United States Desk

Who controls Washington when the map stops being theoretical?

House, Senate, special elections, and presidential drag — projected from public polls and updated as new numbers land.

Live instrument

Lame-Duck Index

69.7 /100

Quacking

7-day delta: -5

Full index →

House 2026

Dem 220 · Rep 215 / 435

D majority probability: 71%

House projection →

Senate 2026

Dem ~51 · Rep ~49 / 100

D majority probability: 58%

Senate projection →

CUSMA Showdown

Leverage gap: +44.5 — Canada commands the table

Goose vs Lame Duck →

Special elections

Active and recent specials

Resolved special primary · June 2, 2026

CA-1 primary archive.

Resolved archive of the June 2, 2026 top-two special primary. Republican James Gallagher finished fi…

U.S. House special · June 16

CA-14 special primary forecast.

California's 14th Congressional District has a certified special-election field. This desk models th…

Result locked · April 8, 2026 · Georgia 14

Fuller won. We called it.

Vote-Scope built this special-election page for a no-polls runoff environment, leaning on first-roun…

Resolved special election · April 16

NJ-11 result archive.

This page is now a resolved archive. Vote-Scope projected Analilia Mejia as the clear favorite befor…

U.S. Midterms 2026 — balance of power

Live odds via Polymarket

U.S. Presidential 2028

Live odds via Polymarket

Model vs Market

Where the model and the market disagree.

Control odds, race-level confidence, and the moments traders panic — set right next to our projection, so the gaps are easy to spot.

Vote-Scope model Vote-Scope
Polymarket / odds Market

Polymarket benchmarks, refreshed daily next to every projection.

Methodology after the signal

How the U.S. projections are built.

House and Senate control, seat fragility, and market disagreement — each built from public polls in a bias-corrected Bayesian model, calibrated on past results, with the simulation only translating votes into seats. All open on the page.

  • House and Senate polls are sourced from the NYT tracker, aggregated with a Bayesian BSTS model, and run through 435 and 100 district-level simulations respectively.
  • The Lame-Duck Index (LDI) combines presidential approval, economic indicators, congressional district margins, and prediction-market odds into a single 0–100 drag score.
  • Special elections use a fundamentals model: incumbency, prior margin, candidate profile, and any available local polling.
  • Prediction-market benchmarks (Polymarket) are shown alongside model output as an independent disagreement signal, updated daily.