Analysis on the MLB strike zone based on data on pitches and calls.
In the recent years, there have been debate and speculation over the possibility of a shrinking strike zone in Major League Baseball. Although only by a small amount - less than 1% from 2015 to 2016 according to an analysis by Jeff Sullivan (http://www.fangraphs.com/blogs/is-mlb-already-shrinking-the-strike-zone/) - there is evidence that suggests a slowly shrinking zone.
Over a one week session dedicated to baseball analytics, which culminated in a personal project that studied an aspect of baseball from data collected from MLB matches, I mapped the pitches that were called balls as well as strikes based on the position of each pitch. The game state was also taken into account, and each graph represents a different state, whether it be no players on the bases, a player only on first, players on first and second, etc. A grid was then created based on the area that contained a majority of strikes, mapping a strike zone on the graph. This process of graphing the pitches was repeated, with the grid placed in the same location on each graph.
The results suggest, based on a qualitative look at the plots, that the strike zone is in fact shrinking, even if it is only by a hair. This can clearly be seen on the graph that details the strike zone when there were runners on second and third. In comparison to the other plots, the area occupied by strikes is noticeably smaller than that of the graph with no runners on the bases. The shrinkage could be caused by umpires altering calls depending on the game state - it would be beneficial for a team to allow balls or walks to strike out "better" batters or prevent multiple runs. Nonetheless, even despite this data, there are numerous improvements and additions that can be made on this data analysis project, including:
- Determining if the shrink in the strike zone is statistically significant.
- Plotting other pitches on the plots and analyzing possible changes between game states.
- Gather more data - the shrinkage observed on the plot with players on second and third base can be due to a significantly smaller sample size than the sample size of the other plots.
If you would like to contribute or suggest improvements to this project, please reach out to me.