Pebble Hunting: The Scioscia About Face

You’ve heard the joke about McGregor the Barbuilder and the one goat, I assume. Mike Scioscia’s one goat is Jeff Mathis. It’s been years since Jeff Mathis was under his aegis, and it’s still the first thing I think of when somebody asks me about Scioscia. A few days ago, Jeff Long asked me how well Scioscia manages his bullpen. I thought for a moment. “Well, he started Jeff Mathis over Mike Nap…”

The thing that McGregor found so exasperating is how permanent that one detail turned out to be. It was only one goat, he’s saying. Since then, model citizen. Learned his lesson. Built a bar, built a pier. No more goats. He’s begging to be reevaluated, for us to treat him not as a snapshot of a life but as a life altogether. So, then, let’s give Scioscia the same fair treatment: Now that we can say, with more precision than we ever could before, how good each of his catchers was at defense, let’s reevaluate the Mathis/Napoli years, and see how Scioscia has done with his catchers since.

We’re going to do four things here: We’re going to take the total defensive contributions—the sum of framing (CSAA), blocking (EPAA), and throwing (TRAA and SRAA)—for each of his primary catchers. We’re going to see whether Scioscia truly did favor his better defender over the better overall player; We’re going to add in the offensive contributions (as represented by PECOTA projections) to see how efficiently he got the most out of his two primary catchers; we’re going to see whether the Napoli/Mathis years were so bad, or whether what we thought we saw actually only looked like carnal knowledge of a goat; and we’re going to see whether whatever decision tree we associate with Scioscia and his catchers exists regardless of the personnel he has to work with, or whether it was specific to Mathis and/or Mike Napoli. Onward!

Step 1: Did Scioscia truly favor his better defender?
We have, since the debuts of Napoli and Mathis, 10 years of Scioscia. In a very simple way of answering this question, we can split his catchers into Primary Starter and Primary Backup:

And then we can sum the total per-season values of the starters and the backups and find that, in fact, Scioscia’s backups have been much better defenders than his starters:

• Starters: -22 runs relative to average, which works out to about run lost every 40 games.
• Backups: +61 runs above average, or a full run gained every 11 games.

While we remember Mathis playing instead of Napoli—which definitely happened—Napoli actually led Angels in games started three times, same as Mathis, and the three Mathis years included the season that Napoli was playing almost everyday (in Kendrys Morales’ absence) at first base or DH. And Chris Iannetta has started the most games behind the dish in all four seasons he has been an Angel, even those seasons in which he has been awful defensively and even those seasons in which he had sturdy competition for the job.

Step 2: How efficiently did Scioscia distribute playing time?
This is a continuation of Stage 1, because knowing Mike Napoli played more games than Jeff Mathis doesn’t mean mistakes weren’t made—Napoli’s 84 starts in 2009 being just ever-so-slightly more than Mathis’ 78, after all.

We know that Scioscia needs two catchers, and we can postulate that a typical split between a non-elite starter and a non-horror-show backup might be something like 120/40. Injuries will interrupt this distribution some, but the closer Scioscia gets to having his better catcher on the field 120 times, the better. Here we’ll again go year by year to see how close to 120 starts the superior catcher got. (Just to repeat, what I counted as “offensive value” was actually just the average of that year’s TAv projection and the next year’s TAv projection, because we can’t expect Scioscia to anticipate up and down offensive seasons any better than a good projection system like PECOTA can. This will benefit Jeff Mathis in his early years, when his projections were more encouraging than his results.)

So, Scioscia’s most valuable catcher got:

• 2006: 59 percent of the way to 120 starts (by Jose Molina; no. 2 catcher)
• 2007: 31 percent (Jose Molina; no. 3 catcher)
• 2008: 75 percent (Jeff Mathis; primary catcher)
• 2009: 70 percent (Mike Napoli; primary catcher)
• 2010: 24 percent (Bobby Wilson; no. 3 catcher)
• 2011: 40 percent (Hank Conger; no. 2 catcher)
• 2012: 48 percent (Bobby Wilson; no. 2 catcher)
• 2013: 50 percent (Hank Conger; no. 2 catcher)
• 2014: 58 percent (Hank Conger; no. 2 catcher)
• 2015: 67 percent (Chris Iannetta; primary catcher)

In only three years did his better overall catcher play the most games. In two cases the no. 3 catcher was actually better than both of the starters in front of him—but, it was defense that set those no. 3’s apart, suggesting that Scioscia didn’t value catcher defense enough.

Further, even in the three cases when our numbers agree that Scioscia got it “right,” he didn’t get his best catchers as many starts as they probably deserved. Taken as a group, the “best” catchers—those 10 listed above—started a total of 627 games, or 63 per year.

