R.A. Dickey’s new knuckleball

Last week, R.A. Dickey announced that he has adjusted his mechanics in an effort to play through pain that in his neck and upper back. The knuckleballer’s velocity is down roughly two mph as a result of the nagging injury, and yet he has elected not to spend time on the disabled list. On Sunday, Dickey went seven innings and gave up three earned runs on four hits. Dickey threw a quality start, though his knuckleball hasn’t appeared to be very effective. We learned Monday that he will have an MRI this week.

To ensure that his knuckleball crosses the zone at its normal level, Dickey said, he has to “start it higher.” Essentially, Dickey has shifted his vertical release point upward and his horizontal release point a bit further away from his right ear. Depending on what time period you choose as a reference point, the shift in release point seems to measure as a few inches rightward and upward—nothing drastic, but still significant.

Dickey’s average knuckleball velocity was down at its lowest level of the season on Sunday, and the trend doesn’t look encouraging.

image

Reason to worry?

On the surface, Dickey obviously isn’t blowing us away in 2013. Dickey’s FIP is a full run higher this year than it was last year, He is walking hitters more often, and yet his strikeout rate hasn’t dropped significantly. These statistics need some time to stabilize, however, and it is more appropriate for us to examine Dickey on a deeper level.

Let’s look at the sabermetric results that Dickey is getting from his knuckleball thus far in 2013. This season, hitters are swinging at fewer pitches outside the strike zone (25.4 percent) than they were in 2012 (33.3 percent). Additionally, when hitters do swing at pitches out of the zone, they are making contact at a much higher rate this year in comparison to 2012. This combination doesn’t bode well for Dickey, and it doesn’t surprise me that he is striking out hitters at a lower rate than he was last year. If Dickey can’t induce hitters to chase the knuckler more often, his lower strikeout rate will be here to stay.

So long as his knuckleball is moving at the level it has in the past, Dickey’s new release point doesn’t present any issues. However, the swing rate data I presented above indicate that perhaps Dickey’s knuckleball isn’t as effective as he works through his back pain.

We can break Dickey’s results down even further. I’ll compare levels of PITCHf/x movement observed for this pitch over Dickey’s past few starts to the levels we witnessed in 2012. If we see any major differences in movement caused by Dickey’s altered release point and velocity, the pain is a relevant issue. If not, Dickey’s disappointing start to the season may start to turn around.

What makes a great knuckleball great?

As you might expect, analyzing knuckleball PITCHf/x data is tricky. John Walsh wrote a seminal piece on knuckleball movement here at The Hardball Times in 2007, and his work will allow us to examine the characteristics of an effective knuckler.

To briefly summarize, a knuckleball typically moves anywhere between -10 and 10 inches horizontally, and between -10 and 10 inches vertically (before accounting for the effect of gravity). While most pitches consistently have well defined levels of movement, knuckleballs obviously don’t. I won’t go into the physics of knuckleball movement here, because a graduate degree in physics isn’t a prerequisite for fantasy baseball success. Alan Nathan has done extensive research on the subject, and I’d recommend checking out his website if you’re interested.

Here’s what we do know about knuckleballs. The more they move (horizontally and vertically), the more effective they are—but you didn’t need me to tell you that. Walsh created classifications for these different levels of movement, and his analysis of Tim Wakefield’s knuckleball revealed that OPS against “large movement” knuckleballs was almost 300 points lower than OPS against “small movement” knuckleballs. Pitch sequencing obviously plays a role here, but comparatively, “large movement” knuckleballs are better pitches than “small movement” knuckleballs. Walsh didn’t have much data to work with back in 2007, but I was able to verify his findings for Dickey.

If Dickey is throwing more small movement knuckleballs this year, we can’t expect him to achieve the same results he did in his spectacular 2012 season. Let’s look.

What do we see?

Below, I’ve plotted horizontal and vertical movement on Dickey’s pitches. Dickey’s entire 2012 season is represented on the first graph, and his two most recent starts are the next two graphs. The green dots on the first graph were classified as knuckleballs, as are the blue dots on the next two. Don’t trust the pitch classifications, however, as PITCHf/x classifications for knuckleballs can often be incorrect. These misclassifications aren’t really an issue for us, though. Almost nine out of 10 pitches that Dickey has thrown this year have been knuckleballs.

image

Recall the Walsh study that I mentioned earlier. Small movement knuckleballs are hit often, and they are often hit harder than the average knuckleball. I’ve defined medium-large movement areas with black rectangles above. You’ll notice that Dickey has been throwing a lot of small movement knuckleballs in 2013. He also threw fewer medium-large movement knuckleballs last Sunday than he has all season. Many factors affect knuckleball movement (weather, wind, etc.), but I’d guess the drop in movement isn’t a coincidence.

