Not All Whiffs are Created Equal

A few weeks ago, I took a deep dive on Luke Weaver, budding Cardinals No. 2 and Adam Wainwright torch-taker. I came out of it reluctant to invest in him at his hefty asking price: per National Fantasy Baseball Championship (NFBC) average draft position (ADP) data, 27th among starting pitchers and 105th overall. His lackluster swinging strike rate (SwStr%) indicated to me he could not sustain his lofty 28% strikeout rate (K%) when pitchers of similar SwStr% caliber had strikeout rates ranging from 18% at worst to 25% at best. The best, in that instance, was Aaron Nola. I doubted Weaver could compare so favorably, primarily because Nola steals strikes in a way few others can.

What I failed to do — what I should’ve done — is check. Duh! I should’ve checked how often Weaver earns called strikes. So I did, and I came away feeling even more scared than before. Turns out, I’m an idiot who botched some simple arithmetic: Weaver falls just outside the top quartile of pitchers in stealing strikes, as opposed to literally 3rd-worst like I claimed in the previously linked Tweet. That would’ve been really bad. Still, this miscalculation and subsequent mischaracterization of Weaver’s ability sent me on a quest of ultimately marginal value.

The more I looked at Weaver’s plate discipline peripherals, the more I realized Weaver doesn’t coerce many swings-and-misses out of the zone. He’s actually pretty bad at it. That means he generates most of his success in the zone, which, to me, seemed unusual. I was nervous that thriving by threading the needle — i.e., pounding the zone (assuming some semblance of command) and hoping for whiffs — might be a dangerous way to live, or that, at the very least, a young pitcher for whom the proverbial book on him isn’t out yet might be subject to regression in this particular area. The freshly FanGraphs-retired Eno Sarris thought thriving in the zone is better than thriving outside the zone:

wouldn’t a really great zone whiff rate be better than getting your whiffs outside of zone? harder to deal with as a hitter?

— Eno Sarris (@enosarris) February 1, 2018

Naturally, he was right.

Weaver makes up for ineffectiveness out of the zone with effectiveness in the zone. I sought to find a quick comp against whom I could compare Weaver’s unique plate discipline peripherals. This rabbit’s hole produced Tony Cingrani, who, with strong in-zone stuff, weak out-of-zone stuff, and a pedestrian whiff rate (9.9%, like Weaver’s 9.7%), could strike out 25% of hitters. It’s not quite Weaver’s 28%, but it suggests Weaver might not regress as much as I suspected. It also suggested to me not all whiffs are created equal, the conclusion to which I have already spoiled.

Using data from all qualified player-seasons by pitchers from the last 10 years (818 in all), I regressed against strikeout rate (1) whiff rate and, separately, (2) whiff rate as in-zone and out-of-zone components. Each produced nearly identical adjusted R²s (0.727 and 0.728, respectively) but, as expected, pretty different outcomes:

(1) xK% = 0.018 + 1.955*[SwStr%]
(2) xK% = 0.017 + 2.148*[Z-SwStr%] + 1.847*[O-SwStr%]

(The “Z-” and “O-” prefixes denote in-zone and out-of-zone. Edit: Shame on me for not being clearer about these calculations, which do not show up organically on the leaderboards. Please refer to this comment below. Also note that Z-SwStr% + O-SwStr% = SwStr%.)

Using the estimated coefficients, we can conclude whiffs in the zone are a little more than 16% more valuable than whiffs out of the zone. This makes no assumption of or judgment about the sustainability of either path to success. However, assuming every pitcher owns his skills and that their outcomes are true representations of those skills, and holding all else constant, a pitcher who records a 10% Z-SwStr% and 0% O-SwStr% should notch a strikeout rate 3 percentage points higher than a pitcher who records a 0% Z-SwStr% and 10% O-SwStr%.

Ironically, I expected these results to really move the needle for me on Weaver. His SwStr% “ratio” leans much more heavily in favor of in-zone whiffs, after all. So I re-calculated his expected strikeout rates using each regression-based formula:

(1) 20.8%
(2) 21.2%

Uhhhhh. Well, here’s a pull-out quote from my previous post:

I see a more likely mediocre-case scenario in which Weaver produces something like a 21% strikeout rate

Maybe I wasn’t so far off after all.

It’s not a matter of nonlinearity, either. I tested a third regression that included quadratic (squared) terms of both Z-SwStr% and O-SwStr% to test nonlinearity; it produced a negligibly better 0.729 R² and an expected strikeout rate of…

(3) 21.3%

Make of it what you will. Weaver is above-average at stealing strikes, so I’ll give him the benefit of the doubt that his K% will clock in above 21%. And all this says nothing of exploring (using multiple regression or other strategies) how the many fragments of plate discipline (such as SwStr% and, say, first-pitch strikes (F-Strike%)) interact with one another. But since Weaver trails Nola in both total whiffs and stolen strikes — Nola, who strikes out 25% of hitters — I have a hard time conceding that Weaver will do more than split the difference between the regression equations and Nola.

Still, two or three additional ticks to his strikeout rate would improve my bearish outlook for him markedly. I squint and I see a top-30 starting pitcher, but I have this sinking feeling we’re overestimating his ceiling, especially in light of a career Minor League K% that never cracked 25% (i.e., against suboptimal competition).