Forecasting a Democratic Resurgence? Partisan Accountability in the 2026 U.S. Midterm Elections

This article forecasts the 2026 U.S. midterm elections, building on the model developed by Algara et al. (2025) to predict the seats won by the president's party from the generic ballot, net presidential approval, midterm status, and the partisanship of the median seat at stake.First, we update our time series of presidential approval and the generic ballot through 2026, showing that the two evaluations remain distinct and diverge sharply this cycle.We validate the model with 34 rolling-origin forecasts per chamber using only pre-election information.Second, we forecast that Republicans are highly likely to lose the U.S. House, winning 196 seats [95% PI: 160, 221] with a 10% probability of retaining control at their 3-point generic ballot deficit.Third, our model forecasts Republicans narrowly holding the U.S. Senate with 51 seats [95% PI: 42, 61] and a 55% probability of control, reflecting an unusually favorable Senate class, although alternative specifications forecast a Democratic majority.Lastly, we measure how the House lines redrawn for 2026 shift partisan terrain, an effect our historically estimated model only partially absorbs, and show that the new maps could decide a close national contest but would not overturn the current House forecast.