Forecast methodology

1 · Ratings, set by hand

Editorial judgment is the input; the math only expresses it.

Every 2026 Senate, House, and Governor race is rated on an eight-point scale — tilt · lean · likely · safe for each party. There is no tossup: every race is rated with a favored side, with tilt the most competitive tier. The rating is a human call, grounded in the structural record Akashic already owns: 132 years of same-office results, the partisan lean of every geography, demographics, incumbency, and seat exposure. We do not chase polls, and we do not re-rate week to week. The conviction is the product.

2 · Ratings become probabilities

One calibrated table — the rating sets the number.

Each rating tier maps to a single probability that the favored party wins, calibrated from the historical hit-rate of that tier across past cycles (Senate, House, and Governor races retro-classified by their prior-cycle margin). That tier probability is the win number shown on every race, list, map, and headline — the rating is the input, and the table below is the only thing that turns it into a percentage. This is retrospective model research, not a forecast Akashic published, and it is not part of the public scorecard. A race may also carry a predicted two-party margin; that margin shades the map, but it never overrides the tier’s probability, so the rating chip and the win percentage can never disagree.

The seat simulation (next) layers on a per-office error spread (σ) so an individual race can drift from its rating when the whole map is simulated together: U.S. House ±14.1 pts, U.S. Senate ±21.8 pts, Governor ±15.1 pts. These are large because a prior-cycle margin is a deliberately weak proxy for a real rating. It tests the mechanics without manufacturing a historical Akashic track record. Calibration version backtest-2.

3 · The seat simulation

Correlated error is mandatory — independence would lie.

The chamber outcomes come from a 50,000-run Monte-Carlo over every contest. Each run draws a single national environment swing — s ~ N(0, ±7.8 pts) — applied to every race at once, plus an independent per-race shock. Modeling races as independent would manufacture an absurdly narrow, dishonest seat distribution; the shared national swing is what makes the spread honest. The variances are set so each race’s own win probability is preserved exactly, no matter how strong the national correlation. From the simulation come the control probabilities, the median seat count, the 80% and 95% ranges, and the tipping-point race.

A tied Senate (50–50) is resolved by the Vice President’s tie-breaking vote, which belongs to the Republican column for the entire 2026 cycle (VP Vance, term to January 2029). So Senate control is scored as Democrats needing a full 51 seats and Republicans 50 — a 50–50 chamber counts as Republican control, not a coin-flip split. The House cannot tie (435 is odd), and the governorships are an aggregate count with no tiebreaker, so an even statehouse split is reported on its own.

4 · One locked call

Published once, immutable, timestamped.

The forecast is drafted privately, then locked and published once — ratings, margins, the simulated distribution, and this methodology, all frozen with a timestamp. A genuine error correction is a new version that preserves the original; the record must always show what we actually said. A forecast is never shown as a result: it carries a persistent prediction banner and a distinct, dashed visual treatment, and it never shares a surface with a live call.

5 · Scored in public

A prospective record, beginning with 2026.

When the results land on the same race pages, every locked call is scored automatically and kept permanently: the Brier score on the win probabilities (overall and by office), a calibration curve (did our 70%-favorites win about 70% of the time?), the directional record with the misses named, and the seat-forecast error — was the actual composition inside our 80% range? Each scored cycle is evidence for the next forecast’s credibility.