Professional
I build and operate AI-augmented systems for making calibrated forecasts — on questions in policy, markets, and other socioeconomic domains — and I keep the results on the public record.
The Work
Forecasting systems
Pipelines that combine multiple AI models with structured data ingestion — legislative tracking, market data, filings, event monitoring — to produce probabilistic forecasts with explicit resolution criteria. Designed so the outputs are scoreable: every forecast can be checked against what actually happened.
Domains
Developing geopolitical situations, legislative and policy outcomes, prediction-market questions (Metaculus and related platforms), and the AI buildout — the semiconductor complex, capability trajectories, and system dynamics — plus other questions where a calibrated probability is more useful than an opinion.
The record
The research archive holds the first era of this work: 2,300+ forecast markets with resolutions and Brier scores, produced by earlier versions of the same pipeline. The current era's track record is being built in public on this site.
Engagements
I'm open to work where this system — or the experience of building it — is useful: forecasting on specific questions, designing AI-augmented analysis pipelines, or research collaborations in policy and markets. If you think there's a fit, describe the problem below.
Get in Touch
Runchey Research produces educational research and probabilistic forecasts. Nothing here is personalized investment advice.