April 26, 2011
The Minnesota Twins have the somewhat unique distinction of being left wanting by Francisco Liriano’s performance—not just his early-season struggles, but his body of work as a pitcher. Manager Ron Gardenire told reporters:
Here’s the question—does pitching to contact allow a pitcher to get deeper into games? And if so, is there any cost, in terms of runs allowed, that offsets the gains of pitching deeper into games?
Let’s take a look at it through the lens of K/9. Why K/9, you may be asking—well, K/9 is just strikeouts per out times 27, which makes it a very good measure of how a pitcher gets his outs, independent of how many outs he gets (this will become important in a second). Let’s look at the relationship between K/9 and pitches per plate appearances, examining pitchers with more than 120 IP from 1993 to 2010:
Yes, utterly shocking, I know. Strikeout pitchers will throw more pitches to the batters they face than other pitchers. (The correlation coefficient, if you’re wondering, is .49.) And it’s pretty simple to see why this is the case—a strikeout requires at least three pitches, while an out on contact requires only one.
So case closed, right? Pitching to contact allows a pitcher to end a plate appearance with fewer pitches, which should help pitchers get deeper into games, right?
But we’re missing a key piece of information—how many outs a pitcher records per plate appearance. Looking at how K/9 correlates with outs per PA:
What’s interesting here is that by using K/9, we’ve chosen a measure of a pitcher’s strikeout ability that is not intrinsically correlated with his total number of outs—if we’d used K/PA instead we’d see a higher correlation. (As it is, we see a robust .44, nearly the same as what we saw for K/9 and pitches per PA.) What the data is telling us is that while strikeout pitchers do in fact throw more pitches per batter they face, they can make it through an inning while facing fewer batters. In terms of pitches per outs, then, the two effects counterbalance each other to some extent—we see a modest .12 correlation between K/9 and pitches per out.
So what does this mean for Liriano? Let’s take a look at outs per contact and outs per noncontact to see how they correlate with rates of contact and noncontact outcomes per plate appearance:
The red blob is outs per noncontact events—K/(K+BB), in other words. What we can see is that there is a weak but real relationship (R of .24) between a pitcher’s (K+BB)/PA rate and his K/(K+BB) rate, so pitchers with higher noncontact rates will tend to have more strikeouts per noncontact than other pitchers. We see no significant correlation, however, between outs per contacted ball and the number of contacted balls a pitcher allows (R of -0.03, so not only is it not significant, it actually runs the other way: pitchers who allow less contact appear to get more outs on contact, although again, this isn’t significant).
But what about Liriano in particular? Is he the sort of pitcher who would benefit from pitching more to contact? Well, for pitchers in the study, we see a .70 outs per contact rate, compared to .66 for outs on noncontact. In other words, assuming that everything else stays the same, the typical pitcher may in fact benefit slightly from pitching more to contact (that, of course, being a rather generous assumption, given the correlation we see). But for Liriano, while we still see a .70 rate of outs per contact, we see a .73 outs per noncontact rate. So, if Liriano begins pitching to contact more (assuming his K/(K+BB) rate does not increase as his contact rate increases, which I think is reasonable given the above), he’s going to see an increase in the number of batters he has to face to get through an inning. Rather than letting him pitch deeper into games, it’s going to make him less efficient in terms of batters he faces. Rather than making him more of a pitcher, it would probably just make him a more ordinary pitcher.