Mastering the Knuckleball Part 1: A Look at the long-established knuckleballer, Tim Wakefield

            Tim Wakefield has been around for what seems like forever.  And since PitchFX data has become available, he is the only pitcher to have thrown enough knuckleballs for us to get any measure of the pitch.  John Walsh  and Dave Allen have examined the pitch to see how such a seemingly batting-practice like pitch can be at least a somewhat effective pitch. 

            This year has seen the emergence of R.A. Dickey, another knuckleballer, for the New York Mets.  Dickey had pitched the last 2 years for Seattle (2008) and Minnesota (2009) as a starter and a reliever, but had never been particularly successful.  This year, things would seem to have changed for Dickey, and he's been the most effective starter for the Mets this year.  In doing so, he's been more impressive this year than almost every year Tim Wakefield has pitched.  This raises the question:  how has Dickey's knuckleball changed this year (as compared to the last two years) and how does it compare to the pitch thrown by Wakefield. 

            This article is the first part of a three-part series that will address these questions by looking at both Wakefield and Dickey, seeing their developments over the last three years, and how their knuckleballs have been (or have not been) effective.  This first part will examine Tim Wakefield.

Figure 1:  The Spin Deflection of Tim Wakefield's pitches over the last three years.  The Red dots represent Wakefield's fastball, the purple dots represent Wakefield's Curveball, and the light blue pitches represent the knuckleball.  The Graphs are from a catcher's point of view, so pitches on the left side of the graphs appear to bend toward Right-Handed Hitters, while pitches on the right side of the graphs bend toward Left-Handed hitters.

Figure 2:  Heat Maps showing the most frequent locations for Wakefield's knuckleball each of the last three years.  The darker the red, the more frequently does Wakefield's knuckler hit that location as the pitch crosses the plate.  The Black Rectangle in the center is an approximation of the strike zone. 

Table 1:  A breakdown of the movement of Wakefield's knuckleball each of the last three years.  Note that while there are some fluctuations each year, They actually aren't that huge.

Tim Wakefield has been throwing the knuckleball for over 15 years in Boston and has gotten his knuckleball routine down pat.  The graphs in Figure 1 show this fairly well*...while the knuckleball moves somewhat randomly, his overall spin-deflection plot for each year looks more or less the same.  As Table 1 shows, over the last three years, 50% of his pitches tend to move horizontally in the range between roughly 2 inches toward Right-Handed Batters to 5 inches or so in on Left-Handed Batters.  Similarly, 50% of his pitches tend to have a vertical spin deflection ranging from roughly -1 inches to +5 inches. 

*NOTE:  The Knuckleball is reported to "knuckle" in mid-air and change direction mid-flight.  This effect is NOT captured by PitchFX data...rather the above data only records the spin** deflection from when the pitch is first recorded (50 feet from the plate) to when it reaches the plate. Thus, it's important to note that some of the effectiveness of the pitch may be missed by this data.
**Can one really use the term "spin deflection" when a knuckleball has barely any spin? 

Now knuckleball pitchers don't generally change their aim with the knuckleball from pitch to pitch, because they can't predict which way the pitch is going to move.  So they tend to instead pick a part of the plate to aim at where the movement of the pitch most often won't end up with it moving too far out of the strike zone.  Figure 2, which shows us where Wakefield's pitches have been located as they have crossed the plate, shows us the end result of this aiming technique, but it's rather curious.  The end result of Wakefield's knuckleball's movement and his aim at a particular part of the plate is for his pitches to NOT be centered at the center of the strike zone, but at a point a little higher and in on right-handed batters.  The end result is that the pitch tends to be up and in to Right Handed Batters and up and away to Left-Handed Batters.  I would wonder if Wakefield could adjust his aim slightly down and toward left-hand batters while keeping his knuckleball form constant.  If so, he might be able to hit a more even area of the plate, reduce his walk rate and fly ball rate, and thus improve his results.  Still, Wakefield has managed to locate his pitches within my approximate strike zone a healthy amount of the time each of the last three years.

Now, since his pitches moving the same roughly each of the last three years and have been hitting the same locations in the strike zone, we might expect for the pitch to have the same results as well.  And Indeed, for the most part we do see this. Take a look at table 2 below:

Table 2:  The Effectiveness of Wakefield's Knuckleball each of the last 3 years.  Explanations of each of the columns are as follows:
Run Value Per 100 Pitches:  The Run Value per 100 Knuckleballs thrown.  NEGATIVE Run Values are good while Positive Run Values are bad (The opposite of fangraphs).
Expected Run Value per 100 Pitches: The Run Value per 100 Knuckleballs thrown that we would expect based upon the batted ball types of balls put in play against Wakefield. 
Whiff Rate:  (# of Swinging Strikes)/(# of Pitches Swung at by Batters)
Swing Rate:  (# of Pitches Swung at by Batters)/(Total Pitches Thrown)
Swinging Strike %:  (# of Swinging Strikes)/(Total Pitches Thrown)
GB Rate:  % of balls hit into play by batters that result in ground balls.
In Play Rate:  % of total pitches (of that type and year) that are put into play.
% of Pitches in the Zone:  % of Pitches in a wide (2 feet wide) strike zone.

As Table 2 shows, none of Wakefield's numbers on his knuckleball change very much from year to year.  The Swinging Strike Stats (Whiff%, Swing%, and Swinging Strike%) all stay around the same area (around 8% Swinging Strikes) and Wakefield's knuckleball has always been an extreme fly ball pitch.  It's true, that in 2008, the knuckleball has a much better run value than either of the last two years, but that number is a mirage caused by Wakefield's .247 BABIP that year (Knuckleball BABIP does tend to be low, but .247 is .032 below his own career average).  As a result, his expected Run Value in 2008 is more in line with his other year results. 

One last thing on Wakefield, as the previously mentioned articles by John Walsh and Dave Allen show, Wakefield's pitch DOES get better results with more movement.  However, there's a catch to this:  If Wakefield gets too much movement on his knuckleball, the pitch misses the strike zone too often to be a valuable pitch.  Thus while it's a generally a positive that Wakefield is able to get a lot of movement on his knuckleball, it is not always the case. 

In any event, Wakefield's numbers make him ideal for a number-5 starter who can eat up innings and save the pen.  His numbers this year (despite his ugly ERA) are basically the same as the last two years, meaning there's no reason he shouldn't continue to have a job in baseball this year or next year.  But his knuckleball is not high-value material and is unlikely to become so. 

The same cannot be said of this year's breakout knuckleballer, R.A. Dickey.  In the next part of this three-part series, I'll explain how he has developed, how his pitches have worked this year, and how his pitches are different from that of Tim Wakefield.