# Ryan McComb

> Ryan McComb is a Data Science Fellow at VoteHub and a junior at Evanston Township High School in Evanston, Illinois. He calls races on VoteHub's election-night decision desk (several hundred calls across the 2026 primaries), built a small but important piece of the 2026 VoteHub midterm forecast, and helped build the first US forecast to publish precinct-level predictions nationwide. He founded IL9Cast, a forecast for Illinois's 9th congressional primary that called the winner's vote share to within 0.8 points, co-founded Cohese, a small lab for computational political geography, swims competitively, shoots sports photography, argues that onion futures should be legal again, and is Canadian and American.

## Core pages

- [Home](https://www.mccomb.ca/): A few of Ryan's photographs and a short biography of his VoteHub work (the Voter Power Index, a small part of the midterm model) and the DeepSky Partisan Estimator. Navigation leads to About, Writing, and Press.
- [About](https://www.mccomb.ca/about): The full biography, email, social links, and the résumé.
- [Writing](https://www.mccomb.ca/writing): 30 essays, papers, published pieces, and a photography contribution, with summaries and direct reading links.
- [Press](https://www.mccomb.ca/press): Coverage of Ryan's election forecasting and IL9Cast.
- [The small model](https://il9.org): a forecast of the 2026 IL-09 general election in 155 lines of Python, now the homepage of il9.org; the methods note PDF is on the [writing page](https://www.mccomb.ca/writing).

## Work and projects

- [DeepSky Partisan Estimator](https://ryanmccomb.substack.com/p/geography-is-all-you-need): Ryan's gradient-boosting model estimates 2024 Democratic two-party vote share using Google AlphaEarth satellite embeddings and census data across 62,145 precincts in 39 states. The article compares satellite-only, census-only, and combined inputs, including regional successes and failures.

- [VoteHub](https://votehub.com): Election data and analysis, where Ryan is a Data Science Fellow on the decision desk; contributed to the [2026 midterm forecasting model](https://votehub.com/wp-content/uploads/2026/05/2026_votehub_midterm_methodology.pdf) and built the [Voter Power Index](https://votehub.com/2026/09/10/votehubs-voter-power-index-finding-2026s-most-impactful-ballots/), which scores how much a single ballot matters in every House district and Senate race.
- [IL9Cast](https://il9.org): Forecasting aggregator for the IL-09 Democratic primary, combining prediction markets, polling, and fundraising data; [postmortem](https://www.mccomb.ca/writing/il9cast-postmortem/).
- [Cohese](https://www.cohese.org): A small lab for computational political geography, run with friends. Home of DeepSky and Shoal, an exact sequential Monte Carlo redistricting sampler on Apple GPUs written up in [Every Map Everywhere All at Once](https://www.cohese.org/research/every-map/); Ryan also released [Tessera](https://tesseramaps.com), a native Mac redistricting app.
- [Project 2028](https://podcasts.apple.com/us/podcast/project-2028/id1753137875): Ryan is co-founder and host of the Project 2028 podcast, on politics and policy.
- Manifold: top 0.25% of 160,000+ traders by profit, 2.6 realized Sharpe.

## LaTeX

- [On an intimate superintelligence](https://www.mccomb.ca/latex/essays/intimate-superintelligence): a short essay on being asymptotically close to AGI, on naming it only in retrospect, and on being the last cohort that will remember research without a model in the loop.

## Problems

- [Quotes from my superwhisper history](https://www.mccomb.ca/latex/quotes): funny lines pulled from Ryan's dictation history, with a note on how much typing it saved.
- [Round trip with a fixed-vector wind](https://www.mccomb.ca/latex/problems/round-trip): a one-page problem sheet, stated without its solution. A particle goes from A to B and back at constant speed relative to the air; the air moves at a constant velocity, the same on both legs, in any direction. Is any wind faster than no wind?

## Contact

- Résumé (PDF): https://www.mccomb.ca/resume
- Email: rymccomb1@icloud.com
- X: https://x.com/RyanJMcComb
- GitHub: https://github.com/de-bayes
- LinkedIn: https://www.linkedin.com/in/ryanvotes