2026 Forecast Model
How Akashic Edge rates and forecasts Senate, House, and Governor races. Every forecast page uses the same core model with office-specific parameters.
We rate every race
Four things go into every rating: how the state or district voted for president in 2024, whether there’s an incumbent (and how strong), whether it’s an open seat, and what the national mood looks like right now. We then translate each rating into a specific win probability for the model.
Statewide races are most sensitive to national wave effects.
District-level races are more localized. ~130 competitive, ~305 safe.
Governors have stronger personal brands; more insulated from national environment.
Ratings become numbers
The ladder is symmetric — Safe Dem and Safe Rep are mirror images. We calibrated these numbers against how similar rating systems have performed historically, and they’re the same across all three offices.
Every possible outcome, counted
Each race is treated as an independent coin flip weighted by its rating. We then add up every combination of wins and losses that could happen across all the races, and get the precise probability of every final seat total. No simulation error, no approximation — runs in under 10ms for all 435 House districts.
Seats not up this cycle just stack on top. Senate D total = locked D seats + D contested wins. Same pattern for governors across all 50 states.
Accounting for a wave
Step 3 assumed every race stood on its own. But a good or bad night for one party tends to be a good or bad night for everyone in that party — a national wave. So we run the whole model 10,000 times, each with a slightly different national environment, and keep track of how often each outcome shows up.
The “wave intensity” control (sigma) sets how wild the simulated environments get. Higher sigma means wider ranges and more possibility for a surprise.
Statewide races are most sensitive to national wave effects.
District-level races are more localized. ~130 competitive, ~305 safe.
Governors have stronger personal brands; more insulated from national environment.
Try your own scenario
Every forecast page has an interactive scenario builder. Flip any race rating — the overall seat distribution recomputes in real time. Want to know what happens if Dems win both Georgia seats? Or if the wind shifts three points? Dial it in, watch the odds move.
The House forecast also lets you play with redistricting. Live probabilities from Polymarket drive the defaults, but you can override any one of them and see what it costs — or wins — for each side.
Data Sources
- State/district partisanship: Akashic Edge elections database (165 years of results, county and block-level disaggregation)
- CD-level presidential margins: Block-level disaggregation of 2024 presidential results onto 120th Congress boundaries
- Redistricting: State-enacted shapefiles (CA, TX, NC, OH) + UCLA Redistricting Data Hub
- Prediction markets: Polymarket live probabilities for redistricting outcomes and individual races
- Polling: Aggregated from FiveThirtyEight, RealClearPolitics, and direct pollster releases
- Candidate/incumbency: Current as of the latest forecast update