﻿ Preliminary aging curve for fastball speed | The Hardball Times

# Preliminary aging curve for fastball speed

Understanding how players age is very important when examining them and their careers. It is essential when people try to look in the crystal ball to determine just how good a hot prospect could be in a few years or if an aging star might have a good year or two left in him. Aging curves have been created for everything from hitting to pitching and even defense. While the aging curves that have been produced are very accurate, they tend to be high on the hierarchy of statistics.

What I mean by this is we have accurate aging curves for things like ERA, but not for things that contribute to ERA like the speed of a pitcher’s fastball or movement on his curveball. This is where the PITCHf/x data come in. By tracking the pitches as they travel toward home plate, I hope we can create aging curves for variables that contribute to things like ERA. In the next few sections, I will be describing how I am creating aging curves with the data. While I have tried to explain the process as simply as I can, there is some advanced math involved. If this isn’t your thing, please feel free to skip ahead to the results.

###### Method

Now that initial 2008 corrections have been made to the PITCHf/x data, we can start to really compare the 2007 and 2008 data. When I say fastballs here, I really mean fastballs and not sinkers, cutters or splitters. I begin by calculating the average speed of each pitcher’s fastball and the variation of the speed for each year. I then subtract the two averages and combine the variations by summing the squares. This gives me a difference between the years and an error for each pitcher. I then can combine pitchers of the same age by taking a weighted average. The ages I am examining here are pitchers who were between the ages of 24 and 33 during the 2007 season. For this study I am using pitchers who have thrown at least 100 fastballs in both 2007 and 2008. I have 143 pitchers in that age range who qualify for the study.

###### Potential issues with these data

Unfortunately, some unresolved issues may affect this study. First, obviously, I don’t have a full year of data for either 2007 or 2008. Through last Sunday’s games, PITCHf/x has tracked nearly 200,000 pitches, which actually is a pretty large fraction of the slightly more than 330,000 pitches it tracked last season. Because I am using weighted averages, though, I am calculating the final error for each age, so while lack of data is an issue, it should be visible in the results.

Potentially the biggest issue is that most of the data from 2007 were recorded in the latter part of the year and in 2008 all of the data are from the beginning of the year. Because I am taking the difference, this isn’t a problem as long as both younger and older pitchers tire at the same rate during the year—it isn’t obvious if either group might tire more during the year. Also, there is a bias in the data because only pitchers who were successful enough in 2007 are now pitching in 2008. Lastly, there have been some questions about comparing data year to year with data corrected using my correction code. Because of the way I am calculating my corrections, I am potentially exposed to changes in the PITCHf/x system from year to year. Loooking at some averages from 2007 and 2008, though, it appears that there have been no major changes in the system. I’ll need to stay on top this as the data roll in.

###### Results

Okay, finally we are ready to look at the results.

I am comparing everything here to 24-year-old pitchers as my baseline. Because of this, there aren’t any error bars in that bin. Each following year is compared to the previous year as described above. It appears that until pitchers reach 28 or 29, they increase the speed on their fastball by about 1.5 mph. After 29, there is a rather sharp decline in fastball speed.

During the next five years, pitchers lose just over four mph. I am a little surprised that the peak is as old as it is. By the time pitchers are in their late 20s, they have thrown a huge number of pitches and have a lot of wear on their arms. The drop when pitchers turn 30 seems very rapid to me. Because pitchers’ aging curves tend to be much flatter than hitters’, this rapid decrease must be compensated for by some other factor.

###### Conclusions

While there still are some things to work out, the initial results look quite promising. As more data become available, things like aging curves will become better and better. I hope that soon the aging curves created using PITCHf/x data will be good enough to help predict things like ERA and strikeouts per game. With the addition of HITf/x, high-quality data for hitters are not far away. These data will help uncover some of the underlying fundamental quantities of the game that, up until now, have been hidden to us.