Election & sports intelligence

A bureau of electoral and sports intelligence.

Not just who’s ahead. What’s moving, what’s fragile, and what the market is missing.

World Cup 2026 — live desk

One match. One trophy. Spain–Argentina.

🇦🇷 Argentina — 53.59%

Argentina came from behind to end England’s run; Spain shut out France. After 102 matches, the first 48-team World Cup is down to a final too close to call — 50,000 simulations put only seven points between them.

Open the World Cup desk →  ·  Beat the model →

Today's headline number

69.7 /100

Quacking

Lame-Duck Index · 7d: -5 · as of 2026-07-14

The flagship Vote-Scope instrument measures presidential drag ahead of the 2026 midterms. Higher = more lame duck. Currently in the Quacking zone.

Open full index →

Model vs market

Where we line up with the market — and where we don’t.

Polymarket and betting odds sit right beside the model, so you can see at a glance where the traders and the math agree, where the market is panicking, and where Vote-Scope sees an edge.

Model output Vote-Scope
Benchmark Market

Refreshed daily, beside every projection.

Methodology after the signal

How the projections are built.

Every projection is built from public polls, an open Bayesian method, and the source data behind it — all on the page.

  • All projections aggregate public polls in a Bayesian BSTS model — house effects measured and corrected, correlations estimated from the data, accuracy calibrated on past elections. Simulation only translates it all into seats.
  • Prediction-market data (Polymarket) sits beside the model as a disagreement detector, not a substitute.
  • Riding and constituency probabilities are conditioned on national and regional swings, then re-aggregated to seat distributions.
  • Source data, methodology notes, and model outputs are public at each desk. Updated on a rolling basis as new polls arrive.