February 29, 2004
Baseball Prospectus Basics
Just Another Out?As we've stated on a number of different occasions throughout the Baseball Prospectus Basics series, one of the goals of performance analysis is to separate perception from reality. Sometimes that means interpreting numbers, and sometimes that means interpreting events with our eyes. Either way, it's about collecting information, and getting a little bit closer to the truth.
Evaluating the importance of strikeouts, especially for hitters, is something that has traditionally fallen into the second category. And it's easy to understand why: baseball is a game that centers around the ongoing conflict between batter and pitcher, and there are few outcomes that capture the drama of that conflict better than a mighty whiff, followed by a long walk back to the bench. On the surface at least, a strikeout appears to be the ultimate failure for a hitter-infinitely worse than a Texas-leaguer or a flyout to center.
From a quantitative perspective, however, there is little evidence to suggest that a strikeout is "worse" than a groundout, popout, or any other means of making an out, with respect to generating runs. Sure, it might look bad-not even being able to put the ball in play-but the fact is that error rates, in this era of improved equipment, are as low as they've ever been. Granted, putting the ball in play, whether in the air or on the ground, can sometimes enable a hitter to advance a runner, but it also increases the chance of hitting into a double-play-a far greater rally-killer than a strikeout.
As a result of all that, the value of "just putting the ball in play" is as low as it's ever been. The following graph illustrates the correlation-or lack thereof-between team strikeouts and team run scoring from 1950-2002:
As you can see by the round, lifeless blob in the middle of the graph, there is virtually no positive correlation between a team's strikeout totals and its runs-scored totals. When it comes to offense, an out is an out is an out.
On an individual level, the evidence against strikeouts as the scourge of the earth only gets more damning. Check out the correlation between Ks and the various elements of offensive production:
Correlation of SO/PA with (all players 1950-2002, 300+ PA)
Metric Correlation ---------------------- ISO +0.388 SLG +0.198 BB/PA +0.125 OBP -0.100 AVG -0.290 OPS +0.106 MLVr +0.005While it might not be overwhelming, there is a distinct, positive correlation between an individual's strikeout rate and a number of useful attributes: hitting for power-as represented in this case by isolated power (ISO, or slugging percentage minus batting average) and slugging percentage (SLG)-as well as drawing walks-as represented by walk-rate (BB/PA). Of course, causation is a sticky subject, so try not to misinterpret the above data as "proof" that increased strikeouts cause an improvement in a player's secondary skills. It's just that where one group shows up, often so does the other.
Notice, also, the virtually non-existent (albeit positive) correlation between strikeout rate and "complete" measures of offensive performance like on-base plus slugging (OPS) and Marginal Lineup Value Rate (MLVr). No matter how you slice it, it just doesn't appear that strikeouts have much of an effect on a team's-or an individual's-ability to produce runs.
But those are hitters. Pitchers, on the other hand, are a completely different story.
Where the value of "just putting the ball in play" has often been overstated for hitters, the opposite has long been the case for pitchers. In their case, a strikeout is most definitely not "just another out." In fact, the ability to create outs for one's self is among the most important skills a pitcher can possess.
Why? There are a number of reasons, but mainly it's because more strikeouts mean fewer balls in play. Fewer balls in play mean (on average) fewer hits surrendered. And with fewer hits surrendered come fewer runs allowed. The steps aren't perfect, mind you, but on a macro level they hold up. The following graph illustrates the correlation between individual strikeout rate and ERA from 1993-2002:
Or, to perhaps give this conclusion some real-world resonance, look at the disparity in ERA between those pitchers with the highest strikeout rates in the league in 2003 and those at the bottom of the barrel:
Pitcher SO/9 ERA ------------------------------------ Kerry Wood 11.35 3.20 Mark Prior 10.43 2.43 Curt Schilling 10.39 2.95 Pedro Martinez 9.93 2.22 Javier Vazquez 9.40 3.24 Pitcher SO/9 ERA ------------------------------------ Joe Mays 3.46 6.30 Danny Graves 3.20 5.33 Aaron Cook 3.12 6.02 Kirk Rueter 2.51 4.53 Nate Cornejo 2.13 4.67The difference isn't accidental. In a nine-inning complete game, Kerry Wood is roughly 30% less reliant upon his defense to convert batted balls into outs than someone like Kirk Rueter or Nate Cornejo would be. That's not just a huge difference, that's a Marlon-Brando-pulling-up-a-chair-to-the-buffet difference.
Strikeout rate also has predictive value. According to a study conducted by Keith Woolner, pitchers with high strikeout rates age better than comparable pitchers (i.e., pitchers who posted similar park-adjusted ERAs at the same age) with low strikeout rates. Bill James also gave this subject some treatment in his most recent edition of the Historical Baseball Abstract when discussing Mark Fidrych, and came to a similar-if slightly hyperbolic, as Tommy John can attest-conclusion: "There is simply no such thing as a starting pitcher who has a long career with a low strikeout rate."
The prominence of the strikeout in Major League Baseball has been increasing steadily over the past 130 years, and it may continue to grow as teams begin to let go of their macho attachment to "just putting the ball in play" on offense, while further valuing pitchers who are self-sufficient on the mound. Like many other developments in baseball, this will be a sign of evolution, and a better game overall will be the result.
Don't fear the strikeout. In many ways it is a harbinger of better things to come.