Created by Ryan McComb
I'm a student at Evanston Township High School with a passion for politics, data science, and prediction markets. I built IL9Cast because I wanted to see real-time market sentiment for the IL-9 primary race and figured some others might find it useful too. In all transparency, I have done some volunteer work for Daniel Biss, but I promise that has not influenced the efficacy of the data presented here. If you want me to make a race forecast like this for your district, I would love to!
Also, to be upfront about it: I used AI to help build the website itself -- the Flask backend, the HTML, the CSS. I'm a high schooler, not a web developer. But the actual data pipelines, the aggregation formulas, the model logic, the methodology -- all me. AI helped me put it on a screen. I wrote the code that makes the decisions.
If you want to talk markets, forecasting, or have feedback on IL9Cast, feel free to reach out.
The Lunch Table
This whole thing pretty much runs out of a lunch table at ETHS. Every day my friends sit there while I talk about polling averages and precinct maps instead of normal stuff. They've watched me pull up the site on my phone a hundred times, told me when something looked weird, and somehow kept showing up even though I've definitely bored them half to death. This has been one of the coolest projects of my life. Not because of the model or the data -- because of the people who let me be way too excited about a congressional primary, never made it weird, and never got annoyed. They believed in it before anyone else did, and that matters more than they probably realize.
Sammy Jain, Teddy Gutstein, Vikram Kelly, Nathan Gregory, Ido Salant, John Spyrson, and Nick Jackson.
There were a lot of late nights. The time Railway deleted the entire data volume at 3am and I had to rebuild everything from CSV backups. The nights spent staring at chart smoothing algorithms trying to figure out why the lines looked wrong. Refreshing the site over and over after a deploy hoping nothing broke. My parents never once told me to stop, even when I was clearly up way too late on a school night working on a congressional forecast that nobody asked me to build. Thank you for that.
Shout-out to Elle Zebala, who I've been going back and forth with about politics since middle school. She'll listen to whatever half-baked theory I have that week and then tell me exactly why I'm wrong. Wouldn't have it any other way -- a lot of how I think about this stuff comes from those conversations.
Thank you to Mr. Patel, my AP Gov teacher, and to Julian Cronin.
And to Matthew Eadie, the local reporter whose tweet about IL9Cast kind of started all of this.
To everyone else at school who's checked out the site, sent me a tip, argued with me about the race, or just let me ramble at them -- thanks. You know who you are.
The Project
IL9Cast is a real-time forecasting platform for the Illinois 9th Congressional District Democratic Primary, scheduled for March 17, 2026. The site aggregates live prediction market data from Manifold Markets and Kalshi and also publishes a precinct-level election model to provide up-to-the-minute probability estimates and geographic insights for each candidate.
This project combines sophisticated market aggregation methodology, a precinct-level election model, and clean data visualization to help voters and political observers understand the race dynamics in real time.
Press
We were also featured in the Capitol Fax newsletter.
Read Evanston RoundTable's coverage: "ETHS student aims to forecast 9th District congressional race using betting market data".
Contact Information
Inspiration
This forecasting platform is inspired by the pioneering work of data-driven election forecasters including:
- Nate Silver - Creator of FiveThirtyEight and modern election forecasting
- 538 - The original data-driven election forecast site
- The Silver Bulletin - Nate Silver's current forecasting publication
- Galen Druke - Podcast host and election analyst at 538
Support IL9Cast
IL9Cast runs on real-time data collection every 3 minutes, persistent data storage, and continuous server infrastructure. If you find this forecast useful, consider helping cover the hosting and operational costs.
