Blog
Analyses you can check
Worked analyses with the question, the method, the uncertainty, and the limits stated plainly, and the code in the open.
How do you measure a play the data cannot name?
Testing the agent's definition of the tush push against human charting, adjusting the comparison, and correcting an error I found in my own tool.
Does establishing the run win football games?
Why the halftime filter does more than it appears to, and what a test on plays called before any score finds.