’s. The October 31 rung settles first; November 3 is the date the ladder is really pricing.

Last week’s Radar listed six items. Only slices of one have paid. That is normal, and it is the point of grading slowly.

Green. The short-dated rungs of the Iran ladder — through October 5 and October 7 — settled Yes, as did the September 30 rung we flagged at 91% two issues ago. And the 57% Brazil favorite on last week’s board finished first on Sunday. That is the first leg, not the contract: it pays on October 25.

Amber — open, and moving the way the board leaned. October FOMC hold, 82% → about 85%. December hike, 73% → about 75%. Google best AI by month-end, 62% → 71%. Iran through October 31, 56% → 75%. None has resolved; amber means the map held, not that it paid.

Carried. September CPI (October 14) and Kalshi’s Supreme Court reply (November 9) have not happened yet. No grade.

Red. None this week — which proves less than it sounds. A favorite at 57% leading round one is consistent with a good price and with a lucky one. Calibration is judged across dozens of settlements, not one green week.

Ask households whether now is a good time to buy a car, a couch or a refrigerator, and fewer said yes this month.

Michigan’s preliminary October survey found that buying conditions for big-ticket goods “plummeted amid high prices and borrowing costs.” The headline sentiment index slipped to 46.3 from 48.1, and the current-conditions gauge fell harder, to 44.7 from 50.9. The steepest drops came among lower-income households and those with smaller stock portfolios — the people with the least cushion against higher prices.

That is the household side of numbers this newsletter usually shows as percentages. Prediction markets price an 85% chance the Fed leaves rates alone this month and about 75% for another quarter-point in December. A household deciding whether to finance an appliance does not see either number. It sees the monthly payment. Michigan’s final October reading lands October 23 — after CPI, before the Fed.

Every contract carries a small print of its own: the resolution rules. They decide what “Yes” means, who certifies it, and when.

Three markets on this week’s board show how different those rules can be. Brazil’s contract resolves on a consensus of credible reporting, with the electoral court as tiebreaker — and it explicitly includes any runoff, which is why Sunday moved the price without ending the market. The “best AI” contract resolves on a single public scoreboard: whichever company holds first place on the arena.ai Text Arena at noon ET on October 31. On the October 8 board, a Google model led an Anthropic model 1534 to 1515 — with the leader still tagged “preliminary.” The Iran ladder resolves No only on a qualifying US strike that lands on Iranian territory; intercepted missiles and threats do not count.

Useful habit: before reading any price, read the question the way the contract does. A headline can feel decisive and leave a contract untouched — or the reverse.

Flávio Bolsonaro finished about two points ahead of Lula, and the market jumped to the high 80s. Is October 25 a formality?

Yes. He beat final polls that had him trailing, and some of the eliminated candidates were right-wing alternatives whose voters, in this view, lean his way — the BBC called the runoff “his to lose.” Momentum, an incumbent who admitted the result was “unexpected,” and a price that held its gains all week point the same direction.

No. A 47–45 split leaves about 9.3 million valid votes — roughly 7.8% — with candidates no longer on the ballot, and turnout can shift between rounds. Pre-vote head-to-head polls had the two level within the margin of error. If pollsters underestimated Flávio once, the runoff surveys may or may not repeat the miss. Lula, in this view, has three weeks and a campaign that says it is “in extra time.”

We take no side. The book is about 87% Flávio. The runoff is October 25.