The fact that Dickey emerged from his previous start relatively unscathed is a minor miracle. Those who watched Sunday’s start can confirm that Dickey threw a far too many hanging knuckleballs. This one might be a strong candidate for “worst pitch of the season” thus far. Even if the right-hander’s MRI results don’t reveal any significant damage, Dickey should be fighting through this pain in the training room—not on the mound.

Print Friendly
« Previous: THT Awards
Next: Prospect pitching injuries remain unpredictable »

Comments

  1. Dave Cornutt said...

    I’m not so sure velocity per se is that important.  Dickey and Wakefield are/were knuckleball pitchers in the Phil Niekro mold; that is, nearly all of their pitches are knucklers, with the occasional fastball.  Movement is obviously an important point with the knuckleball.  Wakefield famously found that he got more movement out of his knuckler if he backed off of the velocity some, and before his time Niekro had made the same observation.

    You wouldn’t think a release point should matter with a knuckleball.  However, in order for a knuckleball to work, the ball has to have a little bit of spin; 1-2 complete rotations from release to catcher’s mitt was what Niekro used to shoot for.  If the ball doesn’t rotate at all, it’s more likely to be a “floater”, and Dickey’s chart from the Yankees game has noticably more of those than the chart from the Baltimore game.  From what I’ve read, the knuckleball pitcher usually uses movement of one finger to impart that little bit of spin to the ball.  If Dickey is experimenting with his mechanics, he may have changed how he is doing that.

    I watched the game against the Yankees.  I thought his knuckleball was moving incredibly well in the early innings.  They were doing a lot of super-slo-mo replays of his pitches on the YES broadcast, and the amount of rotation was obvious.  About the 5th inning, I noticed that the ball seemed to be rotating less.  Overbay hit his game-winning HR in the 7th; IIRC Dickey was at about 90 pitches at that point.  That pitch was indeed a meatball.  (The one that Boesch hit out in the 2nd, I thought was a pretty good pitch.)  If Dickey has changed something that he’s doing to get the rotation, possibly he does not have the strength built up in the part of the hand that he’s using now.  70-80 pitches may be his limit for a while.  Obviously that limits his value as a starter.  But since you don’t tell a reigning Cy Young winner to go to the bullpen while he works out his mechanics, here we are.

  2. Alan Nathan said...

    I would be a bit careful using a scatterplot to draw conclusions about how the 2013 and 2012 movements compare.  The problem is that for 2012, the density of points is so high that it is not easily possible to determine what fraction of pitches show up in the “medium-large” zone.  So it is not obvious to me from these plots that there is any difference between the two years in that fraction.  So, I did the analysis myself.  Rather than look at the movement in the x-z plane, I instead looked at the total movement.  Actually, what I really did is looked at the so-called lift coefficient, which is proportional to the total movement.  What I found is a distribution looking not so different between 2012 (2857 identified kballs) and 2013 (659 kballs).  BTW, thanks to @harrypav for access to his database and pitch id’s.  Here is a link to a standard box plot I made:  http://baseball.physics.illinois.edu/Cl-box-2012-2013.PNG.  Perhaps there are better ways to display the results, but I am far from convinced that there is any dropoff in movement this year.

  3. Noah Woodward said...

    @Alan- I agree completely, and heat maps would’ve been more appropriate. I obviously yield to you on this, as I admire your work.

    I disagree with your selection of all of Dickey’s 2013 starts for comparison with his 2012 starts. Dickey was pulled from his April 18th start, and wasn’t complaining of pain before that date.
    I created the appropriate heat maps, and posted them here. The area of concern, for me, is the area that includes -3 inches of vertical movement and below.

    This only depicts three starts worth of data- but wouldn’t you agree that we’ve seen relatively less x-z movement over the past few weeks?

  4. Alan Nathan said...

    Noah…What I see in the data is an upward shift of vertical movement.  For the 2012 data, I find a mean vertical movement of about 0, with rms=7.7”.  For all of 2013, I get mean of 0.8” with same rms.  For last three games in 2013, I get mean of 1.8” and about the same rms.  By shifting the mean movement up by nearly 2”, there is less downward movement.  However, I find very similar total movement for 2012, all of 2013, and last three games from 2013.  But I agree that the “shape” of that movement has changed in the last three games, with apparently less downward movement but more upward movement.  What I cannot say is how significant that shift is with respect to performance.  BTW, I am using my own movement numbers (drag removed, calculated from 50 ft), so the absolute numbers might differ from yours.

  5. Noah Woodward said...

    Thanks, Alan. I haven’t removed drag, so I might be seeing something different. What was your rms for his last three starts?

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *