COOPER drives bracket models
Nate Silver’s COOPER ratings are a central engine for tournament predictions this March Madness, powering model-driven upset forecasts and bracket probabilities (natesilver.net). Analysts and bracketologists are leaning on COOPER to separate real upset risk from noise after a volatile Elite Eight (natesilver.net).
Nate Silver formally introduced COOPER as Silver Bulletin’s NCAA basketball rating system on March 10, 2026, in a post titled “Introducing COOPER.” (natesilver.net) Silver Bulletin published COOPER power ratings covering all 365 men’s Division I teams in its men’s ratings release. (natesilver.net) A brief “final pre‑tournament COOPER ratings” update from Nate Silver on March 15 noted that Duke edged Michigan in the final ratings, and Silver’s full March Madness predictions were published in mid‑March with updates as late as March 29. (substack.com) COOPER is behind Silver Bulletin’s paywall, but independent modelers have added COOPER to ensemble composites—Probabilis listed COOPER alongside KenPom, Powerrank and T‑Rank in this year’s weighted composite. (probabilis.blogspot.com) Open modeling projects and bracket builders show the same pipeline used by analysts: base power ratings feed a Log5 head‑to‑head probability model, then Monte Carlo adjustments (injury, momentum, luck) are applied to generate round‑by‑round probabilities. (github.com) Major sports outlets published updated odds, upset guides and Elite Eight/Final Four picks across March 15–29 while bracketologists and betting sites incorporated model outputs into their coverage ahead of the Final Four. (cbssports.com)
