The first On Point presidential forecast: where 2028 starts

Two years out is early, and any forecast this far ahead is mostly a statement about the starting position. That is still worth stating carefully, because the starting position after a midterm tells you a lot about what each side has to do. The first On Point presidential forecast runs the 2028 election 4,000 times on a national environment built from our polling, every state's presidential history and the electorate that turned out in 2026. Here is where it begins.

The top line

Against a generic Republican nominee, the Democratic nominee wins 90 percent of the simulations, with an average of 328 electoral votes. The middle 80 percent of outcomes runs from 268 to 395 Democratic electoral votes, so a narrow Republican win is well inside what the model considers possible. The national popular vote averages D +6.7, with an 80 percent range from D +1.3 to D +11.9.

That is a strong position, and it mostly reflects the year Democrats just had. Our final 2026 poll and the national House vote both put Democrats ahead by about thirteen points. Midterm waves fade, and the model already assumes a large part of this one does: the presidential environment it starts from is roughly half the size of the 2026 margin. The question for 2028 is how much more of it fades.

The same six states decide it

The model measures how often each state casts the deciding electoral vote. Pennsylvania, Michigan, Georgia, North Carolina, Wisconsin and Arizona together are the tipping point in 87 percent of simulations. That is the 2024 battleground almost unchanged, which is itself a finding: the 2026 wave did not redraw the map, it moved the whole map toward Democrats.

Pennsylvania sits at the center again. The model has it at D +5.2 and gives it a 24 percent chance of being the decisive state on its own. Further out, Texas at R +2.9 is the largest state that could plausibly move into play, which is why our 2026 miss there matters so much for how we model it.

The split result risk

Because Democrats run up large margins in big states they already win, they can lose the electoral college while winning the popular vote. The model puts that at about 5 percent, against 0.6 percent for the reverse. A Democratic popular vote lead of a couple of points is not enough on its own. The battleground states have to come along.

The primaries

The forecast also simulates both nomination fights, state by state, with real delegate rules and candidates dropping out when they fall behind.

On the Democratic side, Harris starts as a clear front runner, leading the simulated primary in most states, with Newsom and Buttigieg the most likely alternatives. Front runner status this early is fragile: name recognition explains a large share of early primary polling, and it fades once voters meet the rest of the field. On the Republican side, Vance leads Rubio, with Trump Jr. a distant third.

What would change the picture

• How far the wave fades. Every point the national environment moves toward Republicans shifts the battleground roughly one for one. A return to a 2024 style environment would put Republicans back in front in the battleground.
• Who turns out. The 2026 wave brought out millions of voters who rarely vote in midterms, while some reliable presidential voters stayed home. Which of them show up in 2028 is the subject of who votes in 2028.
• The nominees. Candidates matter at the margins. The 2028 forecast page lets you set any pairing and see each candidate's state by state effect.
• A third party. A credible independent pulls votes unevenly by state, which can matter more than the national share suggests.

We will update this forecast as primary polling arrives and as the field takes shape. The model, its assumptions and every revision are published on the site.