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 №13·Week of 2026-10-22 to 2026-10-29·published thursday 07:00 BRT

What the prediction market did

The five published prices hold USD 43.03M and took USD 11.76M of new money between the October 22 and October 29 closes. The House books took USD 8.40M and ended level; the Senate books took USD 2.98M and went four points down and back.

House · Democrats

92.50%

→ 0.00pp on the week

On USD 17.03M, with USD 4.39M of new money. It held 91.50% from October 23 to 27, and on October 29 touched 93.50%, equal to the series high first reached on September 20.

House · Republicans

7.50%

→ 0.00pp on the week

On USD 13.88M, with USD 4.01M of new money. It moved between 7.50% and 8.50% and on October 29 touched 6.50%, equal to the series low. The pair sums to 100.00% at both ends.

Senate · Democrats

68.50%

→ 0.00pp on the week

On USD 6.36M, with USD 1.75M of new money. From 68.50%, the series high since April 14, it lost a point a day to a 65.50% close on October 25, held it three days, and regained a point a day.

Senate · Republicans

31.50%

→ 0.00pp on the week

On USD 4.65M, with USD 1.23M of new money. The mirror, from a low of 30.50% on October 23, the lowest since April 14, to 35.50% on October 26. The pair sums to 100.00% at both ends.

The fifth price, on the election happening on schedule, reads 99.45%, up 0.55pp from the October 22 close of 98.90%, on USD 1.11M. On October 29 it touched 99.50%, the highest point since its series began on July 29.

What the polls did

The spread comes before the average. The average is D+7.64 over 33 rounds from 23 polling firms, against D+7.61 over 31 on October 22 under the same rule, and the rounds inside it disagree by nine points.

Widest

Reuters/Ipsos · D+12

Narrowest

Rasmussen Reports · D+3

Spread between institutes

9 points

Part of the nine points is time. The 15 rounds that closed fieldwork on or after October 16 average D+8.00; the 18 earlier ones average D+7.33. Part is who was screened: the 23 likely-voter rounds average D+7.43 and the 10 registered-voter rounds D+8.10. Both compare groups of polls, not one poll measured twice.

What the press reported

What the fixed-list outlets covered, October 23 to October 29

Alaska ran through the week. On October 23 the New York Times, Reuters, the Washington Post and the Guardian reported a Justice Department lawsuit seeking to block certification of the Senate result if the Republican incumbent, Dan Sullivan, loses. On October 24 USA Today carried Democrat Mary Peltola calling the suit "corrupt"; on October 25 ABC News, NPR and the Guardian reported Blanche, for the Justice Department, defending it. On October 28 Fox News, and Reuters citing the New York Times, reported allegations that Democrats recruited spoiler candidates, and on October 29 the Hill reported a House Republican probe and Politico an investigation ordered by Alaska's governor.

The Senate map and the accuracy of polls came second. On October 24 the New York Times asked why Democrats could find it hard to win the Senate despite strong polls; on October 27 it reported that polling bias varies by state; on October 29 CNN wrote that the range of Senate outcomes is enormous. USA Today reported on October 26 that two polls disagree on how close the Texas Senate race is, and on October 28 the Guardian called South Carolina's a toss-up.

Money and turnout framed the final week. On October 27 NBC News reported the most expensive cycle ever, and on October 24 NewsNation reported early voting surging in battleground states. Gallup, reported by the Hill, found support for reelecting congressional incumbents at record lows on October 28, and the economy, healthcare and the Iran war as top concerns on October 29.

The crossing of the week

Four points down and four back

Senate Democrats fell from 68.50% to a 64.50% low on October 26 and returned to 68.50%; the Republican side ranged from 30.50% to 35.50%. The low came three days after the Alaska lawsuit was reported and two days after the New York Times asked why strong polls might not deliver the Senate. We print the timing: a price records that buyers changed their bids, not why.

The widest leads asked a different question

Tufts with YouGov (D+12 among likely voters, fieldwork October 23 to 27) and The New York Times with Siena (D+11, October 12 to 19) put the candidates' names in the question. So did The Economist with YouGov, whose October 23 to 26 wave ran D+11 or D+9 among likely voters, depending on the screen. Our average counts the party-label question only: D+7.64. One question asks about a party, the other about two named people, and a known name can move a voter either way. The named-ballot rounds are in our data file, outside the average.

On schedule, and a fight over certification

The contract on the election happening as scheduled reached 99.50% on October 29, its series high, in the week a lawsuit sought to block certification of one Senate result. They are different objects: the contract asks whether the vote takes place on November 3, not whether every result is certified afterwards.

How to read this number

What each of these numbers will be checked against after November 3

This is the last issue before the vote; next week's sets each instrument against the result, each with its own yardstick. A control price settles on which party controls the chamber, yes or no, so a price of 68.50% is not wrong if the other side wins: prices are judged across many outcomes, by how often the 70% events happen about 70% of the time. The polling average is a margin in national House vote share, and its yardstick is the national House vote, which takes weeks to count. Neither yardstick is the other's: in 2012 Democrats won more House votes and fewer seats.

Sources

Prediction market, October 30 reading

Polling, primary sources

Press cited in sections 4 and 5

Method

Market prices run from the October 22 close, the last point recorded that day in our daily database backup, to the latest reading, certified on October 30 at 00:28 UTC (October 29, 8:28 pm Eastern) by two captures eight minutes apart. New money is measured from the October 22 close to the October 29 close. Both ends of each binary pair are printed raw and normalised, because a pair that does not sum to 100 moves one side without the other side trading. Series highs and lows are checked against the full series, which begins on April 14 for the Senate and July 29 for the House and the schedule contract. The polling average is a simple, unweighted mean of one round per polling firm per wave over a rolling 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 are listed in the data file and left out of the average; the New York Times/Siena round entered that list on October 22, after Issue No. 12 had printed D+7.71 with it inside, so this issue compares against D+7.61, the October 22 average under the same rule. 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. 13 · The Senate price went four points down and came back, and the widest poll leads came from a different question | AFOS Analytics