April 13, 2005
Lies, Damned Lies
Change-ups and BABIP
At its core, the scientific method is about attempting to explain observed phenomenon. A lot of sabermetric innovations have been the result of a clever researcher playing around with a large dataset, molding the data in a way that nobody has before, and emerging with a result that is both exciting and readily verifiable.
Voros McCracken started by observing that certain well-known pitchers were not very consistent in their ability to prevent hits on balls batted into play; Roger Clemens or Randy Johnson might rank at the top of the charts in BABIP one year, and on the lowest rung the next. As he soon discovered, this was not some isolated characteristic of Clemens or Johnson; throughout the annals of baseball history, the vast majority of major-league pitchers had demonstrated almost no ability to regulate their BABIP.
In that spirit, let me point out a couple of observations of my own:
You probably see where I'm going with this. We usually think of the defining characteristic of the knuckleball as being that it flutters in unpredictable directions. More basically, the knuckleball is simply a lot slower than other pitches, far slower than a conventional change. Takatsu's slow pitch is not a knuckler, but it moves at about the same speed as a knuckler, and it appears to behaving similarly in terms of preventing batters from getting their fair share of hits on balls in play.
The next logical step is to examine whether other pitchers that throw good slow balls--albeit not quite as slow as Takatsu's--have had similar success in hit prevention. Data on just which pitches a major leaguers throws is either maddeningly incomplete or prohibitively expensive, but fortunately, we have the Neyer/James Guide to Pitchers to work with. Page 35 that book contains a list of 18 pitchers with noteworthy change-ups, either included on Rob and Bill's Top Ten list, or in the honorable mention section that follows it.
Here are those 18 pitchers, rated by career Delta-H. For those of you who are unfamiliar with it, I'll save you a click and give you the definition from our glossary:
"Delta-H: The number of hits above or below average for this pitcher, based on his own number of balls in play and his team's rate of hits (minus home runs) per ball in play; (H-HR) - BIP * (team (H-HR)/BIP). Essentially, the Voros McCracken number. For a team, Delta-H should be zero. Positive numbers signify more hits allowed than expected ('bad luck,' if you believe pitchers have nothing to do with the outcome of a BIP), negative numbers mean fewer hits than expected ('good luck')."
Andy Messersmith -181 Mario Soto -145 Pedro Martinez -97 Jamie Moyer -96 Jean Dubuc -91 John Tudor -91 Ellis Kinder -76 Trevor Hoffman -38 Kirk Rueter -12 Nap Rucker** -10 Stu Miller -1 Mark Eichhorn +3 Mike Boddicker +5 John Franco +14 Ed Lopat +21 Bill Sherdel +26 Johnny Podres +35 Doug Jones +63 AVERAGE -37 ** Neyer and James say that Rucker developed his change-up in late career. I'm using the last five full seasons for his Delta-H number.Of the 18 pitchers, seven have posted career Delta-H numbers significantly lower than what you'd expect from luck alone. Doug Jones is the only real outlier on the other side, and everything he threw was slow. On the average, the group of pitchers with killer change-ups posted a Delta-H of -37. That's not an enormous difference--it amounts to just a handful of extra hits prevented per season--but the evidence points in the right direction.
Seeking out another "objective" list of change-up pitchers, I Googled across a USA Today series of season preview pieces from last year, which included a list of top change-ups by division based on interviews with players, managers, coaches and scouts. Those lists included a total of 15 names, three of which--Trevor Hoffman, Pedro Martinez and Jamie Moyer--were already included in the Neyer/James catalog. These are the other 12, along with their career Delta-H:
Greg Maddux -134 Tom Glavine -131 Keith Foulke -75 Brad Radke -63 Mark Buerhle -52 Wade Miller -33 Johan Santana -22 Eric Gagne -14 Kirk Rueter -12 Mark Mulder -11 Mark Redman +10 Matt Morris +18All told, that's 21 of 30 change-up artists with a Delta-H number below the norm. If you think I've missed someone obvious, feel free to look them up yourself on our DT cards.
While some pitchers, especially the relievers on this list, throw a change-up with a lot of movement or otherwise use it as an out pitch, for most others the pitch is part of a grander strategy to upset the hitter's timing and keep him off balance. A hitter might make contact with the off-speed pitch, but he's not going to make good contact, producing a lot of weak grounders and pop-ups. Contrast that with, say, a Rich Harden fastball--that pitch is damn tough to hit, but when you do make contact with it, you're going to hit it a long way. Consider a Kerry Wood curveball: most of the time, you'll swing in the wrong place entirely, but when Wood does leave one hanging, it is liable to be hit very hard. Change-ups are not difficult to hit--they're relatively easy to hit--but the best ones are hard to hit well. The same thing goes with knuckleballs; knuckleball pitchers don't accumulate a ton of strikeouts, but they compensate by not allowing a ton of hits.
It's not my goal to pick holes in McCracken's work; that we need to go out of our way to find exceptions proves his theory's strength. But I do think that I've found one here.