AFOS Weekly · USA-2026 midterms Political Risk Weekly

AFOS Analytics

Real-time political pricing
Prediction market × polls × press
no smoothed averages

Harvard Dataverse · DOI 10.7910/DVN/XRUT8U
United States·Issue №14·Week of 2026-10-29 to 2026-11-05·published thursday 07:00 BRT

What the prediction market did

The five published prices settled with USD 96.22M traded in total, USD 55.47M of it new since the October 29 close: USD 34.61M in the House books, USD 20.50M in the Senate books. On the eve both pairs summed to 100.00%.

House · Democrats

100%

↑ +7.50pp on the week

Settled yes on November 4 at 13:34 UTC, on USD 37.37M. It closed October 29 at 92.50% and read 94.50% on the eve. After polls closed it was certified at 98.20% at 00:35 UTC on November 4.

House · Republicans

0%

↓ -7.50pp on the week

Settled no, on USD 25.93M. It closed October 29 at 7.50% and read 5.50% on the eve.

Senate · Democrats

100%

↑ +32.50pp on the week

Settled yes on the morning of November 4, on USD 17.13M. It closed October 29 at 67.50% and read 76.50% on the eve. On election night the book moved faster than two captures eight minutes apart could confirm, until they agreed at 98.65% at 01:52 UTC.

Senate · Republicans

0%

↓ -32.50pp on the week

Settled no, on USD 14.33M. It closed October 29 at 32.50% and read 23.50% on the eve.

The fifth price, on the election happening as scheduled, settled yes at 100% on USD 1.46M. It closed October 29 at 99.30% and November 2 at 99.70%.

What the polls did

The spread comes before the average. Counting every round that finished fieldwork by election day, the average is D+7.73 over 37 rounds from 27 polling firms, and the rounds inside it disagree by nine points.

Widest

Big Data Poll and Reuters/Ipsos · D+12

Narrowest

Targoz Market Research · D+3

Spread between institutes

9 points

The 29 likely-voter rounds average D+7.45 and the 8 registered-voter rounds D+8.75. Between the eve record and election day six rounds entered: John Zogby Strategies (D+6), Reuters/Ipsos (D+8), Focaldata (D+7), Big Data Poll (D+12), YouGov (D+9) and Rasmussen Reports (D+6). One left: the October 14 to 18 round by Hart Research and Public Opinion Strategies, whose question asked which party voters prefer to control Congress.

What the press reported

What the fixed-list outlets covered, October 30 to November 5

The Alaska lawsuit was dismissed before the vote. On November 1 CBS News, NBC News, The Hill and the Guardian reported that a federal judge dismissed the Justice Department's suit seeking to block certification of the Alaska Senate result if the Republican incumbent lost. On October 30 the New York Times reported Democrat Mary Peltola saying she did not know about the ballot-stacking effort, and Fox News reported Republican Senator John Barrasso alleging that Chuck Schumer and Peltola's campaign schemed to recruit a decoy candidate.

The final days were about turnout and about power after the vote. On October 31 the Guardian reported more than 50m Americans had voted early, and ABC News carried Kamala Harris urging a turnout too big to rig. On November 1 Reuters reported the White House mapping a path to govern without Congress if Democrats won, and on November 2 the New York Times reported Trump floating pre-emptive pardons for allies. On November 3 Reuters reported that prediction markets faced a test as election bets soared, and Axios reported voting-machine trouble in a New York House battleground.

From November 4 the subject was the result. CBS News reported the fallout after Democrats won control of Congress, the Guardian's live coverage carried Trump acknowledging the Iran war's unpopularity, and The Hill reported Hakeem Jeffries taunting Speaker Mike Johnson for having said the polls were wrong. On November 5 CBS News asked whether Republican redistricting had backfired, the Washington Post asked how accurate the midterm polls were, and NBC News reported Democrat Scott Colom criticizing his party for writing off Mississippi as the Senate count there stayed close.

The crossing of the week

The eve record, next to the result

On November 2 at 22:06 UTC we certified House Democrats at 94.50% and Senate Democrats at 76.50%, and published both beside the polling average, D+7.66. Democrats won both chambers. One outcome does not grade a price: 76.50% left roughly one chance in four for the other side, and a single election cannot say whether that was too much or too little. What the record does show is timing: Senate Democrats went from 67.50% at the October 29 close to 76.50% on the eve, before any result was known.

The Senate seat market held the band

At the November 2 close the Senate seat market's bands summed to 104.30%, inside our 95 to 105 gate, and put 57.50% on Republicans holding 47 seats or fewer. NBC News and ABC News have 44 Republican seats called, with Alaska and Mississippi still counting; if both go Republican the total is 46, still inside that band. The House seat market summed to 106.80% that day, outside the gate, so we do not use it.

The average waits for the official count

The polling average is a margin in national House vote share, and the national House vote is not fully counted: western states count for days, and states certify over weeks. We will not set D+7.73 or D+7.66 against a partial tally or a projection. The comparison waits for the official count.

How to read this number

What one election can and cannot say about a price

A control price is a probability, and the contract settles yes or no. On the eve the Senate price was 76.50%. Had Republicans kept the Senate, that fact alone would not have made the price wrong, and Democrats winning does not make it right. Prices are judged over many events, by checking whether things priced near 75% happen about three times in four. Fourteen issues hold one outcome per chamber, and a dated record, written before the vote, that anyone can check.

Sources

Prediction market, settled

Results, as called on November 5

Polling, primary sources of the rounds that entered after the eve

Press cited in sections 4 and 5

Method

Market prices run from the October 29 close, the last point recorded that day in our daily database backup, to settlement: the House contract settled on November 4 at 13:34 UTC, and the Senate and schedule contracts earlier that morning. The eve record is the reading certified on November 2 at 22:06 UTC by two captures eight minutes apart that agreed within 0.20pp; on election night the Senate book failed that test four times before passing at 01:52 UTC, and only certified readings are printed. New money is measured from the October 29 close to the final traded volume. The seat-market bands are read at the November 2 close and used only when they sum to between 95% and 105%. The polling average is a simple, unweighted mean of one round per polling firm per wave over a thirty-day window, with the hierarchy likely voters over registered voters over adults, and it counts only the question that offers party labels: rounds that name each respondent's district candidates, or that ask which party should control Congress, are listed in the data file and left out. From this issue the window ends on election day, November 3, so the average no longer moves as rounds age out. We never subtract a market price from a polling margin: one is a probability of controlling a chamber and the other is a margin in vote share.

AFOS
AFOS Analytics
Global Political Risk Intelligence · GLOBAL BY DESIGN
www.afos-analytics.com · contact@afos-analytics.com · github.com/AFOS-Analytics

Get it by email

AFOS Weekly follows the US midterms every Thursday, written for voters: where the prediction market, the polls and the press disagree, and what each one measures.

No spam. Unsubscribe anytime.

After signing up you choose whether to receive it in English, Portuguese or Spanish.

AFOS Weekly · Issue No. 14 · Democrats won both chambers, and here is what the market and the polls said the evening before | AFOS Analytics