@UmpScorecards is run by college students, at no cost to fans. Please consider donating to help us cover our development and maintenance costs, as well as to support future improvements to the platform. Contributions are greatly appreciated!
The UmpScorecards Patreon
You are not yet logged into a Patreon account, which means that you cannot access our locked content.
About Us
@UmpScorecards is an online platform dedicated to measuring the accuracy, consistency, and favor of MLB umpires. It is not affiliated with Major League Baseball.
Ethan Singer is the founder, co-owner, and technical director of @UmpScorecards. He is an undergrad at Boston University studying Statistics, Computer Science, and Public Policy. You can find more of his work and his resume on his personal website. You can reach him at [email protected].
Ethan Schwartz is the co-owner and business director of @UmpScorecards. He is a recent graduate of the University of Pennsylvania where he studied Mechanical Engineering. You can reach him at [email protected].
Our History
@UmpScorecards started on Reddit in October of 2019, covering the MLB postseason. At the time, graphics were made by hand: data was downloaded manually, a python script was run to generate summary statistics, and the results were entered into an Adobe Illustrator template. These early graphics emphasized the frequency of incorrect calls and not much else. Here is our first ever graphic:
During the following offseason β which was extended because of the COVID-19 pandemic β the @UmpScorecards team worked to investigate the impact of missed calls, in addition to their frequency. This effort culminated in an article posted in the Community Research section of Fangraphs.
Inspired by positive community feedback in response to the Reddit graphics, and to our work to quantify the value of ball and strike calls, we decided to take our idea to Twitter. In August of 2020, we created an automated bot to Tweet out Umpire Scorecards for every game, every day. This is our first ever Twitter graphic:
Our account remained under the radar until September 28th, 2020, when our algorithm calculated that the umpire cost the San Francisco Giants 1.85 expected runs in a game they lost by 1 run to the San Diego Padres. Because the game was higher stakes than most β a βwin and inβ game 162 for the Giants β our Scorecard gained some traction. The Tweet was picked up by NBC Sports San Francisco, and was our first to garner over 500 likes. We finished the 2020 season with 19,000 followers β having started from scratch only three months prior.
Leading up to the 2021 MLB season, the @UmpScorecards team worked to improve our graphics, and to introduce a measure of in-game umpire consistency to our Scorecards. After a successful start to the season, we continued making improvements through the regular season. Most notably, we launched our website β a data archive allowing users to filter, sort, and explore our historic data β during the All-Star break. Our work was highlighted by the Athletic, NBC Sports, Bleacher report, and others. We finished the 2021 season with over 160,000 followers.
Leading up to the 2021 MLB season, the @UmpScorecards team worked to improve our graphics, and to introduce a measure of in-game umpire consistency to our Scorecards. After a successful start of the season, we continued making improvements through the start of the regular season: ultimately, we launched our website β a data archive allowing users to filter, sort, and explore our historic data β during the All-Star break. Our work was highlighted by the Athletic, NBC Sports, Bleacher report, and others. We finished the 2021 season with over 160,000 followers.
The offseason leading up to the 2022 MLB season allowed our team to make more significant improvements, especially to the website. We added historical umpire data going back to 2015 to our archive. We built our own API so that users could explore our data using custom and more advanced filtering. We also updated our measure of umpire consistency. During the All-Star break, we introduced our expected umpire stats: a machine learning approach to estimating an umpireβs performance relative to their peers. Our platform and metrics were covered by the New York Times and USA Today, among others. We finished the season with over 300,000 followers.