AFOS Analytics
Real-time political pricing
Prediction markets × polls × news
no smoothed averages
United StatesThe week of August 17 to 21 produced actual votes, and the two control contracts ended where they began. Florida and Alaska voted on Aug 18, with Angie Nixon winning the Democratic Senate primary in Florida and Dan Sullivan and Mary Peltola advancing through Alaska's nonpartisan primary. In the House contract, Monday's close and Friday's close are identical, 87.50% for the Democrats and 12.50% for the Republicans, 0.00pp on both sides. In the Senate the only move is on the Republican side, +1.00pp, from 48.50% to 49.50%.
What did move over the week was the Senate's other book, and it moved against the binary. In the seat distribution, Republican control is the sum of the 50-50 tie and the majority at 51 or more, and that sum, normalised, went from 47.29% at Monday's close to 47.13% at Friday's, while the binary rose. The distance between two prices for the same outcome went from 1.21pp to 2.37pp in five sessions, and in Monday's confirmed reading it stands at 2.77pp. That is three measurements in a row, always in the same direction.
The largest single-band move of the week sits inside that story: the bucket for exactly 51 seats gave up 2.50pp, from 12.50% to 10.00%. That is the narrowest-majority scenario, and the money left it for 52 (+1.90pp) and 54 (+1.30pp). The market did not change its mind about who controls; it changed its mind about by how much.
⚠️ The house average on the generic ballot stands at D+5.82pp, across 11 polls from 9 institutes, and it is against that base that this issue's price is read.
1.Executive Summary
For a desk, the week reads as follows: the event arrived and the control price did not respond. Two state primaries resolved, with a result the coverage treated as a surprise in Florida, and the probability of House control closed all five sessions in the same place. That does not mean the market ignored the ballot box: it means those two races were not among the ones that decide control, and the price already embedded that.
What changed sits one layer down, in the Senate seat distribution. There the money left the one-seat-majority scenario for the more comfortable ones, and the control aggregate implied by the distribution moved in the OPPOSITE direction from the binary contract. It is the second week in which the two disagree, and the distance between them grew across all three measurements we have.
2.Why this brief does not subtract the market from the polls
This section exists because the most tempting calculation on this page is the one that cannot be made. The control contract and the generic ballot measure different things, and the difference is not one of scale or precision: it is one of quantity.
<strong>Why it matters:</strong> 88.50 minus 5.82 is 82.68, and that number does not exist. It has no unit, it would change size if the poll were expressed on another scale, and it corresponds to nothing that can be checked on November 3. Translating votes into seats runs through 435 districts with different shapes, and in 2012 the Democrats won more House votes and fewer seats. The AFOS panel leaves that edge MUTE on purpose, and this brief does the same.
The legitimate question is not the subtraction, it is coherence: is a market paying 88.50% on Democratic control compatible with a 5.82-point lead at the ballot box? The honest answer is that it depends entirely on geography, and the market publishes its own answer in another contract, the seat distribution, which today puts 25.50% on the band below 190 Republican seats. It is that contract, and not the subtraction, that translates votes into seats.
3.Weighted scenarios for the week
The three scenarios below are not narrative: they are a direct read of the Senate seat distribution, the contract in which the market publishes the probability of each band. The bands sum to 99.50%, inside the 95 to 105 gate that AFOS requires before publishing any distribution. The percentages are the market's, not the writer's, and they are not a recommendation.
The Republicans end with 49 seats or fewer. It is the sum of three bands: 26.50% at 47 or fewer, 13.00% at exactly 48 and 13.50% at exactly 49. This aggregate has been FROZEN at 53.00% since the Aug 23 reading, and it was the only one of the three that did not move over the week. In that case the Senate changes hands and the House, which the market already gives to the Democrats at 88.50%, follows.
The Republicans end with exactly 50 seats and the tiebreak goes to the vice presidency. This is the scenario that makes the binary and the distribution disagree, because it counts as Republican control in the binary and appears as an isolated band in the distribution. It gave up 0.50pp over the covered week and another 0.50pp since. One in four Republican winning scenarios runs through here, not through an outright majority.
It is the sum of the bands at 53 or more, which is the number of seats the Republicans hold today. The market pays 86.45% on them losing at least one. Two of those bands concentrate money disproportionately: exactly 54 seats pays 4.05% and carries USD 0.78M of accumulated volume, more than the 52-seat band, which pays 10.95% with USD 0.60M.
