Schilling’s aching shoulder

On June 18, Curt Schilling got cuffed around by the Atlanta Braves. He
lasted only 4.1 innings, surrendering 10 hits and six runs, walking two
and striking out nobody. The last time Schilling failed to strike out
a batter in a start was in 1993. Of course, anybody can have a
bad day, but this was clearly more than that.

It was obvious to everybody that Schilling’s velocity was down. And not
just down a mile or two per hour, but way down. Here’s what Braves
third baseman Chipper Jones had to say:

I think we all
expected to see 92, 93 (miles per hour) when he gets in trouble, or
even hike it up to 94, 95. The fastball that I saw register the
highest was 89, and that was with the bases loaded. Schill always had
that innate ability to catch another gear when he needs it, and for me
to only see 89 miles an hour tells me, you know, he might be hurt. I
don’t know.

Later, Sox beat writer Gorden Edes of the Boston Globe reported:

In Monday’s 9-4 loss, Schilling threw only one fastball that exceeded
90 miles per hour, a 91-mph pitch to Kelly Johnson with the bases
loaded in the fourth.

Schilling was sent back to Boston for an MRI and the results showed
“no structural damage.” Aside: What is “structural damage”? Is there
nonstructural damage that we should be worried about? Anyway,
Schilling says he hasn’t felt pain in his shoulder, although
clearly everybody is worried that there is some problem.

The Boston pitcher was given a cortisone shot in his right shoulder
and the word is that he’s suffering from tendinitis, which I think is
what they usually say when they don’t know what’s going on quite
yet. Or don’t want to tell.

So, why am I, an analysis guy, writing about Schilling’s medical problems?
Well, I thought I might have a look at the pitch data available from
his difficult start to see if we can learn anything. First, we should
easily see his loss of speed, but we can also check pitch movement,
release point and maybe some other stuff as well.

Velocity

Let’s start with pitch speed, which everybody noticed right off the
bat. Did Schill really top 90 only once in the outing? Yes, and
here’s a graph to show it:

This plot is known as a histogram or frequency plot. It shows the
percentage of time Schilling threw a pitch of a given speed. The red
curve shows the June 18 start in Atlanta, while the blue curve shows
his two previous starts this year for which we have the pitch data.
You can see from the red curve that hardly any pitches he threw on
June 18 topped 90 mph. Normally, as the blue curve shows, Schilling
throws quite a few pitches above 90.

By the way, both Jones and Edes were accurate in their
statements. Jones identified Schilling’s fastest pitch of the night
and he also mentioned that Schilling gets to 93-94 when he needs
to. Edes said only one pitch topped 90 mph, that a 91 mph fastball
thrown to Johnson with the bases loaded. That pitch was
measured at 90.3 mph by Enhanced Gameday and it was indeed thrown to
Johnson with the bases loaded. Johnson flied out.

Pitch selection and movement

So, Schilling clearly was not throwing as hard in his last start, but
we can ask more. Did Schilling modify his pitch selection, based on
his lack of stuff? Are all those pitches in the 78-86 mph range just
slow fastballs or did he go to the slider more often? And did his
pitches have less movement in his last start, in addition to a loss of
velocity?

Before we can answer these questions, we need to look at what kind of
pitches Schilling normally throws. The plot on the right shows the
vertical and horizontal movement for the 162 pitches captured by
Enhanced Gameday previous to his last start (see my previous article,
In
Search of the Sinker for background information on plots like
this).

The green points are fastballs and the black ones are curves, while,
the red points are a mix of sliders and splitters. The splitters are
more toward the left of the red cluster, while the sliders are
toward the right, although there isn’t a clear separation.

Now, we can look at the the same plot for the start of June
18. Actually, what is shown below is the same plot we just saw, with
the additional pitches added (solid dots).

There are a couple things to note about this plot. While Schilling
appears to have used his usual assortment of pitches in his last
start, he tended to throw them slower, which you can see by observing
the two “legends” on the plot.

