A District-Level Forecast of the 2026 U.S. House of Representatives Midterm Elections

We develop a district-level forecast for the 2026 U.S. House of Representatives midterm elections.The model includes four district-level variables (past district vote, whether the race is competitive or unopposed, expert assessments of the likely winner, and the ratio of campaign donations to the Democratic and Republican candidates), two variables that capture national political conditions (national congressional vote intentions and incumbent presidential party if the election is a midterm), and state-level congressional vote intentions to capture state-level conditions.Before-the-fact tests based on data from 100 days (or more) prior to past elections show impressive historical accuracy, with an overall accuracy rate of 96% since 1996.An analysis of races that experts classified as tossups shows that statistical chance does not account for our model's accuracy.100 days before the election, our 2026 forecast gives Democrats a 9-in-10 chance of winning a majority of House seats.We also present a map of the expected outcome for each district.The conclusion highlights how the historical accuracy of the model can be used to understand any potential unexpected outcomes in the upcoming midterm elections.