On Zack Wheeler and pitch counts—I know, I know. "Rob, stop. It's already dead." Hear me out; there's a little more meat left on this bone.
People are discussing these two topics, specifically the confluence of said topics, quite a bit lately. Typically the context of these discussions goes something like, "Did the Mets push Zack Wheeler too hard?" I've already shared how I feel on the matter, and on pitch counts as a barometer for pitcher abuse in general. (Out-of-context spoiler: Not buying.)
That said, there has been a subset of these discussions that go a step further. Rather than indiscriminately pointing to pitch counts at the per-start level, some are taking more of a micro view of the metric and wondering about pitch counts per inning.
Metsblog's Bryan Mangan has been very vocal on this topic, taking to Twitter with questions on Wheeler's per-inning usage and ultimately penning a piece at The Read Zone entitled "No Pitcher Has Thrown As Many Pitches Over as Few Innings as Zack Wheeler This Millenium."
In short, he keys in on a pair of metrics: total pitches and innings pitched. By dividing the former by the latter, he quickly throws light upon the fact that Wheeler has a higher average-pitches-per-inning value than just about anyone in recent memory:
"[S]ince 2000, no pitcher has been pushed to throw as many pitches over as few innings as Zack Wheeler was."
Mangan subsequently slices the data a few different ways, but this is, for all intents and purposes, the core of his argument. As it should be; the statement that Wheeler has been highly inefficient—in terms of pitches per inning—is irrefutable. It is also a more precise way to review Wheeler's effort on the mound than with pitch counts alone.
However, there are some assumptions here that still make me uncomfortable. Specifically, while this methodology brings us from the start down to the inning grain, there is still no distinction between those innings. This essentially means means we're assigning the same value to each inning. But that doesn't feel right to me, based on what I know from watching Zack Wheeler pitch.
Like all pitchers, Wheeler's innings are not created equal—he has easy innings and tough innings. But more interesting to our exercise is where those different flavors of innings fall. Basically, I want to see the shape of Wheeler's innings pitched within a start.
It doesn't take much digging through Wheeler's 2014 game logs—data gathered via Brooksbaseball.net and Baseball-Reference.com —to unearth the first interesting nugget: Zack Wheeler tends to work less hard as the start progresses. Take his start last September 19 against Atlanta as an example:
Wheeler averaged 21 pitches per inning over those first three frames before averaging just 14 pitches per inning over the final three. That is not an insignificant variance. Wheeler pretty clearly found his groove mid-way through the start.
For Wheeler, this game is not atypical. He averaged 18.3 pitches per inning between innings one and three in 2014, but just 16.5 pitches per inning between innings four and six. (I'm only looking at the first six innings because the final three do not contain a representative sample—he didn't often pitch deep enough. [1-3, 1,735 total pitches | 4-6, 1,372 total pitches | 7-9, 177 total pitches])
This first finding is slightly unexpected, but we find a more surprising aberration when we look at the shape of Wheeler's effectiveness in those innings. Specifically, Wheeler's run-prevention also improves as his starts progress.
While the first trend was odd, this one goes very much against the grain of the aggregates for starting pitching in today's game. When a pitcher is exposed to an opponent a second or third time in a game, run prevention, on average, becomes more difficult—this is common knowledge.
But Wheeler outdoes the league as a whole by almost a full run, as measured by ERA, when comparing his first three innings with his next three. And that irregularity only grows wider when Wheeler is compared to his age-peers:
The comp-set of peers consists of 20 starting pitchers that were 24-years-old or younger in 2014 who made at least 12 starts that lasted into the sixth inning but also did not make more than a single relief appearance (as that would skew the run prevention data): Madison Bumgarner, Jake Odorizzi, Drew Hutchison, Sonny Gray, Yordano Ventura, Julio Teheran, Shelby Miller, Danny Salazar, Trevor Bauer, Gerrit Cole, Jarred Cosart, Nathan Eovaldi, Henderson Alvarez, Brett Oberholtzer, Michael Wacha, Jordan Lyles, Kevin Gausman, Tyler Skaggs, Tyler Matzek, and T.J. House.
So what we've learned by looking at the shape of his starts is that Zack Wheeler is an anomaly. Contrary to league splits, the 24-year-old righty consistently improves as the game goes on.
Who knows what this means for his long-term pitching stock, but, to me, there is a clear lesson to be gleaned from this information as it relates to the general consensus regarding Wheeler's relatively high pitch counts. Take this recent tweet, for example:
The difference between 74 and 78 wins to the franchise was nothing. To Collins it was probably his job. That's why Wheeler gets pushed hard.— ihateprospects (@ihateprospects) March 16, 2015
The sentiment here has been pretty common in the wake of the Wheeler injury. Note the operative word: "pushed." That same word appears in the quote from the Mangan piece up above. The inference is that Wheeler must only have reached these pitch totals because he was pushed by a short-sighted manager. But does that really jibe with the story that the data is telling us?
The data confirms that Wheeler is indeed racking up more pitches, per inning, than we'd hope to see. However, he's generally doing the bulk of that damage early in games. By the time the odometer rolls to 100 pitches, Wheeler is, more often than not, cruising. So, logically, is he being pushed? That characterization doesn't make a lot of sense.
To my eyes, manager Terry Collins is almost being pulled to these pitch counts by an anomalous pitcher consistently doing his best work around the time that the rest of the Mets' starters usually get the hook (Collins' staff averaged 97 pitches per start in 2014; the league average was 95). It's tough to make a case that it's in the best interest of the team or the pitcher to promptly call the bullpen as soon as Wheeler hits his stride. (It brings to mind the recent quote from Sandy Alderson—in reference to handling Zack Wheeler—where the general manager described managing somebody "to the point that he's not useful to you.")
Certainly it's incumbent on Collins and his coaching staff to know when to pull the plug, regardless of whether a pitcher is having success. But per last week's companion piece to this one, contrary to popular belief the manager is just not failing in that regard. Wheeler has never exceeded 120 pitches in a game since joining the Mets organization and to parse the potential for damage done below that total is a statistically unsubstantiated exercise in splitting hairs.
What we are left with is a highly talented, highly irregular young pitcher . . . who got hurt. Was that injury the fault of the manager or the organization? I've made my case that it was not. That obviously doesn't mean you can't disagree. It just means that if you do, you're running out of data-centric legs to stand on.