Strikethrowers and control freaksby Sal Baxamusa
August 13, 2007
Last week, we took a look at how many pitches a plate appearance lasts. When looking at the average number of pitches broken down by plate appearance, we found that walks took 5.7 pitches, strikeouts 4.8 pitches, and all balls in play (whether resulting in outs, singles, doubles, triples, or home runs) took 3.3 pitches. When we broke it down by distribution, we found that a ball in play was no more likely to be an out than a hit based on the length of the plate appearance.
Most of the comments that I received asked whether these trends hold when we look at different types of hitters or pitchers. Today, we'll take a look at pitchers who throw a lot of strikes and see if there are any differences. Perhaps pitchers who control the zone will throw fewer pitches per plate appearance across all categories, since they tend to throw lots of strikes.
Let's use the ratio of strikeouts to walks (the familiar K/BB) as a proxy for strike-throwing ability. We'll limit the sample only to pitchers with K/BB greater than three (your Scott Kazmirs, Jeremy Bondermans, Dan Harens, Billy Wagners, Ben Sheetses, etc.) and call them strikethrowers. First, let's look at the average number of pitches per plate appearance, broken down by result.
Result Strikethrowers League-Average BB 5.79 5.67 K 4.77 4.81 Hits 3.33 3.34 Outs 3.30 3.32 All 3.75 3.75 Outs includes only outs on balls in play Balls in play includes home runs
Basically no difference, except for the walks. Just because these guys throw enough strikes to have good K/BB ratios, it doesn't mean that batters are putting their pitches into play any earlier. However, it does look like it takes slightly more pitches for one of these strikethrowers to issue a walk.
Does the distribution of pitches per plate appearances look any different for these masters of the strikezone?
Very similar. The plot for strikethrowers peaks and falls off slightly more quickly, but the differences are practically imperceptible. If we break it down by result, we see that the slight difference stems from walks:
The distribution for balls in play only is virtually identical. Despite the fact that these pitchers throw enough strikes to post an impressive K/BB, hitters apparently don't feel compelled to swing any earlier. Drawing a walk off of these guys is tough, however: a smaller fraction of their walks are of the four and five pitch variety and a greater fraction take six or more pitches.
What if we used a slightly different definition of controlling the strike zone? Let's take a look at pitchers who issue walks to fewer than 5.0% of hitters faced (Jon Garland, Mike Mussina, Roy Halladay, Greg Maddux and company).
Result Control Freaks League-Average BB 5.92 5.67 K 4.79 4.81 Hits 3.25 3.34 Outs 3.25 3.32 All 3.61 3.75 Outs includes only outs on balls in play Balls in play includes home runs
As with the strikethrowers, a hitter really has to work one of these control freak pitchers in order draw a walk. They do tend to throw fewer pitches overall due to the fact that they walk very few batters. And because they walk so few, we can guess that they are throwing tons of strikes. Interestingly, balls in play tend to occur sooner for the control freaks than for the rest of the league. One might interpret this to mean that hitters feel compelled to swing earlier against this group of pitchers.
Distributions tell a similar story:
As with the strikethrowers, the control freaks really make hitters earn their walks. A smaller fraction of their overall walks take only four or five pitches; well over half require six or more pitches. Unlike the strikethrowers, however, their distribution for balls in play does show some difference from the league average. The difference is small, but it can be seen that hitters tend to put the ball in play on the first, second or third pitch a bit more frequently.
So do control freak pitchers compel hitters to put the ball in play earlier? If these data are to be believed, then the answer is yes. But the difference is very, very small. These pitchers conserve about one-tenth of a pitch per ball in play. A pitcher who has a 6 IP, 6 H, 2 BB, 4 K line would do so in 97 pitches using league-average pitches per plate appearance numbers. If we substitute the "control freak" balls in play number of pitches per plate appearance (but keep all else the same), it would take 95.5 pitches. It is a difference of fewer than two pitches in a typical game!
Do you ever hear analysts say that a pitcher goes deep into games because he stays around the strike zone, forcing hitters to put the ball in play early in the count? These data don't support that theory. If we accept that K/BB and BB% are reasonable proxies for pitcher's propensity to stay in the zone, then what is shown here is that these types of pitchers don't "force" batters to put the ball in play early in the count. If they last deep into games, it is because they turn plate appearances into outs, not that they end plate appearances quickly.
One way to contextualize this conclusion is to relate it to the theory of Defense Independent Pitching (also known as DIPs, and its cousins FIP, xFIP, and LIP). But first, a disclaimer: because even mention of the acronym DIPs has been known to start sectarian warfare among baseball fans, I want to make clear that these conclusions don't necessarily support DIPs. They are, however, consistent with what we might expect given the theory of defense independent pitching.
DIPs tells us that a pitcher has little control over the outcome of a ball put in play. If that is true, then it stands to reason that the hitter has a great deal of control over the outcome of a ball in play. These data would suggest that there is little a pitcher—at least the types studied here—can do to influence even the decision of a hitter to put the ball in play. That is, the pitcher doesn't simply lose control when the hitter puts bat to ball, he loses control the moment the ball leaves his hand (at least when viewed over the sample size of an entire season). That's a subtle but critical distinction.
This is all conjecture, of course—educated conjecture, but conjecture nevertheless. Scientific prudence demands additional information; specifically, we should check if certain types of hitters do show significant differences from league average in the length of their plate appearances. We'll get to that next time.
References and Resources
The data here are pitch-by-pitch data from 2006 as supplied by Retrosheet. Many thanks for all their hard work.
Sal Baxamusa is a graduate student in chemical engineering. He can be reached here.