Data
The files this site is built from, as they stand now. Each is the same file the pages read, so a figure on the site can be checked against its download.
Full district snapshot archive
Immutable bundles retain the published forecast, every district row, measures and MO-2, with checksums and recorded field changes. The archive begins at its stated baseline; earlier headline history cannot reconstruct missing district records. Files do not include simulation draws or copies of every external source.
Full immutable district records begin at this baseline. Earlier headline history is not a full snapshot archive.
- Nov 3, 12:24 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 3, 4:44 am CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 10:35 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 10:10 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 8:53 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 2:18 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 12:04 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 12:04 pm CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 11:57 am CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 3:22 am CT: District records and changes (JSON) · Seats CSV · Checksums/provenance
- Nov 2, 3:22 am CT — archive baseline: District records and changes (JSON) · Seats CSV · Checksums/provenance
Files
| File | Generated | Size | What is in it |
|---|---|---|---|
| forecast_2026.json Forecast run | Nov 3, 12:24 pm CT | 7 KB | The current run: parameters, fitted coefficients, and for each chamber the expected seats, 10th, 50th and 90th percentile and the chance Republicans keep two-thirds; the scoring rule; vintage ids. |
| seats_2026_forecast.csv Seat table | Nov 3, 12:24 pm CT same run as the forecast | 66 KB | One row per seat on the ballot: candidates, incumbency, 2024 lean (L_*), model inputs, expected Democratic two-party share (mean), seat error (sd_seat), chance a Democrat wins (p_dem) and the simulated 10th, 50th and 90th percentile of the share. |
| seat_hist_2026.json Seat distributions | Nov 3, 12:24 pm CT same run as the forecast | 43 KB | For each contested seat, counts of simulated elections in 100 one-point bins of the Democratic two-party share, plus how many draws fell outside the bins. |
| measures_2026.json Ballot measures | Nov 3, 12:24 pm CT | 11 KB | The five statewide measures: wording, polls used, expected yes share with its 10th and 90th percentile, and the chance each passes. |
| congress_2026.json US House | Nov 3, 12:24 pm CT | 10 KB | The congressional forecast (MO-2): chance a Democrat wins, expected share and range, and the fundamentals, polls and market inputs blended. |
| forecast_history.json Forecast history | Nov 3, 12:24 pm CT | 8 KB | One row for every version whose rounded figures changed: the chance Republicans keep two-thirds in each chamber, expected seats, measure chances and MO-2. Headline figures only, not the full files. |
| calibration.json Backtest calibration | Nov 3, 12:24 pm CT same run as the forecast | 310 KB | From the historical backtest (2018 to 2024): for each model, ten bins of predicted chance with seats, Democratic wins, average predicted, observed frequency and Wilson 80% and 95% intervals; bins under 10 seats are flagged. Not a forecast of 2026. |
| nowcast.json Election-night nowcast | Nov 3, 12:24 pm CT | 157 KB | The live model's snapshot: status (pre-returns or counting), each contest with its counted fraction and projection, and the seat shift it implies. Rebuilt hourly before returns and on every capture on election night. |
| comparators.json Published comparison ratings | Nov 3, 12:24 pm CT same run as the forecast | 60 KB | Outside legislative ratings with their own source dates and URLs. Probabilities appear only where the source publishes them; these are comparisons, not model inputs. |
| finance.json Campaign finance context | Nov 3, 12:24 pm CT same run as the forecast | 29 KB | Dated candidate and committee finance records with the supplied source attribution. These contextual records are separate from model win probabilities. |
Generated times are in Central time and are read from each file; the two files with no time of their own carry the time of the forecast run they came from. Shares are of the two-party vote unless a field says otherwise. Probabilities are simulated frequencies from 100,000 simulated elections.
Earlier versions
The forecast is re-run every hour, and a new version is recorded only when its rounded figures change. Every such version is a row in forecast_history.json, with its time and headline figures (chamber chances and expected seats, measures, MO-2). The downloads above are always the latest full files; earlier full files are available only from the full archive’s recorded baseline onward. Method and inputs are on Methodology; each modelling choice is on Decisions; how the approach has been tested is on Evaluation.
Licence and citation
Original text: Creative Commons Attribution-NonCommercial 4.0. Data files: Creative Commons Attribution 4.0 (reuse with credit, including commercial use). The election results inside them are public records.
The files contain public election records (candidate lists, incumbents, past results and, on election night, vote counts) alongside the model’s own figures. The records are public records of the Missouri Secretary of State and other cited sources; sources are listed on each page.
Suggested citation. Bogard, Daniel. “Missouri Election: forecast data, version of Nov 3, 2026.” https://www.missourielection.com/data. Accessed [date].