House Forecast


Cory McCartan

The House map

Each district is shaded by the probability of a Democratic or Republican win. Cross-hatching indicates an incumbent who has a 50% or higher chance of losing. You can hover over a district to learn more.

Public opinion over time

This chart shows the model’s best estimate of how the generic congressional ballot has evolved over the past few months. The darker and lighter bands show 50% and 80% credible intervals.

How the national environment affects outcomes

The chart below shows the range of outcomes we forecast conditional on different national environments. For example, if Democrats win 51% of the vote nationwide, they are expected to eke out a majority on average, but could win anywhere from around 185 to around 250 seats. The colored box covers the middle 50% of possible outcomes, the thick line covers the middle 80% of outcomes, and the thin line covers 98% of outcomes.

The 435 House races

You can search by district state, number, candidate, incumbency, rating, or contestedness. Try searching for “Pelosi”, “WA 7,” “open,” “safe,” “contested,” “lean rep.,” or “tossup” seats.

†Estimates do not account for rank-choice voting.

How the odds have changed

The model is regularly re-run as new data and polls come in. The charts below track how the model’s election-day forecast has changed over time.

National polling

The most recent generic congressional ballot polls are shown in the table below. The Impact column roughly measures how the poll is currently affecting the model—whether it is pulling the forecast towards Democrats or Republicans, and by how much.

50% and 80% credible intervals are presented throughout.

A detailed write-up of the model, along with code and data, are available here.

Data are courtesy of FiveThirtyEight, Data for Progress, VEST, the ALARM Project, Daily Kos Elections, the MIT Election Data + Science Lab, and IPUMS.