Also, if you look closer at the red
points, there seem to be fewer, at larger negative values
of horizontal movement—Schilling seemed to avoid the splitter
in that start. The fastballs also show less movement in this last
start, although I believe that is consistent with them being slower.

Release point

It is sometime said that a pitcher will lower his arm slot if he
has a shoulder injury. The Enhanced Gameday measurements include the
(x,y) coordinates of the release point for every pitch. This graphic
shows Schilling’s release points for the Atlanta start (red points) and
his previous starts (black). The y-axis records the height above the
ground (in feet) of the release point. The horizontal position is
measured relative to the center of the pitching rubber.

Wow, those red points are consistently shifted with respect to the
black ones. Schilling doesn’t appear to be lowering his arm slot, but
he does seem unable to get proper arm extension. But you know what?
When I saw this plot, I was a little suspicious about the data. THT’s
John Beamer has
already shown that we need to be aware of possible
inaccuracies in the data. I think the release point is one of the
harder things to get right across stadiums, while other measures like
velocity, movement and location are probably more accurately recorded.

In any case, I thought it prudent to check the release point data by
searching for other pitchers who have data recorded in Atlanta plus
other parks. The plots below show the release points for three additional pitchers, red dots for starts in Atlanta,
black dots for starts in other parks.

So, it looks like my suspicions were justified. Release points for all
these pitchers when pitching in Atlanta differed significantly from
their release points in other parks. There seems to be inconsistency
among some of the other parks as well, which makes it hard to
correct for the Atlanta effect. Therefore, we need to
take the release point data for Schilling with a grain of
salt or maybe a few. (You can easily see some yips in the release
point measurements in David Wells’ data: Those points over on the left side of the
plot don’t make much sense, unless Boomer decided to throw
a few pitches righthanded, in a blow-out perhaps.)

The release point data might be hinting at something, though. Perhaps
we can’t trust the shift in release point that we see, but I’m
guessing that the large spread in Schilling’s red points compared to
those of the other pitchers is probably a real effect. It looks to me like Schilling’s release point was wandering around that
evening in Atlanta, much more so than in his previous starts.

Command

Obviously, a loss of command
doesn’t necessarily imply an injury; some days you just don’t have
good command. Still, I thought it’d be interesting to look at
what happened on each of Schilling’s pitches, both in his last start and
in previous starts. Here is a table of the numbers:

Pitch Results
            Previous      June 18
            --------      -------
Ball        49 (.30)*     30 (.38)
Called      40 (.25)      11 (.14)
Swinging    14 (.09)       3 (.04)
Foul        21 (.13)      15 (.19)
In Play     37 (.23)      21 (.26)
Total      161            80
*Numbers in () are the fraction of total.

“Called” and “Swinging” refer to strikes, of course. It’s clear from
these numbers that Schilling also had a tough time commanding his
pitches in Atlanta. He threw more balls (in terms of percentage) and had
many fewer called and swinging strikes—he induced only three
swing-throughs in 80 pitches that night.

He allowed a higher
proportion of balls in play and of those 21 balls in play, 10 went
for hits, including two doubles and a home run—a scary combination
of loss of stuff and loss of command. Of course, we shouldn’t be
surprised that Schilling was rattled a bit when he realized that his
fasty was coming in 5 mph underweight.

Final thoughts

Some people, including Sox
manager Terry Francona, have mentioned age as a possible reason for
Schilling’s struggles in Atlanta. That seems pretty odd to me, since
Schilling was only 12 days older than when he threw a one-hitter in
Oakland, and nobody was worried about Schilling’s age then. I suppose
Francona was just dodging the question about a possible Schilling
injury.

We won’t really know what happened to Schilling until he tries to
start throwing again in a couple of weeks, or maybe not even then.
I, as a Sox fan, am pretty
worried, although I guess being worried is part of being a Sox fan in
the first place.