AFOS Analytics
Real-time political pricing
Prediction markets × polls × news
no smoothed averages
United StatesThe previous issue closed with the House pair summing to 101.00% at Monday's certified reading and said what to watch: if the sum persisted, the House delta would come to need two measures. It did not persist. The pair went back to 100.00% that same night and closed both edges of this week at that value, so the House remains the book in which raw and normalised say the same number. The contract on Democratic control of the House rose 1.00pp, from 90.50% to 91.50%, with a top of 92.50% on 22 October at 08:00 UTC, the highest since 7 October. The signal of the week of 19 to 23 October is in the Senate, and it has two parts. The first is about the series: the contract on Democratic control of the Senate touched 68.50% on 22 October at 14:30 UTC, and the contract on Republican control touched 30.50% on 23 October at 06:30 UTC. Neither value had appeared before in the series recorded since 14 April: one is the top and the other is the floor of the whole of it, and both fell in the same week. Friday's close stood at 67.50% and 32.50%. Here the two measures disagree, because the pair sum went from 101.00% to 100.00%: raw, the Democratic side rose 3.00pp and the Republican side fell 4.00pp, and normalised both moved exactly 3.64pp, one in each direction. The raw figure makes it look as if the Republican side lost more than the Democratic side gained; the implied probability says it was the same size. The second part is about shape. The market publishes two books on control of the Senate, the binary and the distribution of Republican seats, and in the previous issue the two House books had done opposite things, one frozen and the other reorganising. The two Senate books moved the same way. The band of 47 Republican seats or fewer, which was already the most expensive in the book, went from 37.13% to 40.63% of its mass, and the share of the bands in which Republicans end up with 49 seats or fewer, without a majority even with the Vice President's tie-break, went from 65.09% to 68.36%. The seats book sums above 100% and is therefore read for its shape and never for its level, and no number from it is compared with the binary's price. What is being asserted is the direction: two books on the same question moved the same way in the same week, and the seats book took in USD 0.22M of new money over the period.
1.Executive Summary
The three numbers above are of different natures and do not combine. The first two are contract prices, read live and confirmed by a double reading 8 minutes apart, with a tolerance of 0.20pp. The third is an arithmetic average of published polls, in a 30 day rolling window. This issue measures how each of them behaved in the week of 19 to 23 October, with the delta anchored to the previous Friday's close, 16 October, and it refuses the subtraction between them.
2.Why this brief does not subtract the market from the poll
This week the trading desk question comes through the Senate, and it comes with a record: if the market got to pay 68.50% for Democratic control of the Senate, the highest value in the series, and the poll average for the House gives a Democratic advantage of 7.60 points, do the two readings agree? The answer is still that the question has no numerical answer, for two reasons. The first is one of units, and it holds for either chamber. The second is specific to the Senate: there is no national poll that measures the vote for it, because only a third of the seats are up and each one is a state race.
Subtracting 7.60 from 91.50 would produce the number 83.90, which is not a probability, is not a vote point and would change in size if the poll were expressed on another scale, for example as a share of the two party vote. The 2012 case is the short historical example: Democrats won more House votes and fewer seats, because converting votes into seats depends on geography and on district design. This brief reports the two readings side by side and never the difference between them.
What can be compared this week, and it is not a subtraction either, is the behaviour of two books from the SAME market on the Senate: the binary on control and the distribution of Republican seat bands. They are not the same measure and this house does not convert one into the other, because the band book persistently sums above 100% and therefore serves to discuss the SHAPE of expectations, never to extract a level. In the previous issue the two House books disagreed in behaviour. In this one, the two Senate books agreed in direction, and that is what the scenarios section measures below.
3.Weighted scenarios for the week
The scenarios below are not narrative: they are read in the prices of contracts the market already publishes. In this issue they use the Senate control binary, the distribution of Republican Senate seat bands and the calendar contract. Every number from the distribution comes from the close of 23 October, when the book summed 104.60%, and from the edge of 16 October, when it summed 103.70%, both inside the 95% to 105% gate, and the bands are cited as a SHARE OF THE BOOK, normalised by the sum at that edge. 🔑 The rule of method: a band book summing above 100% carries overpricing, and this house measured across 671 captures of this book that the overpricing is approximately uniform across the bands. Uniform excess means that normalising preserves the internal PROPORTION, and that is why shape can be discussed. It does not mean the normalised share is a calibrated probability, and that is why no number from this distribution is compared with the binary's price. In the Senate the threshold does not fall inside a band, as it does in the House: a Republican majority is 50 seats or more, because the Vice President breaks ties, and 50 is a band of its own in this book, so all of the mass can be assigned to one side or the other.
