Pitcher Greeks: First Principles, Stuff Models, and Derivatives
A 10-minute talk on grading pitches with continuous stuff models and their sensitivities, using MLB Statcast data (2020–2026).
Professional Summary
I’m a quantitative trader at Chicago Trading Company and a baseball
researcher in my free time. My recent work develops “pitcher greeks”:
continuous, differentiable stuff models whose sensitivities show how run
value responds to changes in a pitch’s velocity, movement, and approach
angles.
Education
TODO: Your degree
2020-09-01
2024-05-31
TODO: Your school
Interests
Pitch quality (“stuff”) modeling
Sensitivity analysis & interpretable ML
Sports analytics
⚾ My Research
I build models of pitch quality — how much of run prevention is attributable to a pitch’s physical characteristics alone. My current work treats stuff as a continuous index per pitch type and borrows the language of options pricing: each pitch gets “greeks,” the local sensitivities of expected run value to velocity, induced movement, and approach angles, evaluated on the manifold of pitches that actually get thrown.
Please reach out if you’d like to talk pitching models 😃
Publications
Talks