Meet the humans behind the Twitter bots that track balls and strikes – The Athletic


An old Acer laptop sits on top of a mini-fridge in a 17th-floor dormitory at Boston University. He has a job. Every morning at 10:30 am, the laptop comes alive and running simple Python code, fetching data from an Internet library, generating charts and spitting them out to over 140,000 followers on Twitter. Then the laptop goes to sleep. It’s a day in the life of a bot.

Its creator is often in class at this time of the morning. Ethan Singer is a second year student studying statistics, computer science and public policy at the BU. He never imagined that so many people would follow a referee report.

The @UmpScorecards bot is the result of a project Singer started three summers ago while still a high school student in Bethesda, Maryland. A friend showed Singer a study by BU speaker Mark T. Williams that found Found MLB umpires missed 34,294 ball / hit calls in 2018. Singer wanted to know more about the impact of bad calls – quantifying their value against gross volume – so he created a tool to analyze how individual missed calls affect the operating expectancy and, after two years of DIY, forwarded it to Twitter.

“Now,” Singer said, “the tweets are just sending from my dorm.”

The bot is fully automated. It draws on detailed data released daily by the MLB and extracts five of the 89 values ​​for each field: the horizontal and vertical position of the ball as it crosses home plate; the top and bottom of the strike zone; and whether the throw was called a ball or a strike. That turns into a scorecard for every umpire, every day of the regular season and playoffs, listing their accuracy, consistency, crucial missed calls and which team those helped or injured.


Comments are closed.