ABS Challenge Tracker
MLB's 2026 robo-ump challenge system, tracked all season: which umpires get overturned the most, which teams challenge best, and whether challenge success actually correlates with winning.
jasonhorne.org
Where the box score meets the data science: live analytics dashboards, machine learning models, and AI tools for baseball, basketball, and fantasy sports.
Live dashboards
MLB's 2026 robo-ump challenge system, tracked all season: which umpires get overturned the most, which teams challenge best, and whether challenge success actually correlates with winning.
Is it real or is it luck? Every Braves hitter and pitcher graded on expected vs actual production from pitch-level Statcast data, with rolling trends and an auto-flagged watch list for fading bats and dying fastballs.
Every payroll, record, and postseason result for 450 team seasons. Spending explains about 15% of winning and buys a playoff berth far better than a ring: top-five payrolls reach October 60% of the time and still win it all only 9% of the time.
Models and tools
A stacked ensemble trained on 18 years of KenPom and Barttorvik data to predict the NCAA tournament, inspired by Jared Cross's first-place Kaggle solution. Full methodology and code.
An AI assistant that preps drafts, runs live draft support, and analyzes weekly matchups and trades using parallel research agents pulling current data. Built as a reusable Claude Code skill.