augur

A pre-registered study for the NFL Big Data Bowl 2027, testing whether a pass rusher's burst at the Combine predicts how fast he gets off the line in real games.

Category

Data Science

Industry

Sports Analytics

Role

Solo Competitor

Stack

Python, Polars, NumPy

Football field with player tracking traces and a burst curve after the 10-yard split
What I built

I measured a Combine burst residual after the 10-yard split and compared it with regular-season get-off across 25,278 rushes from 98 players in the 2023 to 2025 draft classes, holding out one class at a time to test each prediction fairly.

Challenge and solution

Sports data makes it easy to find patterns that aren't really there. I pre-registered the analysis, kept a public changelog, and ran a synthetic-outcome pilot and garbage-trait controls to prove the method can tell a real signal from noise.

Approach

Every decision was written down before the results came in, so the finding stands whichever way it lands.