The binary on Democratic control of the Senate closed the week at 67.50%, against 64.50% at the close of 16 October, and touched 68.50% on 22 October at 14:30 UTC, the highest value in the series recorded since 14 April. The path is not monotonic: it fell first, to 63.50% on 19 October at 18:30 UTC, oscillated between 63.50% and 64.50% for two days and rose in four steps between 21 October at 16:00 UTC and 22 October at 14:30 UTC. In the seats distribution, the share of the bands in which Republicans end up with 49 seats or fewer went from 65.09% to 68.36% of the book's mass, an advance of 3.27pp of share, and the advance was concentrated in the band most comfortable for the Democratic side: the band of 47 Republican seats or fewer went from 37.13% to 40.63%, with 38.50% and 42.50% in raw price, and touched 44.50% on 23 October at 04:30 UTC. The bands of 48 and 49 seats, together, stood practically still, from 27.96% to 27.72% of the mass. In words: the seats book not only concentrated expectations on the Democratic side, it concentrated them on that side's widest result.
The binary on Republican control of the Senate closed at 32.50%, against 36.50% at the close of 16 October, and touched 30.50% on 23 October at 06:30 UTC, the lowest value in the whole series recorded since 14 April. It did not stay there: an hour later it was at 31.50%, and at 09:30 UTC the same day at 32.50%, the closing value. In the seats distribution, the share of the bands of 50 Republican seats or more, which is the Republican majority with the Vice President's tie-break, went from 34.91% to 31.64% of the book's mass. It gave ground in both parts: the band of exactly 50 seats, the tie, from 12.05% to 10.99%, and the bands of 51 or more, from 22.85% to 20.65%. This is a change of shape, measured in the share of a book carrying overpricing, and not a second price for the same outcome: it says which way the seats book's expectations moved, and not by how much the probability of control changed.
The contract on the election happening on schedule closed the week at 97.70%, 1.30pp below the 99.00% of 16 October, which takes the complementary outcome from 1.00% to 2.30%. Until Friday afternoon it had spent the week between 98.80% and 99.10%. The whole fall happened in five hours on 23 October: 99.10% at 18:30 UTC, 98.40% at 20:00 UTC and 97.70% at 23:30 UTC, the lowest value in the series since 18 September. It is still the lowest volume book of the five this brief follows, with USD 0.08M of new money in the week.
4.Indicator Grid
| Contract | Current | Δ week | Vol USD acc. | Implied reading |
|---|---|---|---|---|
| House, Democratic control | 91.50% | +1.00pp | USD 13.30M | It rose from 90.50% to 91.50% on 19 October at 17:30 UTC, touched 92.50% on 22 October at 08:00 UTC, the highest since 7 October, and returned to 91.50% on Friday. The pair summed to 100.00% at both edges: raw and normalised say the same. New money in the House book over the period: USD 3.05M. |
| House, Republican control | 8.50% | -1.00pp | USD 9.89M | The exact mirror of the Democratic side, with the pair at 100.00% at both edges: 9.50% until 19 October, 8.50% after, a floor of 7.50% on 22 October at 08:00 UTC. The series floor is still 6.50%, from 20 September. |
| Senate, Democratic control | 67.50% | +3.00pp raw, +3.64pp normalised | USD 5.36M | Top of the whole series recorded since 14 April at 68.50%, touched on 22 October at 14:30 UTC. Week's floor at 63.50%, on 19 October at 18:30 UTC: an amplitude of 5.00pp across 55 readings. The pair sum went from 101.00% to 100.00%, and that is why the raw figure sits 0.64pp below the normalised advance. |
| Senate, Republican control | 32.50% | -4.00pp raw, -3.64pp normalised | USD 3.86M | Floor of the whole series recorded since 14 April at 30.50%, touched on 23 October at 06:30 UTC. Here the raw figure OVERSTATES the fall by 0.36pp, for the same reason that makes the Democratic side look smaller: the pair sum lost 1.00pp over the period, and the side that closed it was the Republican one. |
| Election happens on schedule | 97.70% | -1.30pp | USD 0.81M | The lowest value in the series since 18 September, at Friday's last point, 23 October at 23:30 UTC. The whole of the week's fall happened in its final five hours, from 99.10% at 18:30 UTC. |
| Republican Senate seats, distribution of 11 bands | sum 104.60% at the close | from 103.70% on 16 October | USD 3.40M | Inside the 95% to 105% gate at both edges of the week, with a clean partition across the eleven bands, no overlap and no gap. It is the more heavily traded of the two seat books this brief follows, with USD 0.22M of new money in the week, and the band of 47 Republican seats or fewer remains its most expensive, now with 40.63% of the mass. |
5.Liquidity and market structure
6.Electoral calendar through November
