Today The Economist launched its first real-time midterm model, which uses statistical forecasting to predict how many seats in the House of Representatives each party is likely to win in this year’s US midterm elections.
View the model here:
Applying cutting-edge machine-learning techniques to political science, the model combines information from polling, past elections, special elections, fundraising, ideology and “fundamental” factors like the economy and incumbency. It has been trained on every election cycle since 1942 and nearly 6,500 historical district races. The model will conduct 4.35 million simulated elections every day until the vote, live-updating to incorporate up-to-the-minute data.
The Economist’s midterm model currently predicts that Democrats have a 2 in 3 (or 65%) chance of taking the House, and holding an average of 222 seats, or 4 more than is needed for a majority. It shows that there is a 95% chance that the Democrats will hold between 206 and 241 seats.
The Economist’s data team plans to launch a similar model to forecast Senate results later in the year once primaries are complete and more polls become available.
Find out more about The Economist’s midterm model here.