Introduction: in conjunction with an internship that I am currently working on with the pitching consulting company, connected performance https://www.connected-performance.com/ I am helping them build the framework for their own stuff+ model to use to evaluate their clients and use data to help determine ways to improve their pitching repertoire. To effectively measure pitching performance, selecting appropriate dependent and independent variables is crucial. This post presents a thoughtful analysis to determine the best variables for an in-depth evaluation of pitching prowess.
Abstract:
This post aims to explore the most suitable dependent and independent variables for a quantitative analysis of pitching performance. After considering various options, the dependent variable is suggested to be overall run values, which better represents a pitcher's effectiveness. For independent variables, we propose a set of essential factors commonly used in stuff+ models. By following this approach, we can assess pitching performance in a consistent and comprehensive manner.
Dependent Variable:
After careful consideration, the dependent variable will be overall run values. This choice captures a pitcher's ability to prevent runs and aligns well with the overall objective of pitching effectively. While considering the potential year to evaluate, using 2021 data seems reasonable due to its closer run environment to 2023, with minor adjustments to account for any changes in run expectancy.
Independent Variables:
The independent variables in this study will be chosen based on their significance in various stuff+ models. The list includes:
a. Velocity: The speed at which a pitcher throws the ball.
b. Vertical Release Height: The vertical position of the ball's release from the pitcher's hand.
c. Vertical Break: The upward or downward movement of the pitch.
d. Horizontal Break (Absolute Value): Accounting for left- and right-handed hitters.
e. Horizontal Release Height : Influencing the hitter's perspective.
f. Difference in Break (Vertical - Horizontal): Eliminating dead zone fastballs.
g. Extension: The distance a pitcher's release point is from home plate.
Data Considerations:
When computing run values, using 2021 data is recommended, considering its similarity to the 2023 run environment. While there might be slight variations due to changes in stolen bases, this is a reasonable approximation. However, if extending the analysis to college baseball, a different run expectancy matrix may be needed, contingent on data availability.
To interpret results effectively, multiply the computed run values by 100, consistent with how Baseball Savant presents the data.
In conclusion, this post establishes overall run values as the most appropriate dependent variable for evaluating pitching performance. The selection of essential independent variables ensures a comprehensive and robust analysis. By adopting this approach, researchers can gain valuable insights into pitchers' abilities and better understand their contributions to the game.
留言