In 10 years, Scioscia’s nos. 1 and 2 catchers—we’ll exclude the no. 3 catchers, because no team wants to start its no. 3 catcher if it can help it—combined to produce* around seven runs above average. Which is fine, but which (arguably) left a lot of runs on the table. Scioscia’s “best” catchers produced, over those 10 years, about 76 runs above average. If we imagine a scenario where Scioscia a) measured hitters’ offensive talent the same way PECOTA did, b) measured catchers’ defensive talent the same way we now do, c) valued catcher defense as much as our WARP model does, and d) aspired to give his best catcher 120 starts per year, those same 10 catchers would have produced about 147 runs above average. Meanwhile, the below-average alternatives would have played a lot fewer games and exerted less drag on the totals. In a world in which a, b, c, and d were true—and, obviously, assuming our metrics are right—Scioscia more than a win per season on the table by not playing his right catchers.

But not because he was too obsessed with catcher defense. Arguably because he wasn’t obsessed enough.

3. Were the Mathis/Napoli years really so bad?
Not so bad. There are two points favoring Scioscia’s choices: 1) For a couple years at least, Scioscia was justified in believing that Mathis would hit more than he did; 2) Napoli was really an awful catcher.

To point 1, remember that Mathis was a very good hitting prospect, at least for a catcher. At age 20, he hit .315/.380/.493 between High- and Double-A. When he debuted, PECOTA saw a below-average hitter, but only by a little, and that’s impressive enough coming from a 22-year-old catcher. It took a couple years for PECOTA to give up on his offensive potential, and it’s fair to allow that it would take a couple years for Scioscia and the Angels to, as well.

To point 2, Napoli was roughly average behind the dish in his first two seasons, and in those seasons he was, indeed, Scioscia’s primary catcher. But from Y3 until he was traded after Y5, he was much worse: -12.2 runs, -7.3 runs, -8.6 runs—and, remember, that’s in effectively half-time play. When he hit like he did in 2008, he was still an extremely valuable catcher. But that was an outlier offensive season, and as a .280-TAv hitter he was treading water when he wore the gear. Over the five years he shared a job with Mathis, he trailed his counterpart by 50 runs. He made that up with his bat but, again, it wasn’t necessarily foreseeable that he would. If both had hit to their PECOTA projections, Mathis would have been the better player.

By year five—when Napoli was established Bad at defense, and Mathis was established Historically Bad at the plate—Scioscia had his power to choose more or less taken from him by injuries: Napoli had to play first (in place of Morales), the arguably superior-to-Mathis Bobby Wilson was out with a broken leg after a collision at home, and even Mathis missed two months with a wrist injury.

4. Has Scioscia repeated the Mathis/Napoli experience since?
Actually, the opposite. Chris Iannetta was, in his first three years as an Angel, terrible defensively—worse than Napoli was. He was nearly 40 runs worse than average in those three years, but he projected to be the best hitter on the team, and he played. Meanwhile, Hank Conger was the Angels’ best defensive catcher since Jose Molina, and arguably the more valuable catcher (by a pretty good margin), but he was stuck in a backup role. It wasn’t always smooth—Iannetta was reportedly a source of tension between the manager and the front office in 2012—but Scioscia played him and, eventually, Iannetta improved. Last year, he was 10 runs better than average. And now he’s a Mariner.

If there’s a real surprise here—besides the degree to which Scioscia’s Mathis/Napoli decisions look at least plausible, if still ultimately damaging—it’s how many strong defensive catchers Scioscia has refused to give significant playing time to. Jose Molina wasn’t quite yet the 50-Run Receiver when he was with the Angels, but—well, actually he was, saving 31 runs in 108 starts between 2006 and 2007. But Scioscia kept him in a reserve role, handing the starting and primary backup jobs to rookies and (presumably) allowing Molina to be traded in midseason in 2007. The Scioscia of popular caricature would have been the Rays before the Rays, giving Molina far more starts than would have seemed at the time to make sense. Bobby Wilson and Hank Conger were both kept in backup roles despite excellent defense, the sort of defense that got them jobs on the framing-obsessed Rays and Astros. Maybe Scioscia doesn’t value defense the way we do. Maybe he doesn’t evaluate it the way we do. Maybe when he says “catcher defense” he means managing a pitching staff, and all our fancy –AA stats are still missing the one component that would reveal Scioscia’s methods.

My guess, though, is that he was trying to balance offense and defense all along, and in some ways he was more right than we realized, and in some ways he was more wrong than he realized.

*I can’t stress enough, because I know the wording is confusing, that “produce” means defensive value + expected offensive value. I didn't bother with what they actually produced on offense, because that’s a standard of exactness and future-seeing that I don’t expect a manager to meet.