September 17, 2020
KEY POINTS FROM THIS ARTICLE
— The Trial-Heat and Convention Bump Forecasting Models have an excellent record for accurate predictions of the presidential elections going back to 1992.
— Forecasting models depend on applying electoral history to the current election, but 2020 is historically abnormal (at least, in the period since 1948).
— The greatest challenge for forecasting this year is in how the catastrophic second quarter GDP should be treated.
— Based on President Trump’s approval ratings, in general and on the economy, as well as the projected third quarter GDP growth rates, the forecasts should depend exclusively on the preference polls, and they point to another extremely close election.