Peter Dizikes of the MIT News Office highlights one of the finalists in the MIT Sloan Sports Analytics Conference, currently being held in Boston. The paper, “A Data-driven Method for In-game Decision Making in MLB”, was developed by John Guttag and Ganeshapillai Gartheeban.
Guttag and Gartheeban’s developed a model that suggested when managers should take their starting pitchers out of the game by using data from the first 80 percent of the 2006-10 seasons to develop a model. Then they tested their model against the results in the final 20 percent of the season.
Dizikes summarizes some of the important findings:
The study finds that from the fifth inning on, in close games, pitchers who were left in games when the model recommended replacing them allowed runs 60 percent of the time, compared to 43 percent of the time overall.
Over 21,538 innings, the Guttag-Gartheeban model disagreed with the manager’s decision regarding his starting pitcher 48 percent of the time. About 43 percent of the time, the manager left the starting pitcher in when the model indicated he should be replaced. In just 5 percent of the cases did managers pull starting pitchers when the model suggested they should stay in the game
Guttag and Gartheeban do note that their methodology doesn’t consider every factor and recognize that the manager is considering a lot more than just the current game situation. Hypothetically, leaving Justin Verlander in with a 110 pitch count in a 2-1 game in the bottom of the eighth in late September when the Tigers are only in the lead by a game could be justifiable when most objective models would suggest taking him out.
That being said, many managers can improve simply by not having their pitchers throw unnecessary innings, like the seventh inning after their team has taken a 10-1 lead.