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 №4·Week of 2026-08-20 to 2026-08-27·published thursday 07:00 BRT

What the prediction market did

The eight midterm contracts hold USD 18.21M, up from USD 17.84M last Thursday. Every move on the week happened on the Republican side of a book.

House · Democrats

88.50%

→ 0.00pp on the week

On USD 5.56M, the deepest book here. This is the ceiling of the series, never exceeded, and it dipped to 87.50% on August 25 and 26 before returning.

House · Republicans

12.50%

↑ 1.00pp on the week

On USD 4.25M. It moved up on August 21 and held there for the rest of the window.

Senate · Democrats

50.50%

→ 0.00pp on the week

On USD 2.38M. It spent August 23 to 25 at 49.50% and came back on August 26.

Senate · Republicans

50.50%

↑ 1.00pp on the week

On USD 1.65M, the thinnest of the four control books. It sat at 49.50% for eight days and moved on August 27.

The fifth published price, on whether the election happens on schedule, reads 96.30%, up 0.55pp, on USD 0.37M. The House series opens on July 29, 2026 and the Senate on May 30, 2026, so every superlative here means since those dates.

What the polls did

The disagreement comes before the average, because it is the finding. Seventeen polls from thirteen institutes sit inside the thirty-day window, averaging D+6.00. The newest fieldwork in the file ended on August 17.

Widest

John Zogby Strategies · D+11

Narrowest

Morning Consult · D+4

Spread between institutes

7 points

Zogby surveyed 921 likely voters through August 5. Morning Consult surveyed 29,719 registered voters through August 16, the largest sample here, and sits at the bottom: size does not buy agreement. The sharpest comparison is a single day, when four houses finished fieldwork on August 17 and came back Emerson College D+8 on 1,000 likely voters, against Echelon Insights D+5 on 1,002 likely voters, and The Bullfinch Group and Reuters/Ipsos at D+5 on registered voters. Emerson and Echelon both screened for likely voters, so the gap is not who was asked.

What the press reported

What the fixed-list outlets covered, August 21 to August 27

The week had one actual election. CBS News reported on August 26 that Senator Darline Graham won South Carolina's Republican Senate runoff against Ralph Norman, to fill the seat of the late Senator Lindsey Graham. ABC News reported on August 25 that MAGA Inc. spent $827,000 backing her on the final day; the New York Times on August 21 had President Trump telling supporters to pretend he was on the ballot.

Alaska produced the week's oddest development. Fox News and ABC News both reported on August 26 that a second Dan Sullivan, a Republican sharing the incumbent senator's name, advanced from the primary and will appear on the same November ballot.

Two fights moved into the courts. ABC News reported on August 26 that Democratic states sued to block the President's order limiting mail voting, and on August 27 that a Maryland court barred a Democratic-backed redistricting amendment from the November ballot. Fox News reported the same day that a redistricting push in another Democratic-led state hit a legal snag.

Underneath, two structural threads: ABC News on August 27 on Republicans facing a midterm verdict as the Iran war runs longer than promised, and the Washington Post on August 21 on Democrats trailing far behind in cash on hand.

Our measurement: Both hold, on different clocks. This week the market did not agree: Democratic Senate control finished at 50.50%, where it opened on Thursday. Across the season it agrees plainly, having traded as low as 39.50% since May 30. An improvement can be real on the scale of months and absent on the scale of days.

The crossing of the week

When both sides get more expensive, nobody changed their mind

Each chamber has two separate contracts with separate order books, one on each party controlling it. A week ago each pair added to exactly 100.00%. Today the House pair reads 88.50 plus 12.50 and the Senate 50.50 plus 50.50, both 101.00%. The Republican side rose a point in each chamber and the Democratic side did not move. A genuine shift toward Republicans would show the Democratic price falling, because those are the same event seen from two sides. Instead the point was added, not transferred. What moved is the distance between two books.

The week had a real election, and the national prices did not blink

South Carolina voted on August 25 and was called on August 26. Neither Democratic control price moved on either day. A Republican runoff for a Republican-held seat in a Republican state settles who carries the party's banner, not which party holds the chamber. It is the third consecutive week in which a primary resolved and the control contracts stayed put.

The poll this brief said was missing has arrived, and it is the outlier

Last week's edition declared that Emerson College and The Hill had fielded a national congressional ballot on August 16 and 17, that it was absent from the index we read, and refused to type it in by hand, because a hand-added number cannot be reproduced by the next reading. It is now in the file, collected the ordinary way: D 51% x R 43%, a thousand likely voters. It is also the highest of the four houses that finished that day. Waiting cost a week and produced a number the reader can check.

How to read this number

Two contracts, one outcome, and the gap between them

A pair of contracts on opposite sides of the same event should add to 100%. The extra point in a 101% pair is not a forecast: it is what it costs to hold both sides at once, and it grows when the two books stop agreeing. That is also why we write that the market is paying 88.50% rather than that the probability is 88.50%. The first observes a price. The second is a claim about the world.

Sources

Prediction market

Polling, primary sources

  • Index of published polls · The index lists which polls exist; every number stays attributed to the institute that produced it, en.wikipedia.org/wiki/2026_United_States_elections
  • Emerson College Polling · The national congressional ballot fielded August 16 and 17, declared as missing in issue No. 3 and read from the index for this one, emersoncollegepolling.com/
  • Morning Consult · The 29,719-respondent survey fielded through August 16 that sits at the bottom of this week's range, pro.morningconsult.com/

Press, outlets cited in sections 4 and 5

Method

AFOS reads three signals and does not merge them: prediction market prices, published polls, and press coverage. The house average of the generic ballot is a simple mean, unweighted, over a rolling thirty-day window, and when one institute publishes more than one cut of the same round only one enters, by the hierarchy likely voter, registered voter, adults. Market prices come from a certified capture, two readings eight minutes apart that must agree inside 0.20pp before anything is published. Band markets are published only when the bands sum between 95% and 105%; the popular-vote-margin market is currently above 150% and is collected daily for the series without being published as a reading. This brief never subtracts a market probability from a polling lead, because they are different quantities. Open source, Apache 2.0.

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. 4 · Republicans got a point more expensive in both chambers, and Democrats did not get cheaper | AFOS Analytics