4.Indicator Grid
| Contract | Current | Δ week | Vol USD acc. | Implied reading |
|---|---|---|---|---|
| House, Democratic control | 88.50% | 0.00pp | USD 5.49M | Closed all five sessions of the week at 87.50%. The 88.50% is from Aug 24 and the gain came over the weekend, outside the covered window. |
| House, Republican control | 11.50% | 0.00pp | USD 4.22M | Mirror of the previous one. Closed the week at 12.50%, the same as Monday, despite the Aug 18 primaries. |
| Senate, Republican control | 49.50% | +1.00pp | USD 1.65M | The only binary that moved over the covered week, from 48.50% to 49.50%. It sits in an exact tie with the Democratic contract. |
| Senate, Democratic control | 49.50% | 0.00pp | USD 2.37M | Closed the week at 50.50% and gave up 1.00pp after it. Carries more volume than the Republican contract in the same market. |
| Senate, exactly 51 Republican seats | 10.50% | -2.50pp | USD 0.31M | Largest single-band move of the week. The one-seat-majority scenario lost weight to 52 (+1.90pp) and 54 (+1.30pp). |
| Senate, 50-50 tie | 11.50% | -0.50pp | USD 0.29M | The band the binary counts as Republican control and the distribution isolates. It is the arithmetic source of the divergence between the two books. |
| House, fewer than 190 Republican seats | 25.50% | n/a | USD 0.08M | Leading band of the House distribution. The ten bands sum to 103.90%, inside the 95 to 105 gate. |
| Election happens as scheduled | 96.05% | 0.00pp | USD 0.36M | Calendar contract, stable. It is the institutional floor against which everything else in this grid is read. |
| Generic ballot, AFOS average | D+5.82pp | -0.10pp | 11 polls, 9 institutes | ⚠️ The fall is COMPOSITION, not voting intention: the 30-day window rolled, one poll dropped off the edge and none came in. |
5.Liquidity and market structure
6.Electoral calendar through November
| Date | Sample | Why it matters | |
|---|---|---|---|
| Aug 25, Tuesday | Republican runoff in South Carolina | 1 Senate seat | Ralph Norman against Darline Graham for the nomination to the late senator's seat. Mark Sanford endorsed Norman on Aug 21. |
| Sep 15 | Delaware closes the primary calendar | last state | From then on the candidate field is set in every state, and the market moves to pricing a defined contest. |
| Nov 3 | General election | 435 House seats and 1/3 of the Senate | The date on which every contract in this grid resolves. Until then AFOS measures the distance between price and poll, without arbitrating between them. |
There is no mandatory poll registry in the United States, so this calendar covers ELECTORAL EVENTS and not release dates. The cadence of the institutes is each house's own decision, which is why the generic ballot base can go days without a new entry with nothing having failed. The Aug 18 primaries in Florida and Alaska have resolved and leave the calendar in this issue.
7.Watch list, week triggers
- The distance between the Senate's two books, today at 2.77pp three measurements in a row in the same direction, from 1.21pp on Aug 17 to 2.37pp on Aug 21 and 2.77pp now. If a fourth confirms, it stops being thin-book oscillation and becomes a feature of the pair.
- The exactly-51-seats bucket, which gave up 2.50pp over the week that is the one-seat-majority scenario. Continuing to give ground while 52 and 54 rise means the market is pricing margin, not outcome.
- The South Carolina runoff on Aug 25 it sets the Republican nominee for a Senate seat. It is the next ballot-box event before Sep 15, and the only one of the week.
- The next entry into the generic ballot, and what it does to the average the D+5.82pp average is simple, unweighted, over a 30-day rolling window. A new entry moves the number, and so does one dropping off the edge: this issue's -0.10pp variation is entirely window composition.
- The popular vote margin contract, which remains REJECTED its fourteen bands sum to 152.35%, far outside the 95 to 105 gate. It is the only market that would measure the same quantity as the poll, which is why it is collected every day even while rejected, to build the series. It is never published as if it counted.
8.Methodology
AFOS reports the three signals separately: prediction market price, generic ballot polls with sample and fieldwork dates, and press coverage from a fixed list of outlets. It does not build a composite index and it does not subtract one quantity from another.
Every published price goes through two independent readings eight minutes apart, and is released only when they agree within 0.20pp. This issue uses the reading of Aug 24, 1:24 PM BRT, with an approved verdict. The weekly variations use each day's CLOSE, not the first and last point of the window, because the second choice changes the number without changing the fact. The House series starts on Jul 29, 2026, when collection was switched on, so any superlative about it is always 'of the series', never 'of the cycle'.
A distribution is only published if its bands sum to between 95% and 105%. In this issue the Senate sums to 99.50% and the House to 103.90%, both inside. Accumulated USD volume is cited as context and is not guidance; book depth is not cited, by editorial decision of May 21, 2026. Code and method are open under Apache 2.0.
9.Additional reading · macro coverage
AFOS is the primary source for this analysis. The references below are complementary context and do not support any of the measurements above.
- The Associated Press · Coverage of the Aug 18 Florida and Alaska primaries, including Dan Sullivan and Mary Peltola advancing through the nonpartisan primary., apnews.com/hub/us-elections
- Politico · Readings of the primaries in three states and of the nomination picture for the Senate races., www.politico.com/news/elections
- The Hill · Tracking of congressional ballot polling and of the fundraising picture for both campaigns., thehill.com/homenews/campaign/
- Cook Political Report · District-by-district ratings, the reference for the votes-to-seats translation this brief does not perform., www.cookpolitical.com/ratings
- FEC, campaign finance data · Official record of campaign fundraising and spending, an institutional counterpoint to the money coverage cited during the week., www.fec.gov/data/
Get it by email
AFOS Tradeoff is the weekly read on Brazilian political risk, for people who decide on data rather than headlines.
No spam. Unsubscribe anytime.
After signing up you choose whether to receive it in English, Portuguese or Spanish.