June 15, 2006
Lies, Damned Lies
Justin Verlander has been the talk of baseball and rightly so--ten quality starts in 13 tries will do that for a rookie pitcher. But just as the mainstream media is jumping on the Justin Verlander bandwagon, many statheads are jumping off of it. The reason is a low strikeout rate (5.25 K/9) that places him with the John Thomsons and Mark Hendricksons of the world, and not the Jake Peavys and Scott Kazmirs. Just how much of a concern should Verlander's strikeout rate be? And what can we expect from him going forward?
If we asked PECOTA this question, it would recognize that Verlander's BABIP has been relatively favorable, and that he's pitched especially well with runners in scoring position, and probably spit out an ERA projection in the very low 4's. But idiosyncratic players can sometimes trick PECOTA, and after a fuller review of the evidence, I've come to conclude that Verlander is pretty damn idiosyncratic.
For one thing, while Verlander might have a Mark Hendrickson strikeout rate, he doesn't have Mark Hendrickson stuff. His fastball doesn't have a ton of movement, but it checks in consistently at 95-97 MPH, and sometimes as high as 100 MPH, and you don't need a ton of movement when you throw that hard. His downward-breaking curveball certainly looks like it should be an out pitch (we'll get to this in a moment). And Verlander struck out more than a batter an inning in the minor leagues. So why isn't he getting the results in the strikeout department?
I decided to watch a condensed game version of Verlander's May 17th victory against the Minnesota Twins. This was an unusual outing in which Verlander walked nobody, struck out nobody, and gave up no runs. Here is a list of all the outcome-determining pitches that Verlander made in that game (players reaching base are indicated in boldface):
Inning Outs Batter Count Pitch Result ----------------------------------------------------------------- 1st 0 Stewart 1-1 Fastball (96) f-3 1st 1 Castillo 2-1 Fastball (95) 1B 1st 1 Mauer 1-0 Fastball (95) 8 1st 2 Hunter 1-1 Fastball (97) FC-4 2nd 0 Cuddyer 2-0 Fastball (94) F-3 2nd 1 Morneau 2-2 Fastball (97) F-6 2nd 2 White 1-1 Fastball (97) 6-3 3rd 0 Batista 3-2 Fastball (96) f-3 3rd 0 Punto 1-0 Fastball (95) 1-3 3rd 0 Stewart 2-1 Fastball (96) 2B 3rd 0 Castillo 1-1 Fastball (98) 1-3 4th 0 Mauer 2-2 Fastball (96) 6-3 4th 1 Hunter 2-1 Fastball (97) L-3 4th 2 Cuddyer 0-1 Curve (78) HBP 4th 2 Morneau 0-1 Curve (83) 1B; 9-5 * 5th 0 White 3-2 Fastball (97) F-5 5th 1 Batista 2-0 Fastball (95) f-2 5th 2 Punto 0-0 Curve (84) 2-3 6th 0 Stewart 0-1 Curve (81) f-3 6th 1 Castillo 2-2 Fastball (98) 6-3 6th 2 Mauer 1-2 Fastball (100) 1B 6th 2 Hunter 0-0 Fastball (98) 1B 6th 2 Cuddyer 1-0 Fastball (97) 8 7th 0 Morneau 2-0 Fastball (93) 7 7th 1 White 1-0 Fastball (94) 6-3 7th 2 Batista 1-2 Fastball (99) 8 8th 0 Punto 1-1 Fastball (96) 1B; CS 1-3 8th 1 Stewart 1-0 Curve (81) 6-4 8th 2 Castillo 3-2 Fastball (96) E-5 8th 2 Mauer 0-0 Curve (81) 4-3 * Cuddyer out trying to advance to 3rdHelpful? Not especially. But we see that Verlander, in spite of having inherently good command, actually gets behind in the count quite a lot; like many pitchers, he relies almost exclusively on his fastball when he does. We also see that perhaps he is able to use this fastball to induce a lot of weak contact. There are four foul ball outs in this game (designated with the lower case 'f'), three infield popups, and a lot of meekly-hit balls to shortstop. If you watched this game, you wouldn't be a big believer in DIPS. Verlander was throwing a lot of good pitches, and he was getting a lot of good results--those results just didn't happen to include any strikeouts.
Let's look at a couple of more pieces of information from ESPN's Inside Edge data (warning: subscriber only). Firstly, as we inferred from his start against the Twins, Verlander almost never throws his curveball when behind (only 3% of the time; he also rarely throws his changeup). That is not so abnormal.
Secondly, Verlander gives up a .310 batting average when he does throw his curveball. That is abnormal, especially for a pitcher that throws his breaking ball only on favorable counts. By comparison, opponents are hitting .167 against Jeremy Bonderman's slider, .207 against Barry Zito's hook, and .119 against Bobby Jenks' big curve. The simplest explanation for why this is the case is probably the right one--pitchers like Zito often place their breaking ball where it can't possibly be hit, whereas Verlander tends to throw his in the zone. (The 'other' simple explanation--that Verlander's curve isn't as good as Zito's or Jenks's--is probably also true. It's a major league pitch, but it isn't a 'wow' pitch, at least not in its present incarnation.)
The third interesting piece of information is Verlander's pitch distribution. These are Verlander's results against right-handed hitters when he throws within the strike zone.
Most pitchers work the inside part of the plate a bit more often than they work the outside part. But Verlander takes this to an extreme. Nearly half his strikes are in the 'jamming zone'--either straight inside or low and in. By comparison, here are Jeremy Bonderman's numbers:
And Roy Halladay's:
Verlander is very deliberate about where he throws his pitches, particularly his fastball (his curve can occasionally flatten out or get away from him). If he throws a pitch outside the strike zone on a 3-1 count, it's probably because he wanted to. Conversely, if he throws a hittable pitch when ahead in the count--which he does far more often than most pitchers with his velocity--it's also because he wanted to. The relative dearth of strikeouts are not because of a lack of stuff, but because of the decisions Verlander makes throughout the at bat. This is different from a pitcher like Jon Garland who has trouble missing bats.
Whether Verlander's strategy is optimal or not is hard to say. One thing we don't know enough about is how batters' results differ based on where a pitch is thrown in the strike zone. If Verlander's BABIP is .240 because he's throwing down and in so much, then his approach becomes more interesting. However, I think he would almost certainly benefit from throwing his curveball in the dirt more often with two strikes, or perhaps by developing a splitter.
What's especially interesting is that Verlander does manage to rack up strikeouts when he most needs them. Thus far in the 2006 season:
Oh, and that PECOTA will be low.