Archive · Live forecasting ended March 17, 2026

Illinois 9th District
Democratic Primary

A retrospective of IL9Cast: how prediction markets, a precinct-level model, and campaign-finance data tracked the March 17, 2026 primary — and how the forecast compared to the actual result.

Primary Winner
Biss
29.6% of the vote
Model's Call
77.2%
Win probability assigned to Biss
Election Day
Mar 17
Democratic Primary, 2026
Snapshots Collected
--
-- data points archived
How the model did

The precinct model was published two weeks before the primary. It called the correct winner, put the top three candidates in the right order, and landed within about two points on every top-three vote share.

Biss predicted 31.6% → actual 29.6% −2.0 pp
Abughazaleh predicted 24.3% → actual 25.9% +1.6 pp
Fine predicted 19.7% → actual 20.4% +0.7 pp

The model assigned Biss a 77.2% win probability. He won. See the full comparison on the Model page →

Project Postmortem
Essay

IL9Cast: A Postmortem

A look back at what the model got right, what the markets missed, and what it was like to run a one-person forecast for a 15-way primary.

Read on mccomb.ca →
Explore the archive
Aggregated Odds
Prediction Markets
Weighted aggregation of Manifold Markets and Kalshi through election night — the full archived time series.
View markets →
Candidate Profiles
The Candidates
Profiles, final vote shares, and a look back at every candidate who ran in the IL-9 Democratic primary.
View candidates →
Precinct Simulation
Precinct Model
100,000-simulation Monte Carlo across 436 precincts — now paired with actual results so you can see how the model performed.
View model →
Featured Analysis
Case Study

Bid-Ask Spreads Even Moses Can't Part

A trader pushed Laura Fine to 30% on Kalshi. The spread said 14–16%. Every other tracker parroted the spike. Here's why IL9Cast didn't flinch.

Read the full analysis →
About this project
IL9Cast was an independent hub for the Illinois 9th District Democratic Primary. From January through election day it aggregated prediction market data, published a precinct-level election model, tracked fundraising across every candidate, and on March 17 covered the race in real time. It now lives on as an archive.
  • Manifold Markets 40% weight in aggregation formula
  • Kalshi 60% weight (last price, midpoint, liquidity-adjusted)
  • Precinct-Level Model 100,000 simulation runs across 436 precincts (see how it performed)
  • Open Data Full historical dataset available for download
See model vs actual Archived Markets Download Data
Illinois Flag Footer