keyboard_arrow_uptop

It’s a feeling we’ve brought up a lot this winter: Pitchers seem like they are capable of changing our minds much more quickly than hitters can. A bump in velocity (like Ubaldo Jimenez) or a move to the bullpen (Will Smith) or a return after years away (Scott Kazmir) can radically reshape how we assess a pitcher's future.

Which suggests we should be open to radically different projections for a pitcher from year to year, and yet! There’s a case to be made that this unpredictability should actually make a projection system even more conservative when evaluating pitchers. As quickly as a pitcher can change our minds about him, he can unchange our minds, lose the velocity or move back to the rotation or undergo a surgery that fixes him. So, when in doubt, regress.

It usually takes quite a bit of time for PECOTA to really change its mind about pitchers. Of the 819 pitchers for whom PECOTA issued projections in both 2013 and 2014, 766 are projected to this year produce a WARP within 1.0 WARP of what they were were projected to produce last year. Of the 53 whose projections changed by more than one win, 45 saw their projections drop. So what must a pitcher do to make PECOTA fall suddenly in love? Answers imminent!

Repeating the exercise we did with hitters earlier this week, here are the pitchers whose 2014 projections are most improved from their 2013 projections:

LASTNAME FIRSTNAME 2014 projection 2013 projection Change
Fernandez Jose 4.6 0.5 4.1
Wacha Michael 3.2 0.5 2.7
Wood Alex 1.7 -0.3 2
Gausman Kevin 1.4 -0.2 1.6
Ryu Hyun-jin 2.3 0.8 1.5
Gray Sonny 1.4 0 1.4
Corbin Patrick 1.6 0.5 1.1
Cole Gerrit 1.9 0.8 1.1
Perkins Glen 0.7 -0.3 1
Salazar Danny 0.7 -0.3 1
Hunter Tommy 0.8 -0.2 1
Wisler Matt 0.9 -0.1 1
Collmenter Josh 1.4 0.4 1
Quintana Jose 1.6 0.7 0.9
Uehara Koji 1.8 0.9 0.9

So think about Billy Ray Valentine in Trading Places. If you had to project how well he would do as a managing director at a billion-dollar commodities brokerage, you would start with the limited information you have: his performance as a street-level hustler. How is he, as a street-level hustler? A grade-65 street hustler? Well, billion-dollar business environments are more difficult than street hustles. More-educated competition, more advanced training required, etc. So you’d discount his grade as you increase the level of difficulty. And, because the skills needed for street hustling might not translate perfectly to the board room, you’d probably hedge a bit and make your projection a little more conservative. Further, you don’t know that much about his street-level-hustle work; you just started watching him this very afternoon! So regress a little more.

And here is what we knew about Jose Fernandez before 2013:

  • 79 innings of low-A, 1.59 ERA
  • 55 innings of High-A, 1.96 ERA

Those are very good innings! But do you appreciate what Low-A is like? The oldest hitter Fernandez faced in Low-A was 25. There are guys on the other team wearing blue jeans because their mom forgot to wash their uniform. This was his team’s theme song:

I mean, to go from an environment as low-budget, amateurish, and sparsely populated as that one, all the way to the Marlins…

[Mugs for camera]

Anyway, that’s more or less what we’re looking at with the top six here: Six Billy Ray Valentines. Each had a short resume entering 2013: All six made their major-league debuts last year, all were drafted (or signed) in 2011 or later, and only Sonny Gray had started a game higher than High-A. It would have been really hard to draw any aggressive conclusions about their true major-league talent level based on their minor-league performances to that point.

Further, other than Michael Wacha—who had thrown just 21 innings, each of them amazing, before 2013—and perhaps Gausman (15 innings before 2013), each pitcher either maintained or improved his raw numbers (or came close) despite the big level jumps. Here’s how Ryu’s final two years in Korea matched his first year in the majors, for instance—

  • 9.8 K/9, 2.4 BB/9, 0.7 HR/9, 2.94 ERA
  • 7.2 K/9, 2.3 BB/9, 0.7 HR/9, 3.00 ERA

And here’s Gray, making the jump from Double-A in 2012 to the majors last year:

  • 5.9 K/9, 3.4 BB/9, 0.5 HR/9, 4.14 ERA
  • 9.4 K/9, 2.8 BB/9, 0.6 HR/9, 2.67 ERA

So basically these top six pulled off all three elements of the Billy Ray Valentine story: They gave PECOTA a lot more data to work with; they gave PECOTA more reliable data to work with, as there’s no mystery about how major-league performance will translate to the major leagues; and, in making the jump, they also got better. Started crushing pork-belly markets from day one. All of which makes it much easier for PECOTA to throw an All-Star-level WARP on their futures.

