Command Statistics: A sortie into the unknown
Corey Kluber and Zach McAllister are, at first glance, quite similar pitchers. Each were received in a trade for role players (McAllister for Austin Kearns, Kluber for Jake Westbrook), neither of the two were widely-heralded prospects, and their 2013 ERAs were even within a tenth of a point of each other (Kluber at 3.85, McAllister at 3.75). Going into 2014, the first-glance predictions for each might therefore be similar.
Yet the perception of each in the saber community could not possibly be more divergent.
While Kluber strikes out substantially more batters than does McAllister, it's also true that Kluber walked far fewer batters in 2013 than McAllister (2.02 BB/9 for Kluber, 3.82 for McAllister). Given Kluber's excellent strikeout-to-walk ratio – in other words, given that Kluber excelled at those things which were most within his control – contrasted with his entirely average run prevention, Kluber's considered an underrated pitcher going into 2014. Taking the opposite tact, given that McAllister's strikeout-to-walk ratio was below-average contrasted led to an above-average rate of run prevention, the saber-oriented equally consider McAllister a candidate for regression.
Yet while the strikeout rates are easily accounted for by the rate at which each misses bats (career 7.6% swinging strike rate for McAllister, 10.4% for Kluber), their command is rather more difficult to pinpoint: as it stands - Baseball-Reference.com lists both pitchers as having 64.1% career strike percent. The strikes-per-pitch ratio, a widely-disseminated box score statistic says exactly nothing about command – only about control.
The control-command distinction is among the more widely-discussed scouting principles. Control is a baseline necessity for Major League pitchers – it refers only to the ability to consistently throw the ball over the plate. If a pitcher lacks control, said pitcher does not throw strikes at a particularly high rate. Zach McAllister and Corey Kluber, as evidenced by their identical strike rates, both have entirely adequate control.
It's worth noting that control is difficult enough – putting a 90+ MPH pitch consistently within a three-foot-square area is difficult as is; command, however, requires even greater precision. Command requires the ability to throw the ball – on demand – to any given segment of the strike zone. Pinpoint command doesn't necessarily mean the ability to hit a spot precisely – merely within, as Jim Bouton described it, the ability to throw the ball within a one-foot-wide box. Yet command allows one the ability to garner those paramount first pitch strikes and, given a competent game-caller behind the plate, put oneself in a position to prevent runs.
Yet while control is handily quantifiable, command is much more labor-intensive: there exists no readily-available statistic that quantifies whether a pitcher hit or missed his catcher's target, and current pitch-tracking technology tracks only where the ball actually goes – not where it's supposed to go.
It is with this in mind that, for Kluber's March 12th start against the Padres and McAllister's March 13th start against the Royals, the author tracked 'Targets:' the number of times that the pitcher threw the ball where the catcher called for it. These numbers are taken entirely from my own perception of whether the pitcher delivered the ball where the catcher wanted it – another might watch the game and derive rather different numbers.
Indeed, the flaws inherent in 'targets' are numerous and substantial: it's in large part arbitrary, it's binary (in the sense that there's no scaling of 'exact hits' to 'arguably hit the target'), it's impractically time-consuming on a large scale, it does not account for catcher ability, among numerous others. Yet while the deep flaws of 'targets' substantially limit the precision of the conclusions one can draw from the statistic, 'targets' is still a way of quantifying the otherwise unquantified: command.
The table below lists the number of pitches, targets, and total strikes both Kluber and McAllister threw in their respective contests:
Before drawing unduly ambitious conclusions about the above data, the sample-size caveats must be taken into account: namely, this was only one game, a spring training game at that, with all data being unofficial and collected by the author alone.
That said, there are two facts that are immediately apparent. The first of these is that, even despite his walks, McAllister was not unspeakably worse than Kluber as far as actual strikes thrown. While the gap exists and is substantial between the two outings, the incredibly small sample size, coupled with the career similarities between Kluber and McAllister in the matter of strike rates, might well lead one to believe that there is little substantial difference on the strike rate front.
This same similarity stands in sharp relief to the target rates – an extremely decisive gap. Given the imperfections and arbitrariness of the quantification, one might excuse a small gap out of hand, yet even factoring in that concern, the difference between these two pitchers is astonishing. While the author has not collected target data, nor does he intend to, for the league as a whole – meaning that, at present, one cannot situate McAllister or Kluber within the league as a whole – the above data seems to indicate that there exists a large gap between Kluber and McAllister, not only as far as missing bats, but also in each pitcher's ability to command the strike zone.
