Verducci predicts Nats fans may be reliving this ugly moment. Photo credit unknown.
SI sportswriter Tom Verducci published his 2013 iteration of the “Verducci Effect” article this week, and our own Stephen Strasburg is on his watch list.
The questions we’ll address in this article are these: Should we be worried as Nats fans about Strasburg? And, do you even believe in the Verducci effect?
Cutting and pasting from his article, here’s the Verducci Effect defined (or the “Year-After Effect” as he calls it, so as humbly not to refer to the theory by his own name as the rest of us do):
The Year-After Effect, as I called the risk after a big innings jumps, is not a scientific, predictive system. It’s a rule of thumb to identify pitchers who may be at risk because of a sharp increase in workload. The older the pitcher, the bigger the body type and the closer to the 30-inning threshold is their increase, the less they seem to be at risk.
Of importance here is his own admission; this is a “rule of thumb,” not a scientific analysis, he limits his candidates to pitchers age 25 or younger, and he picks players instead of doing an across-the-board analysis of all eligible players (this is important as we talk about whether the effect is statistically supported). He identified 14 such players in his 2012 iteration of this analysis and 9 of them suffered injuries or “significant regression.” He has a similar track record in his previous years; of the 69 pitchers he’s identified in the last 7 years as being at risk, 55 of them have now suffered injury/posted significantly worse ERAs. That’s about an 80% succeessful prediction clip.
Strasburg pitched 68 MLB innings and another 55 1/3 in the minors for a total of 123 1/3 2010 innings. He threw 159 1/3 in 2012, for a total increase of 36 innings from his previous professional high, or a 22.6% increase. Generally speaking Verducci’s threshold is in the 15-20% range. Strasburg isn’t the most “at-risk” pitcher on the list; that would be Chicago’s Chris Sale, who amazingly threw 121 more innings in 2012 than he ever had before, being converted from the bullpen to a starter last year. But Strasburg definitely increased his workload in 2012, and he’s likely to be increasing it again in 2013, with no stated limit on innings for the coming season. If he averages the same number of innings per start this year as he did last year (5.69) he’ll end up with roughly 187 2/3 innings in 2013, which would be another 15% increase over his previous career high. Most likely we’ll see him averaging closer to 6ip per outing, which would put him at about 200 innings and representing another 20% increase in innings.
Do we think this is dangerous territory? Should we be worried? All signs point to “No;” there’s not a person in the baseball world who would claim that the Nats have been anything but ultra-conservative with Strasburg since the moment he was signed. His surgery, his recovery, and especially the heavily criticized “shut-down” in September of last year. The team chose to be less competitive in the short term in order to attempt to be a better, stronger team in the long term. Given his gradual ramping up of innings and his carefully managed recovery, I expect to see a similar season that Jordan Zimmermann just gave, two years removed from the same surgery. No injuries, a strong season but with some evident fatigue at season’s end.
A better question may be this: does the Verducci effect actually exist?
This January 2012 Deadspin.com article pretty nicely summarizes all the criticisms associated with the effect. A few more links are in this article at AmazingAvenue. This study done by The Hardball Times that looked at ALL pitchers age 25 or younger, divided the pitchers into two groups (those who did and did not throw 30 more innings and thus usually qualify for Verducci analysis) and studied the results. They found that the overall performance didn’t seem to be different between the two groups.
So, if the effect doesn’t exist, why does Verducci have an 80% prediction success rate? If the statistical differences between the two groups are identical, then why isn’t Verducci’s prediction rate closer to 50%?
The answer lies in the following statement: Verducci’s articles don’t present themselves to be a macro statistical analysis, and they doesn’t approach the problem in the same way that statisticians do. Instead, he finds candidates that qualify and then passes judgement based on his professional opinion about whether they’re a “watch candidate.” Which I think is a perfectly fine way to do an analysis piece like this. Of the 11 pitchers he selected this year, 4 are from 2012 playoff teams, another 2 are from teams in the ultra-competitive AL East, and 2 more experienced such extremely high jumps in innings that even a non-statistical observation would conclude they’re injury risks for 2013. I don’t think this kind of analysis is unreasonable frankly. He clearly “cherry picked” these candidates but for good reason; they were for the most part either severely driven or were pitching a lot of higher-than-average leverage innings for the bulk of the year, all the while throwing deeper into seasons than ever before.
Its a combination of statistics and opinion; most critics of the theory use 100% statistics to claim that the effect doesn’t exist. But that’s the rub that I keep coming back to when talking about the use of statistics in baseball; human behavior (aka, baseball players) doesn’t operate on a spreadsheet, and statistics cannot and will not entirely predict all situations in the future. You can’t just say that the effect doesn’t exist because you can’t prove it exists statistically. In this case, there’s clearly an analysis/opinion portion of the effect that takes into consideration immeasurable factors that (in Verducci’s opinion) lead to more stress and a higher probability of injury. Plus, Verducci admits that “body type” and age do factor into his opinion; meaning that a guy with a big body (and by inference he likely means that a bigger body that takes stress off the shoulder) is more likely to be able to weather a larger workload. Roger Clemens versus Tim Lincecum. I’d also assume he’s looking at mechanics along the way (and Verducci is on record for being critical of Strasburg’s mechanics, as I discussed in this March 2011 post). Statistics can’t measure mechanics, or body type, clearly portions of Verducci’s analysis.
Let me put it a different way: would anyone be the slightest bit surprised to see Sale come down with a shoulder injury in 2013? I certainly would not. And that’s the essence of the article, to provide a baseball opinion, and one of the reasons I still put stock into it while others waste time trying to disprove it.
Lets just hope Strasburg is one of the 20% he’s wrong about…