Corbin’s big projection jump comes from marginal improvements across the board: His strikeout projection goes from 7.0 to 7.3; his walks drop from 2.6 to 2.3; his BABIP forecast goes from .318 to .309; his home runs from 1.1 per nine to 1.0; his groundball rate from 46 percent to 47. There are plenty of pitchers who had ERAs lower than expected, or who made PECOTA reevaluate one component of their games, but Corbin outperformed expectations everywhere. Corbin, by the way, had a 26 percent chance of a breakout last year, according to the 2013 PECOTA run. The median breakout rate is 25 percent, so Corbin wasn’t yet one of PECOTA’s favorites. Here are the breakout chances of all the names in that table up there:

Quintana, like Corbin, just improved everything all at once. His component projections:

2013: 6.1 K/9, 3.3 BB/9, 1.2 HR/9, .298 BABIP
2014: 6.6 K/9, 2.6 BB/9, 1.0 HR/9, .292 BABIP

Summing those small gains adds up to an ERA three-quarters of a run lower. If Corbin and Quintana are the pitchers who won PECOTA over with broad improvements, Salazar is the one who hit PECOTA between the eyes with a cattleprod: His projected strikeout rate, on a per-batter basis, is nearly double what it was a year ago. On a per-inning rate, only Kyle Zimmer got a larger reevaluation:

K increasers:

LASTNAME FIRSTNAME '14 K/9 projection '13 K/9 Projection Change
Zimmer Kyle 8.2 4.1 4.1
Salazar Danny 8.1 4.7 3.4
Alvarez R.J. 10.3 7.2 3.1
Siegrist Kevin 8.1 5.4 2.7
McCullers Lance 7.2 4.7 2.5
Chapman Aroldis 13.6 11.3 2.3
Rosenthal Trevor 9.5 7.3 2.2
Diekman Jake 9.4 7.3 2.1
Gausman Kevin 8.1 6 2.1
Norris Daniel 6.8 4.7 2.1

At levels High-A or lower, Salazar’s career strikeout rate is 6.9 per nine innings. At Double-A or above—all coming post Tommy John surgery—it’s 10.9. Salazar’s projected ERA for 2013 was 6.01.

Perkins, Hunter, and Collmenter are all answers to the rarely-asked question of what’s more valuable, a great reliever or a very, very poor starter. It’s odd to see Perkins’ -0.3 WARP projection for 2013, since he had already established himself as an excellent closer and put the horrible history as a starter behind him. But, as we know from horror movie sequels, just because the monster has been vanquished doesn’t mean the townsfolk ever recover their sense of peace—and it doesn’t mean the monster won’t be back:

  • 2011-2012: 9.8 K/9, 2.5 BB/9, 2.52 ERA
  • 2013 projection: 6.0 K/9, 2.7 BB/9, 4.94 ERA
  • 2013 season: 11.1 K/9, 2.2 BB/9, 2.30 ERA
  • 2014 projection: 7.6 K/9, 2.4 BB, 3.77 ERA

If you ever want to reverse-engineer PECOTA to figure out how much it weights the more distant past, start by examining the difference between Perkins' and Nathan’s projections for this year. In the past three years, they’re very similar, with Perkins carrying narrow advantages:

  • Perkins: 10.4 K/9, 2.4 BB/9, 0.7 HR/9, 166 ERA+
  • Nathan: 10.1 K/9, 2.5 BB/9, 0.8 HR/9, 151 ERA+

Yet Nathan’s projection for 2014 is much more favorable than Perkins’: 10.1 Ks per nine, 2.86 ERA. Don’t spook PECOTA. It remembers.

Here, for fun, are the guys PECOTA turned against:

LASTNAME FIRSTNAME 2014 Projection 2013 Projection Change
Ogando Alexi 2.2 4.5 -2.3
Sabathia CC 3 5.2 -2.2
Beckett Josh 1 2.9 -1.9
Santana Johan 0.4 2.3 -1.9
Marcum Shaun 0.8 2.6 -1.8
Rodriguez Wandy 0.5 2.2 -1.7
Kennedy Ian 1.5 3.2 -1.7
Saunders Joe -1 0.6 -1.6
Medlen Kris 2.4 4 -1.6
Nolasco Ricky 0.7 2.3 -1.6
Johnson Josh 1.8 3.4 -1.6
Harrison Matt 0.9 2.4 -1.5
Haren Dan 1.8 3.3 -1.5
Morrow Brandon 1.3 2.8 -1.5