Kluber's command in his game against the Padres, combined with Kluber's excellent 2013 walk rate, gives the impression that Kluber has yet unfulfilled promise, promise that we saw before his finger injury but promise that did not translate to excellent run prevention. Kluber's entirely impressive command absolutely makes one believe that he is one of the most underrated players in the league as the 2014 season approaches.
On a broader note, even without the above data, strike rate is extremely limited as a metric. Strike rate, as a widely-disseminated statistic, is the easiest way of communicating a pitcher's control, but it's frequently and mistakenly perceived as a command statistic. Even excluding the above data, this misinterpretation is problematic. Strike rate includes not only called strikes, but also swinging strikes, foul balls, and even balls put in play. It's a perfectly serviceable way of communicating whether a pitcher was doing a 2012 Ubaldo Jimenez impersonation, but it reveals no more detail than that. While this is an implication not necessarily of the data, the flaws of strike rate illustrate a need for a quantification of command.
Nor, moreover, is first pitch strike rate, tracked by Baseball-Reference, necessarily an indicator of command – in 2013, Kluber had a lower first-pitch strike rate than McAllister. Additionally, while Baseball-Reference has a thoroughly detailed list of pitch summary statistics, none of them indisputably describe command.
While the above details an extant dearth in statistics, the following assertions are entirely more speculative and based on wholly unproven premises – the implications of which, however, are too enormous to not address.
First, it's not entirely the case that targets translated into strikes – there were several pitches which both McAllister and Kluber threw that hit the catcher's target, but were called for balls. This is not inherently a problem – chase pitches would be far less effective if thrown within the strike zone, and hitting the black but not getting the call does not indicate poor command.
Yet the latter of these two points illustrates how tightly-intertwined pitcher and catcher are. If we are given a pitcher with known command, we can with yet greater precision quantify catcher framing abilities – abilities which recent studies by Messrs Pavlidis and Brooks have illuminated in revelatory depth – as well as possibly take some step toward quantifying catcher game-calling abilities, hitherto limited to mediocre metrics like Catcher ERA.
Finally, one should return to the chart above and note that, while Kluber had an identical number of strikes as targets, McAllister had ten less targets than strikes – in this entirely microscopic instance, the relationship between targets and strikes differed between the two pitchers. If it turned out that the relationship between targets (command) and strikes (control) differed between pitchers, this relationship could be quantifiable; if the relationship between command and control were quantified across the spectrum of Major League pitchers, one might plot this correlation against other pitching statistics and thus determine whether the relationship between command and control was itself a factor in a pitcher's success – in the most wildly-momentous scenario, it would give us a new avenue with which to view pitchers.
While each of these two points are entirely premature and speculative, they remain open questions, which only better, more complete, and less-arbitrary data collection can either prove or disprove. In the more concrete sense, it gives a concrete ability to relate to the often broad strokes of the saber community – in this case, it helps offer some insight as to why Kluber's walk rate is superior to McAllister's, and in showing that, demonstrates why ambitions for Kluber's 2014 so far outstrip projections for McAllister.
The use of saber numbers, as iteratively articulated, is simply to give some quantity to those causative factors that lead to an effect: good or poor performance. These factors are pivotal no matter how one views the sport, be that factor quality of contact or swinging strike rate – or command, perhaps the single most fundamental pillar of the game.
John can be reached on Twitter at @JHGrimm. He can also be reached by e-mail at email@example.com.
Z-Mac was substantially below league average; Kluber, above league average. Consequently, Kluber had a better strikeout rate, and McAllister a worse strikeout rate.
This leads one to believe that McAllister's going to have an ERA closer to 4.25 than his 3.75 of 2013. Assuming he doesn't start whiffing more/walking fewer. It also leads one to believe that Kluber will have an ERA closer to 3.25 than 3.85.
Assuming the defense isn't one of the worst in the majors again, which it was in 2013. A bit of an adventurous assumption, to be sure
I seem to recall a lot of times last year where the catcher targeted a low fastball and McAllister came in with one at the letters and got a K.