Knuckleballer Steven Wright was pitching pretty well in 2016 before he was sidelined with a shoulder injury, but so far in 2017 he has been simply bad (ERA of 8.03, 2.094 WHIP). After his most recent game, Brian MacPherson noted on Twitter:
Steven Wright is throwing what in effect is a different knuckleball than he threw last year, at least in terms of the way it's moving: pic.twitter.com/TAMKVd19kS
— Brian MacPherson (@brianmacp) April 23, 2017
Jason Mastrodonato noted the same thing:
I'm not *totally* sure how to interpret these horizontal movement charts for Steven Wright, but I know that 2017 is different…(via Brooks) pic.twitter.com/YRqcq7Cl50
— Jason Mastrodonato (@JMastrodonato) April 23, 2017
There’s no doubt that MacPherson and Mastrodonato are correct. Let’s compare PITCHf/x for Wright from 2016 and 2017. First, we can outline the approximate borders of various pitch types that Wright throws: four-seam fastball (“FF”), two-seam fastball (“FT”), knuckleball (“KN”; Wright throws two different types, fast and slow), and curve (“CU”). On a plot of speed vs. horizontal movement, Wright’s slower knuckleball can’t be distinguished from his curve, but they become more or less separate clusters when comparing speed vs. vertical movement:
Now, using the same outlines as for the 2016 pitches, let’s look at 2017:
There’s a very obvious difference in Wright’s faster knuckleball. Instead of filling the 2016 oval, all the pitches are clustered over to the left of it in the horizontal movement chart.
This is bad. It’s bad because the knuckler is effective due to its randomness, and in 2017 Wright’s knuckler is not random, it’s moving in a predictable way. Putting some numbers on it, in 2016 Wright’s knuckleball had an average horizontal movement of 0.2 inches with a standard deviation of 4.8 inches; in 2017, it’s been -2.2 +/- 3.3 inches:
Is it possible that it’s misleading to compare pitches from 24 games in 2016, to four games in 2017? Perhaps Wright had different horizontal movement from one game to the next, and his knuckleballs only had the broad spread of movement over the season as a whole. Here is 2016, game by game, only looking at the knuckleball clusters:
And 2017, game by game:
This doesn’t help Wright’s story. Throughout 2016, in each game his knucklers had wide and random horizontal movement, filling most of the oval each time. In 2017, each game shows the relatively tight cluster of pitches, all over in the left side. (There might be a little exception on 4-17-17, when we do see a few pitches over in the right part of the oval. Perhaps coincidentally, that was Wright’s only effective game of this season; he went six innings and gave up one earned run.)
In fact, looking at the 2017 game-by-game movements, another problem jumps out. Look at the vertical movement cluster. Again, the knuckleball should be moving randomly in the vertical plane, as well as the horizontal, and that’s what happened in 2016. But in 2017, that has only happened in two games. On April 12, his knuckleball tended to move down relative to the path it would follow according to gravity alone (that is, it had negative vertical movement), and on April 22, it moved “up”, rising like a fastball.
The consistent trends in horizontal and vertical movement show that Wright is putting spin on his pitches. (PITCHf/x offers “spin” information, but that isn’t useful for a knuckleball, since PITCHf/x “spin” is not directly measured, but is inferred from pitch movement.) A knuckleball with spin is not a knuckleball, it’s just a predictable 75-mph pitch. That’s a fastball from a modestly talented 14-year-old pitcher, and major-league batters facing a high school freshman pitcher would do just about what they’ve done to Steven Wright this season.