A prior version of this post was published earlier in the day. The current version has been updated to correct omissions in the data. As it pertains to the notion of an injury nexus related to 50% attrition, the conclusion is not altered.
Over the last two days, I’ve expressed some minor concern about Michael Pineda, but that sentiment was based solely on very limited data. More specifically, the reduced radar gun readings from Pineda’s first outing piqued the worry-prone side of my baseball psyche. However, considering the myriad of plausible explanations for this very limited data point, chances are there is probably nothing to see here…at least not until the next time Pineda takes the mound.
In a recent blog post, however, Will Carroll, who gained notoriety for his work tracking injuries at Baseball Prospectus, suggested that concern over Pineda was not only warranted, but a given. Carroll’s pessimistic view wasn’t based on early radar gun readings, but rather a 2002 study he conducted with Nate Silver, who has since gone onto to greater fame as a political prognosticator for the New York Times. Although Carroll isn’t noted as a statistical guru, Silver’s reputation in the field is strong, so, any study claiming to show an injury trend among young pitchers with his name attached is worthy of closer inspection.
Although the original post detailing the study included a methodology, it did not provide the data. Perhaps that’s just as well because some of the parameters didn’t seem appropriate anyway. In particular, Carroll and Silver lumped together all seasons from 1946 to 2002, which, considering the many changes in the game during that span (e.g., usage patterns, training techniques, rule changes like the DH and slope of the mound, etc.), probably skews the analysis toward being an “apples to oranges” comparison.
The basic finding of the Carroll/Silver study was the attrition rate (defined as throwing 50% fewer innings in a subsequent season) for pitchers throwing at least 150 innings in a season was much higher for younger hurlers, particularly those below the age of 24 (the study only included pitchers with an ERA+ of at least 90 in an attempt to filter out innings declines caused by performance). Carroll’s recent post was based on that conclusion from 2002, prompting him to state “Pineda is below the ‘injury nexus’”.
Even if the data in the Carroll/Silver analysis is assumed to be accurate and relevant, the conclusion is hardly a strong one. After all, the population defined by the criteria includes only 457 seasons by pitchers under the age of 24. Because the sample is relatively small, a whole host of variables would have the potential to skew the data. What’s more, the attrition rates for age-22 and age-23 pitchers aren’t significantly higher than their more mature counterparts in the first place (click on the study link above for a graph depicting the rates), so unless restricted to a much a younger subset, the conclusion seems to be an exaggeration.
In an attempt to replicate the study using more recent data, I pulled together every season of 150 or more innings thrown by a pitcher under the age of 24 (using “season age”, as reported by baseball-reference.com). In order to increase the sample, however, pitchers with an ERA+ lower than 90 were not excluded (an alteration that would bias the results toward Carroll’s conclusion), but specific seasons were removed if the innings declined was related to obvious ineffectiveness, a pitcher being converted to a reliever, or other extenuating circumstances (e.g., Rick Ankiel’s control issues, Dwight Gooden’s drug suspension, etc.). So, what does this more recent snapshot say about the “injury nexus”? Well, it appears as if the “line in the sand” has been washed away.
Note: Based on a population of seasons (244) by pitchers throwing 150 or more innings before turning 24 (using season age) minus 19 specific exclusions. Attrition rates defined as subsequent seasons with fewer innings equal to or greater than the percentage cited.
Since 1982, there have been 244 seasons in which a pitcher under the age of 24 threw at least 150 innings. Unlike the Carroll/Silver study, which found attrition rates above 20% for pitchers age-21 and 22, this more recent sample yielded much lower percentages. In fact, only eight of 98 pitchers age-22 and younger experienced an innings decline of at least 50%. What’s more, even lowering the threshold to 25% failed to produce rates for this segment as high as those suggested by the Carroll/Silver analysis. Of course, without comparing the rates in my analysis to the entire pitcher population, we can’t make a relative comparison, but based on the data from the under-24 age group, if there is a link between age and “high attrition”, it is either very small, or the threshold is much lower than suggested by Carroll/Silver.
Note: Based on a population of seasons (244) by pitchers throwing 150 or more innings before turning 24 (using season age) minus 19 specific exclusions. The rate of change compares the number of innings thrown in an eligible season to the number of innings thrown in the following season.
What does this mean? Perhaps the changes in how pitchers are developed in the minors and utilized in the majors have mitigated some of the risks involved with throwing a lot of innings at a young age? Or, maybe the inclusion of older data in the Carroll/Silver study skewed the numbers even more than I first thought? Regardless, there is one clear conclusion: based on the limited data involved, there is no way to statistically define a pitchers’ “injury nexus”, at least not based on the parameters cited above.
Now that we have determined Michael Pineda isn’t destined for injury because of his age and the number of innings he threw in 2011 (the Carroll/Silver study set 150 innings as the red flag barrier, but in his recent post, Carroll cited 190 innings; Pineda’s 171 innings in 2011 falls right in the middle), does that mean he is out of the woods? Not yet, according to Carroll.
Aside from the “injury nexus”, Carroll also suggests that Pineda will have to survive Joe Girardi’s poor record when it comes to keeping young pitchers healthy. To support this claim, Carroll points to five examples during his Yankees’ tenure. Interestingly, Carroll only names four pitchers in his analysis, which could be because Joba Chamberlain qualifies twice under his criteria, but either way, the sample is once again too small to yield a meaningful conclusion. Besides, two of the four (Ivan Nova and Ian Kennedy) are healthy and coming off very strong seasons, while the other two (Joba Chamberlain and Phil Hughes) debuted before Girardi’s tenure (and, in Hughes’ case, developed an injury beforehand) and have been the subject of organization-wide monitoring. Carroll uses a similar sample of five pitchers to further damn Girardi’s handling of young pitchers with the Marlins, but again overlooks mitigating factors. To Carroll’s credit, he concludes by saying Girardi (and Larry Rothschild, whose handling of young Cubs’ pitchers is called into question) doesn’t deserve blame, but by that point, it is too late.
Carroll’s recent musings conclude by highlighting Joe Maddon as a Good Sheppard of young pitchers. However, by doing so, he disregards his own parameters. For example, Carroll cites Matt Garza and James Shields as two pitchers “brought through the injury nexus” by the Rays, even though they both debuted with Tampa at the age of 24 (i.e., the age with the lowest attrition rate in the Carroll/Silver study). Also, Maddon is given credit for protecting the arm of Jeremy Hellickson, but, considering 2011 was his first full season in the majors, such a conclusion seems premature.
Circling back to Pineda, it’s only natural for Yankees’ fans to have some concerns about the young right hander, but they should be tempered by common sense. Two innings worth of radar readings are not a definitive sample, nor is there evidence that Pineda is at risk because of an “injury nexus” exacerbated by Joe Girardi’s handling of young pitchers. If anything, history seems to be on Pineda’s side, but ultimately, his ability to stay healthy will depend on a variety of factors, many of which are specific to the 23-year old and impossible to predict.