Forecaster scorecard
Each forecaster compared here, FutureBallot included, is scored race by race on certified counts.
Four of 488 races scored so far. A race is scored on this page once its state certifies the count and the results page shows it certified; a race that meets those conditions but cannot be matched to stored result rows, or that meets only one of them, can be named as held back. 484 races are not scored yet: still being counted, awaiting certification, or held back. Scored so far: Senate: one of thirty-five; Governor: one of thirty-six; House: two of 417. The certification statuses are as of November 16, 2026, 10:32 p.m. Pacific Standard Time (November 17, 06:32 UTC).
This page is not the registered scoring. That scoring is first published on Accuracy once at least 95% of the sealed forecast's races for each office are certified; this page shows each race as its count is certified, and this page's running totals are not the registered result.
Where FutureBallot missed
A race is a miss when its favorite lost, when the result fell outside the 90% range FutureBallot sealed before Election Day, or when the winner came from outside the two sides the forecast measured. Every miss among the races scored so far is listed here:
- Idaho's 2nd District. Mike Simpson won by 11.4 points; FutureBallot's expected outcome (Mike Simpson to lead by 21.7 points) came with a 90% range of +12.3 to +31.2 points for Mike Simpson's margin. The result fell below the 90% range.
The forecasters compared, on the races scored so far
Besides FutureBallot, the registered scoring compares, race by race, Race to the WH's chances and the ratings of Cook, Sabato's Crystal Ball and Inside Elections. Silver Bulletin and Decision Desk HQ are not part of the registered comparison, because their terms forbid collecting their numbers by machine; the registration allows only an informational comparison of the two, from publications that cite their figures. Split Ticket is registered as a probability forecaster, but it published ratings rather than probabilities, and the registration's ratings comparison names only Cook, Sabato's Crystal Ball and Inside Elections. Split Ticket therefore does not appear on this page.
Each forecaster is scored only on what it published. For FutureBallot and Race to the WH, a favorite is right or wrong by the wrong-call rule in the registered scoring code, described under How the scorecard works, below. A rating other than a Toss-up names a favorite and a strength (Safe or Solid, Likely, Lean or Tilt) but no chance. Here a rating is scored only on whether its favorite won, which is the registration's wrong-call count for ratings, and is never turned into a probability. The registered comparison of ratings, reported on Accuracy, also uses the strength, as a ranking measure (AUC). A Toss-up, or a call that names no side, names no favorite and is counted in the No favorite named column. A race with a favorite is counted as not scorable when the certified count cannot settle it for that forecaster. On a candidate race, for FutureBallot and Race to the WH, that means the count lacks the first-named candidate's share or the top share; on FutureBallot's party-level rows, not scorable means the favorite's party or the winner's party is not recorded. For a rating, not scorable means the count names no single winner with a party: no share is recorded, the top two tie, or the winner's party is not recorded.
| Forecaster | Races | Favorite right | Favorite wrong | No favorite named | Not scorable | Margin and range |
|---|---|---|---|---|---|---|
| FutureBallot | 4 | 4 | 0 | 0 | 0 | Average absolute margin error 5.4 percentage points; 3 of 4 results inside the 90% range |
| Cook Political Report | 2 | 2 | 0 | 0 | 0 | A rating only; no margin or range |
| Sabato's Crystal Ball | 4 | 4 | 0 | 0 | 0 | A rating only; no margin or range |
| Inside Elections | 4 | 4 | 0 | 0 | 0 | A rating only; no margin or range |
Race to the WH's scores will appear in this summary once they cover at least five scored races; so far they cover four. That is this page's own rule, not part of the registered scoring: over a single race, Race to the WH's Brier score and the result would reveal the chances it gave the two candidates. FutureBallot finds no terms of use on Race to the WH's site. Where FutureBallot has an archive of Race to the WH's chances, those chances are used privately for this comparison only and are not republished, race by race or in the download. When Race to the WH's figures appear here, they are computed from those chances over the scored races that Race to the WH forecast.
