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 №12·Week of 2026-10-15 to 2026-10-22·published thursday 07:00 BRT

What the prediction market did

The five published prices hold USD 28.77M and took USD 3.56M of new money since the October 15 close. Two thirds of it, USD 2.34M, went to the House books, which moved two points. The Senate books took USD 1.19M and moved four to five.

House · Democrats

92.50%

↑ 2.00pp on the week

On USD 11.77M, with USD 1.54M of new money. It held 90.50% for four closes, added a point on October 19 and another on October 22. The week's 44 readings ranged 2.00pp; the series high, 93.50% on September 20, is one point above.

House · Republicans

7.50%

↓ 2.00pp on the week

On USD 8.42M, with USD 0.80M of new money. The mirror: the pair sums to 100.00% at both ends, so the raw and normalised moves are the same.

Senate · Democrats

68.50%

↑ 4.00pp on the week

On USD 4.50M, with USD 0.62M of new money. It closed October 15 at 64.50%, dipped to 63.50% on October 19 and 20, then rose three points on October 21 and two on October 22, first touching 68.50% at 14:30 UTC. The pair opened summing to 101.00% and ends at 100.00%, so normalised the gain is 4.64pp.

Senate · Republicans

31.50%

↓ 5.00pp on the week

On USD 3.38M, with USD 0.56M of new money. It fell 5.00pp raw and 4.64pp normalised, to the lowest point the series has recorded since April 14.

The fifth price, on the election happening on schedule, reads 98.90%, down 0.20pp from the October 15 close of 99.10%, on USD 0.70M.

What the polls did

The spread comes before the average. The average is D+7.71 over 32 rounds from 25 polling firms, up 0.13pp from D+7.58 over 41 rounds a week ago, and the rounds inside it still disagree by ten points.

Widest

Reuters/Ipsos and Quinnipiac University · D+12

Narrowest

Harvard/Harris Poll/HarrisX · D+2

Spread between institutes

10 points

The ten points are mostly not time: Quinnipiac closed fieldwork on September 27 and Harvard/Harris on September 28, one day apart. Sorted by who was screened, the 20 likely-voter rounds average D+7.10 and the 10 registered-voter rounds D+8.80. That compares two groups of polls, not the same question twice.

What the press reported

What the fixed-list outlets covered, October 16 to October 22

The Senate map led the week, and the coverage pulled both ways. Kansas was described as competitive four times: by Politico on October 17, which called it "a sleeper Senate battleground", by the Washington Post on October 20, by NewsNation on October 21 and by the Hill on October 22. On October 18 CBS News called Texas a toss-up in its own poll. In the other direction, on October 20 the Washington Post reported Cook moving Minnesota's Senate race toward the Republicans, and on October 22 the Guardian asked whether Democrats could lose a Senate seat there.

Approval was the second thread. On October 18 the Washington Post reported Trump's approval at a record low, and on October 19 Reuters reported a record low of 31% in a Reuters/Ipsos poll. On October 21 the New York Times reported a wide Democratic lead in its new poll with Siena.

Money followed the map. On October 19 Axios reported a multimillion-dollar boost for Democrats' long-shot Senate races and NewsNation reported both parties shifting funds to close races; on October 20 the Guardian carried Trump saying he will have spent $400m to support Republicans.

And voting is under way. NBC News on October 16 and the New York Times on October 17 were tracking the early vote. On October 20 the Hill reported North Carolina's mail-ballot restrictions will stand through the election, and on October 22 Fox News reported a federal judge blocking voter-fraud probes in battleground states.

The crossing of the week

Five points in two days, on a third of the money

From the October 20 close to the October 22 reading, Senate Democrats rose 5.00pp raw and 4.36pp normalised, on USD 0.42M of new money in the Senate book. Over the same two days the House book took USD 1.21M and moved one point. More money did not buy more movement. A price moves when buyers accept a higher price, not in proportion to the dollars that change hands, and the two Senate books together hold less than half the House books' total volume.

The price moved one way, the Senate coverage two ways

In the week the Democratic Senate price set its series high, the Washington Post reported Cook moving Minnesota toward the Republicans, and four outlets wrote that Kansas had become competitive for the Democrats. We set no Senate polling average beside the price: only a third of the seats are up, each one is a state race, and no national question measures them all. The price and the coverage describe different objects, and we report both.

Three polls above the average, and it moved 0.13pp

Reuters/Ipsos (D+12), The New York Times with Siena (D+11) and CNN with SSRS (D+10) closed fieldwork on October 18 and 19, all above the D+7.71 average. The nine rounds that entered this week pushed the average up by 0.18 to 0.32pp; the 18 that left through the thirty-day edge pulled it down by 0.05 to 0.19pp. Each part is a range because it depends on the order of the calculation. Net: +0.13pp.

How to read this number

When some votes are already in, a poll is part count and part forecast

By late October some ballots are already cast. The CNN poll published on October 21 found about 13% overall saying they had already voted, breaking 70% to 29% for Democrats, while those yet to vote split 46% to 41%. Both groups sit inside the topline. An early-vote tally counts who has voted so far, not who will vote, so it is not a forecast of the result. The reading for a voter: when a poll or a tracker cites early votes, ask which of the two groups the number describes.

Sources

Prediction market, October 22 reading

Polling, primary sources

Press cited in sections 4 and 5

Method

Market prices run from the October 15 close, the last point recorded that day in our daily database backup, to the latest reading on October 22, certified at 20:56 UTC by two captures eight minutes apart; October 22 has not closed. Issue No. 11 printed 63.50% for Senate Democrats, the last point recorded before it went out; the day closed at 64.50% at 23:30 UTC, and this issue measures from that close. New money is measured over the same window. 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. 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 when a firm publishes more than one row in the aggregating index. The split of this week's polling change between new rounds and rounds leaving the window is computed both ways round, which is why each part is a range. 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, and in 2012 Democrats won more House votes and fewer seats.

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. 12 · The Senate price set its series high on a third of the House's new money | AFOS Analytics