Writing
Notes on data, machine learning, building things, and occasionally comedy.
A rumor mill in Discord-bot form, and my biggest project to date.
Turning the league's Elo ratings into moneylines, spreads and a vig, then betting the favorite every week for eight seasons. The bettor comes out $72.93 ahead.
Twenty-five ways to rate a fantasy league, scored against eight seasons of real matchups. The one I'd been using all along won.
Predicting who comes down with a rebound from tracking data alone — what was broken in the 2017 pipeline, what the rebuilt models score, and what the movement model is and isn't worth.