Race by race
Margins are in percentage points of all votes cast, from the side of FutureBallot's favorite (where it gave each side one half, from the side of the first candidate it named): a plus sign means that side finished ahead. The range is the sealed 90% range for that margin. The raters' calls are as Wikipedia transcribed them on November 1, 2026, 10:06 p.m. Pacific Standard Time (November 2, 06:06 UTC); that capture is the last one FutureBallot archived before polls closed. A rater with no call in that capture shows as not rated.
Senate
| Race | Certified result | FutureBallot's favorite | Margin | 90% range | Cook | Sabato | Inside Elections |
|---|---|---|---|---|---|---|---|
| Idaho Senate | Jim Risch (R) won | Jim Risch (R) 97% chance to win; won | forecast +16.9, result +22.9, off by 6.1 | +2.6 to +31.0 result inside | Solid R won | Safe R won | Solid R won |
Governor
| Race | Certified result | FutureBallot's favorite | Margin | 90% range | Cook | Sabato | Inside Elections |
|---|---|---|---|---|---|---|---|
| Idaho governor | Brad Little (R) won | Brad Little (R) >99% chance to win; won | forecast +25.7, result +24.7, off by 1.1 | +10.7 to +40.7 result inside | Solid R won | Safe R won | Solid R won |
House
| Race | Certified result | FutureBallot's favorite | Margin | 90% range | Cook | Sabato | Inside Elections |
|---|---|---|---|---|---|---|---|
| Idaho's 1st District | Russ Fulcher (R) won | Russ Fulcher (R) >99% chance to win; won | forecast +31.3, result +35.3, off by 4.0 | +21.9 to +40.8 result inside | not rated | Safe R won | Solid R won |
| Idaho's 2nd District | Mike Simpson (R) won | Mike Simpson (R) >99% chance to win; won | forecast +21.7, result +11.4, off by 10.3 | +12.3 to +31.2 result outside | not rated | Safe R won | Solid R won |
How the scorecard works
- The forecast scored. FutureBallot's forecast here is the last one sealed before Election Day, sealed at November 2, 2026, 3:07 p.m. Pacific Standard Time (23:07 UTC); the SHA-256 fingerprint of that forecast begins
4f82e8ad9ea77d80(sealed forecasts). Any comparator numbers shown are the last copies FutureBallot archived before polls closed. - Official counts only. A race is scored once every line of its deciding count is certified by the state and the results page shows it as certified. A race that goes to a runoff is scored on the runoff, once that count is certified. A race whose sealed forecast the scorer cannot match to stored result rows is never scored; once the results page shows it certified, it is named here as held back. Unofficial counts are on the results page and are never scored.
- The scorer's rules. The scorecard applies, race by race, the code that carries out the scoring FutureBallot registered before the election. That code also keeps a rule from FutureBallot's earlier scoring of past elections, which the registration does not write out: a margin is not scored where a third candidate won, or where the two sides the forecast measured took under 80% of the vote, and the favorite is still scored. The registered results, with the pass marks set before the election, are first published on Accuracy once at least 95% of the sealed forecast's races for each office are certified, are labelled provisional while any race remains uncertified, and are marked final when every race is certified.
- Favorite right or wrong. For FutureBallot and Race to the WH, the count follows the wrong-call rule in FutureBallot's registered scoring code on candidate races. The favorite counts as right when the first-named candidate's share was at least the largest any candidate took and the candidate was favored, or the share fell short of the largest and the candidate was not favored. On a party-level row, whose sides are parties, this page counts the favorite as right when the favorite's party won (the registered scorer does not score such rows). The wrong-call rule can still count the favorite as right when a candidate outside the two sides the forecast measured wins; such a race is listed among FutureBallot's misses above.
- No race is dropped. No race is left out because of how it turned out. Each race is scored once its state certifies the deciding count, the results page shows it certified, and the scorer can match the sealed forecast to stored result rows. A race that goes to a runoff is scored on the runoff, once that count is certified. A race is named as held back and not scored when the results page and the stored result rows disagree. The results page shows it certified while the stored result rows are not all certified, or the scorer cannot match its sealed forecast to them; or the stored result rows are certified while the results page shows a different status or does not list the race.
- Download. forecaster-scorecard-2026.csv has FutureBallot's score for every scored race on this page (columns and license).