| Date | Sample | Why it matters | |
|---|---|---|---|
| 3 Nov, Tuesday | Midterm election day | 435 House seats and a third of the Senate | It is the only date on this grid that resolves the contracts, and it is eight days from this issue. Every price cited here is the implied probability of an outcome that becomes known on that day. |
| 19 Oct, Monday | Reuters/Ipsos closes fieldwork with a 12 point Democratic advantage among registered voters | 1,279 registered voters, fieldwork 16 to 19 October | D 47% x R 35%, with 19% choosing neither, and it is the round most favourable to the Democratic side within the average's window, tied with the September Quinnipiac. On the same day the binary on Democratic control of the Senate fell to the week's floor, 63.50%, at 18:30 UTC. The two measure different quantities, and the brief does not link one to the other. |
| 20 Oct, Tuesday | Judge keeps North Carolina's mail-in ballot restrictions in place through the election | 1 state, voting rules | It is a voting rule in a state with a Senate race, and not a price event. None of the five contracts in this brief moved more than 1.00pp that day. |
| 21 Oct, Wednesday | The New York Times and Siena publish the October national survey, with the named ballot | likely voters, fieldwork 12 to 19 October | D 52% x R 41%, but with the NAMES of the candidates in each respondent's district, which is a different question and is therefore kept out of this brief's average, with the pollster's own wording on record. On the same day the binary on Democratic control of the Senate took two of the four steps of its rise, at 16:00 and at 20:30 UTC. The series does not allow saying which market reading came before the publication, and the brief records and does not attribute. |
| 23 Oct, Friday | The Justice Department asks for certification of Alaska's Senate race to be blocked if the Republican Dan Sullivan loses | 1 state Senate race | The report came out at 19:06 UTC. The contract on the election happening on schedule was at 99.10% at 18:30 UTC, went to 98.40% at 20:00 UTC and to 97.70% at 23:30 UTC. That is the order of the timestamps, and order is not cause: the lawsuit concerns the certification of a state result, and the contract concerns the election happening on the scheduled date. The Senate binary closed the day where it stood at 09:30 UTC. |
The grid above has one date that resolves a contract, 3 November, and four that resolve none. The four are included because they are verifiable and dated, and they come with the measurement alongside. The case of 23 October is the most instructive of the week precisely because the order of the timestamps is clear and, even so, it does not support the causal reading: the contract that moved is not the one the news describes.
7.Watch list, week triggers
- The calendar contract, at 97.70% at the week's close It spent the week between 98.80% and 99.10% and gave back 1.40pp in five hours on Friday night. At this Monday's certified reading, 26 October at 04:05 UTC, it stands at 98.05%: it has recovered part of the fall and remains below the range in which it spent the week. If it returns to 99%, Friday's fall will have been an event of hours in a low volume book; if it stays below, the quietest of the five books becomes the one that moved most in the last week before the vote. That is the difference the next issue measures.
- The Senate binary, at 64.50% at this Monday's certified reading The week's advance was handed back in full between Friday's close and the certified reading of 26 October at 04:05 UTC: Democratic control of the Senate stands at 64.50%, the same value as the close of 16 October, with the pair at 100.00%, and Republican control at 35.50%. The top of the series is still the 68.50% of 22 October, and the Republican side's floor the 30.50% of 23 October. The question for the next issue is whether the seats book also hands back the mass it moved into the band of 47 seats or fewer, that is, whether the two Senate books also move together on the way back.
- The band of 47 Republican Senate seats or fewer, at 40.63% of the book's mass It is the book's most expensive band and the one that absorbed the week's advance, while the bands of 48 and 49 seats stood practically still. If it keeps growing, the seats book's expectations concentrate on a wide Democratic result, and not only on a majority. If it hands mass back to 48 and 49, the book keeps its direction and loses breadth.
- The Senate pair sum, at 100.00% at Friday's close Last week it went from 99.00% to 101.00%, and this week from 101.00% to 100.00%: in both, the side that closed the sum was the cheap one, and the result was a raw figure that measures the move poorly. While the sum swings 1.00pp a week, the Senate delta needs both measures, and the normalised one is the one that gives the size of the move.
