"OF COURSE, ONE GREAT IDEA GUARANTEES NOTHING": Few people in recent years have contributed as much to our understanding of baseball as Vörös McCracken, who figured out around the turn of the century that while pitchers could control strikeouts, walks and home runs, "There is little if any difference among major-league pitchers in their ability to prevent hits on balls hit in the field of play."
It's counterintuitive, but a decade's worth of research have proven it's largely true. There is no consistent year-to-year skill in "getting them to hit balls at the fielders" -- it is random. (Knuckleballers and extreme fly-ball pitchers are the outliers.)
All of which is to tee up this remarkable piece of journalism from Jeff Passan, on McCracken's journey from obscurity to the Red Sox, to where he is now.
Miinor qualification: a pitcher's fly ball/line drive/ground ball tendencies are repeatable and correlate to hit prevention. Beyond that, McCracken's insight is not even really counterintuitive. If you think of the field of play as an arc ranging from 0 degrees (the LF foul line) to 90 degrees (the RF foul line), the differences between a hittable spot of the arc and a sure-out spot of the arc are pretty narrow -- maybe +/- 5 degrees on either side of the infield, a smaller spot up the middle, and a couple of spots in the outfield that might be larger or smaller depending upon the quality of the outfielders and how well hit the ball is. The pitcher can influence (not control) whether the ball is hit to the left or right by lateral placement, speed, and curvature, and can influence (not control) whether the ball is hit to the outfield or infield by vertical placement and sink/rise, but all of the fine tuning on the particular angle is in the hands of the batsman. The notion that a pitcher can throw a sinking pitch and know that it's going to be hit to the 25% mark instead of the 30% mark, irrespective of the contributions of the last variable (the hitter) is pretty far-fetched. It was only ever counterintuitive, and it only now retains popular currency, because earlier generations of players and writers created a myth about it and then invested themselves in the truth of that myth. If you started baseball from scratch today and polled smart people about this question, McCracken's revolutionary insight would be obvious. It's sort of like the Coase Theorem in that respect.
ReplyDeleteOne thing I've wondered when watching pitchers get into "bad times" and walk two or three batters in a row - why don't they intentionally throw a pitch the batter will hit and hope the outfield makes the out? Wouldn't that be better than a walk?
ReplyDeleteBecause, I imagine, the same lack of control that leads to more walks leads to "hittable pitches" becoming home runs.
ReplyDeleteThanks for this. I wasn't terribly invested in the myth mostly because the myth seemed preposterous to me. I'm glad to know I'm not the only one for whom this was an obvious insight.
ReplyDeleteThere are a lot of "pitch-to-contact" pitchers in the major leagues who do just that. Few walks, few strikeouts, high contact rates. But batting average on balls in play (BABIP) is roughly .250 for ground balls, .150 for fly balls, and .700 for line drives (plus or minus a bit).
ReplyDeleteSo let's say you're a league average pitcher, and of balls put in play against you (both hits and outs), you give up about 40% ground balls, 40% fly balls, and 20% line drives. And let's also say that you get about 15 strikeouts and 15 walks per hundred batters faced.
Those numbers are arbitrary, but they're not good. You're giving up too many walks. So your pitching coach says "cut down on your walks." The only way you can do this is by giving up on strikeouts -- no more nibbling around the corners, no more reaching back for your really good stuff. And let's say you go all the way down to zero strikeouts and zero walks.
All else being equal, you would have gone from 36 baserunners and 64 outs to 30 baserunners and 70 outs (excluding defensive contributions, which you shouldn't exclude, but whatever). All else isn't equal, though. First, some of the additional 9 expected hits are going to go for extra bases. Second, by moving all of your pitches (not just pitches to 30% of the batters) toward the center of the plate, you're increasing the likelihood of line drives and home runs, which means that you're really looking at a difference of fewer than 9 runners (and possibly a negative difference) and a potential increase in total bases. The projection is really complicated. There are some pitch-to-contact pitchers who have had great success and long careers (Derek Lowe, for example). Some have been up and down (Ryan Franklin). Some are cataclysmic disasters (Horacio Ramirez).
Like the Coase theorem, though, it took somebody getting up and saying "wait a second, this is obvious," and people going "THE EARTH IS FLAT, GALILEO, NOW SHUT UP." It's obvious in retrospect and might be obvious in a vacuum, but it's not obvious if you've always been taught otherwise.
ReplyDeleteThis also suggests that fielding by pitchers is also constant, which I wouldn't have expected. Either that, or the number of balls hit to pitchers is so small as to be statistically insignificant.
ReplyDeleteThat number may not be statistically insignificant, but you also have to sift out the balls that even poor-fielding pitchers turn into outs. So the sample you're looking for consists of (a) balls hit just inside and outside of standard pitcher range, plus (b) pitcher errors (I think that's it -- I can't recall ever seeing an instance where a pitcher, rather than a ball, was late to first base when a ball was hit to a first baseman).
ReplyDeleteGiven that pitchers' required range is so small (both because of positioning and because of the time for recovery from follow-through), the margin in (a) seems like it would have to be tiny. Pitcher errors (especially throwing errors) might be larger than that, but they don't happen so much that I would imagine they are statistically significant.
Should have looked at this before posting. Bill James reports that average putouts by pitchers in the modern era are .58 per game, which we'll call .6. There are about 6.6 strikeouts per team per game, which leaves 21.4 outs per batted ball (foul-outs included). These numbers aren't perfect, but assuming strong correlation between balls fielded per position and putouts, it looks like pitchers get slightly less than 3% of the balls hit in play. So if the difference between a great-fielding pitcher and a poor-fielding pitcher were 10% of the balls hit to the pitcher (which seems high to me, but I wouldn't know), it would make an expected difference of three points in the BABIP. To the extent that that seems marginally significant, you also have to consider that random fluctuations would mask it.
ReplyDelete<span>Should have looked at this before posting. Bill James reports that average putouts by pitchers in the modern era are .58 per game, which we'll call .6. There are about 6.6 strikeouts per team per game, which leaves 21.4 outs per team per game on batted balls (foul-outs included). These numbers aren't perfect, but assuming strong correlation between balls fielded per position and putouts, it looks like pitchers get slightly less than 3% of the balls hit in play. So if the difference between a great-fielding pitcher and a poor-fielding pitcher were 10% of the balls hit to the pitcher (which seems high to me, but I wouldn't know), it would make an expected difference of three points in the BABIP. To the extent that that seems marginally significant, you also have to consider that random fluctuations would mask it. </span>
ReplyDeleteI saw AJ Burnett forget to cover first in a play against the Phillies this year. Can't find video.
ReplyDeleteI wonder if a pitcher's (putouts + assists)/game is consistent over time?
I don't know about assists, but the point of the Bill James article I saw that said that current pitcher putouts are .58 a game was that pitcher putouts/game have not changed significantly since the late 1800s.
ReplyDeleteAnd -- crap, I didn't count assists at all. Which are like four times putouts. So the difference in BABIP might be as much as 15 BABIP points, which does start to look modestly significant, though I still think that individual year-to-year variability would mask the effect.
ReplyDelete