2028 presidential election model
—
A state-by-state forecast that blends four inputs — historical results and fundamentals, head-to-head polls, prediction markets and early race ratings — then simulates the Electoral College 10,000 times. Nominees aren’t chosen and the 2026 midterms haven’t happened, so the ranges are wide on purpose.
State by state
Each tile is colored by the model’s expected margin and labeled with the Democrat’s win chance. Hover or tap a state for the four inputs.
What each input says
Implied national popular-vote margin from each input, and the blended result.
Electoral vote outcomes
Democratic electoral votes across 10,000 simulations, in 10-vote bins; the line marks 270.
Tipping-point states
How often each state casts the decisive 270th electoral vote.
Every matchup
Same model, each candidate pairing. Click a row to view it.
| Matchup | D win | D EV | Pop. margin | Polls | Market D | Polls used |
|---|
State inputs and forecast
Margins in points (positive = Democratic lead). History, Polls and Markets each apply that input’s national estimate to the state’s partisan lean; Ratings converts the early race rating to a margin. Click a header to sort.
Weight explorer
Change how much each input counts and how uncertain the race is. This re-runs 5,000 simulations in your browser for the selected matchup from the same inputs; it isn’t saved and doesn’t change the published forecast above.
Polls used
Hypothetical 2028 general-election polls. Newer and larger polls count more.
| Pollster | Ended | Democrat | Republican | Margin | Pop. |
|---|
Prediction markets
Prices in cents (= implied %). A nominee’s win-if-nominated chance is their winner price divided by their nomination price.
| Contract | Market | Kalshi | Polymarket | PredictIt | Used |
|---|
Fundamentals
The history input averages the last three national results and adds a damped penalty for the president’s approval. The generic ballot is shown for context only.
| Measure | Value | As of |
|---|
Model settings
Every constant the model uses.
| Setting | Value |
|---|
How the model works
- Partisan lean. Each state’s margin relative to the national popular vote, weighted 60% on 2024, 30% on 2020 and 10% on 2016. Maine and Nebraska’s congressional districts are modeled separately.
- Four national estimates. History averages the last three national margins and adds a tenth of a point per point of President Trump’s negative net approval. Polls average 2028 head-to-heads, weighted by recency and sample size; a matchup with few polls borrows from the all-matchup average, and a battleground-only poll nudges the seven swing states. Markets convert the party or nominee odds into the national margin that would produce that win chance on this map. Ratings turn each early Safe, Likely, Lean or Toss-up call into a margin.
- Blend. Each state’s expected margin is a weighted average of the four (35% history, 30% polls, 25% markets, 10% ratings by default; districts have no rating).
- Simulate. Each of 10,000 elections draws one national error shared by every state, a regional error shared within five regions, and a state error, so states move together the way they do in real elections. A 269–269 tie is counted separately.
Limits: no major forecaster (Cook, Sabato’s Crystal Ball, Inside Elections) has published 2028 presidential ratings yet, so the ratings input uses 2028 Tracker’s early map. There are almost no 2028 state polls. Ossoff has said he isn’t running, though markets still price him. Nebraska may vote on moving to winner-take-all; the model keeps the current district system. A weekly refresh pulls new polls, odds and ratings.
Where the state is decided
—
Claude’s read of the county map, Oct 2026.
Every county: where it stands and which way it moved
Across: 2024 margin. Up: shift from 2020 to 2024 (up = toward Democrats). Circle size = 2024 votes. Key counties are colored and labeled.
Key counties
Chosen by rule for each state: the two biggest Democratic and Republican vote banks (won by 10+ points), the largest bellwethers (within 4 points of the statewide margin, on average, in 2016–2024), the largest close counties (under 10 points in 2024), and the two biggest net-vote movers from 2020 to 2024. The even-race benchmark is the county margin consistent with a tied state; a county reporting above it means the Democrat is running ahead of a 50–50 pace.
| County | Why it matters | Share of vote | 2016 | 2020 | 2024 | Shift 20→24 | Net votes 2024 | Even-race benchmark | 2028 projection | Net votes 2028 |
|---|
Net votes from each key county
Votes the county added to the Democratic (right) or Republican (left) column: hollow = 2024 actual, solid = 2028 projection at 2024 turnout.
Top swing counties across the battlegrounds
All ten states, ranked by votes cast in counties decided by under 10 points in 2024, weighted toward the closest. Click a row to open its state.
| County | State | 2024 votes | 2020 | 2024 | Shift |
|---|
Election night
—
Last update: —. Results come from AP, state election offices, Decision Desk HQ and the New York Times; headlines from major outlets. Each county is scored against its even-race benchmark, so an early county tells you how the whole state would land if the rest of it behaved the same way.
Race board
Statewide count so far, and the on-pace estimate: the statewide margin implied by how reporting counties are running against their benchmarks, weighted by the vote counted. Early mail or Election Day votes can skew it either way in the first hours.
| State | Race | Democrat | Republican | Counted | Current margin | On-pace estimate | Counties in | Call |
|---|
Key counties tonight
Implied statewide margin = the county’s current margin minus its even-race benchmark. Positive means the county is running ahead of a tied-state pace for the Democrat.
| County | Why it matters | Benchmark | Counted | Current margin | Implied statewide |
|---|
Headlines
Newest first, gathered at each update.
| Time | Source | Headline |
|---|