- The 10.00pp dispersion across pollsters in the generic ballot window Within the same 30 day window, Reuters/Ipsos measured a Democratic advantage of 12.00 points with fieldwork ending on 19 October, and Harvard/Harris measured 2.00 points with fieldwork ending on 28 September among likely voters. The two rounds with fieldwork inside this issue's week are 6.00 points apart: 12.00 from Reuters/Ipsos and 6.00 from ActiVote, ending on 21 October. They are two very different instruments, and the difference starts with the undecided: the first leaves 19% outside the two parties, the second leaves none.
8.Methodology
Prices read on AFOS's own route with the country parameter and an uncached reading, and confirmed by the capture lock: two readings 8 minutes apart, released only if they agree within 0.20pp, with the two timestamps required to be distinct. This issue's certified reading is from 26 October 2026 at 04:05:50 UTC, and the real interval between the two timestamps was 8.0 minutes. Each contract's series is read from the database backup and never from the history route, because that route caps the window at 90 days and in the Senate contract that would hide the start of the series, which is 14 April. The piece of series the daily backup does not yet cover is closed with the route's intraday points and checked by a positive control, series by series, comparing the two sources over the window in which both have data: in this issue all five series closed with identical points on both sides, and that is why every superlative here holds for the whole series and not only for what the backup already had. Closes are the last recorded point of each day, and top and floor are intraday, because a daily close hides extremes. The week's delta is anchored to the previous Friday's close, 16 October, which is the convention of this United States series, and the tool that calculates it was required to reproduce the numbers published in the previous issue before being used here. A band distribution is cited only when the book's sum falls between 95% and 105% at the edges used, and it is cited as a share of the book, normalised by the sum at that edge. A band book summing above 100% carries overpricing, and this house tests where it sits by comparing the normalised share of the book's centre against the raw sum across the series: if the centre's share falls when the sum rises, the excess is in the tail; if it rises with it, the excess is in the centre; with no clear relationship, it is approximately uniform. Measured across 671 captures of the Senate seats book, the correlation is -0.40, inside the range this house treats as no clear relationship, between -0.5 and +0.5. In that case normalising preserves the internal proportion, and that is why the distribution here supports statements about SHAPE and never about level, and why no share of it is compared with the price of a binary contract. The generic ballot average is a simple arithmetic mean, unweighted, in a 30 day rolling window, with one round per pollster per wave by the hierarchy likely voter, registered voter, adults, and with a wave published under another instrument, such as the ballot with the names of the district's candidates, kept out of the calculation and declared with the pollster's own wording. Book depth is not cited in this brief, by an editorial decision of May 2026.
9.Additional reading · macro coverage
The items below are the outlets and pollsters cited in the sections above, with the address of the story or document itself. Items the collection only reached through the aggregator were dropped from this list, because the aggregator's address does not point to the outlet.
- The New York Times, 23 Oct · The Justice Department's request to block certification of Alaska's Senate race, www.nytimes.com/2026/10/23/us/politics/justice-dept-alaska-senate-race-sullivan.html
- Siena Research Institute, 21 Oct · The October New York Times/Siena national survey, with the named ballot, sri.siena.edu/2026/10/21/new-york-times-siena-national-survey/
- Ipsos, 19 Oct · The topline of the October Reuters/Ipsos survey, with the House vote among registered voters, www.ipsos.com/sites/default/files/ct/news/documents/2026-10/Reuters%20Ipsos%20News%20Issue%20Survey%20October%202026%20Topline_2.pdf
- The Hill, 20 Oct · North Carolina's mail-in ballot restrictions kept through the election, thehill.com/homenews/state-watch/6161350-judge-upholds-nc-mailballot-rules/
- NPR, 23 Oct · Ohio as the state that could decide control of the Senate, and its rural vote, www.npr.org/2026/10/23/nx-s1-6012809/ohio-senate-governor-election-rural-voters
- The New York Times, 23 Oct · Roy Cooper and North Carolina's Senate race, www.nytimes.com/2026/10/23/us/politics/roy-cooper-north-carolina-senate-election.html
- The Hill, 22 Oct · Kansas's Senate race, thehill.com/homenews/campaign/6166006-adam-hamilton-kansas-senate/
- Fox News, 23 Oct · The Senate Republican leader facing the possibility of losing the House, www.foxnews.com/politics/thune-weighs-nightmare-scenario-gop-democrats-threaten-flip-house
- The New York Times, 24 Oct · Why Democrats could find it hard to win the Senate despite their polls, www.nytimes.com/2026/10/24/us/politics/democrats-expectations-blue-wave.html
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.