Pitching Backward: The Not-So-Sudden Breakout of October’s Best Lefty Reliever

Where the Tigers and Dodgers and Nationals and A's have failed this offseason, the Orioles haven't, and won't, because their manager can call for Andrew Miller. The 2014 postseason has been a tragic and entertaining reminder that bullpens are in fact important, and that elite bullpens—or at least bullpens that perform elitely—can cover for other deficiencies on a roster. Miller has arguably been the best reliever so far this postseason, which shouldn’t come as a surprise unless you opted to not watch any baseball over the previous seven months.

A few years ago, when I first started looking more deeply into the statistical side of baseball, I thought that I would try to solve one of the biggest mysteries in baseball. I thought that by using linear regressions and key statistics I could identify which relief pitchers were going to break out the following season. My model was never able to accurately predict much of anything, so I shelved it.

Recently I’ve been thinking about this quite a bit, inspired by the spectacular season that soon-to-be free agent Andrew Miller has put together. Miller has been a major leaguer since 2006, bouncing around between the starting rotation and the bullpen for the first few years. Recently he’s locked down a role in the bullpen, though it wasn’t until 2014 that he really came into his own as a shutdown reliever.

Over his career Andrew Miller has posted 3.3 WARP, with more than half of that value coming in 2014. His 1.7 WARP this season (1.2 with Boston and 0.5 with Baltimore) was third best among qualifying relievers behind Dellin Betances and Wade Davis. This is a significant improvement over his first three full seasons in Boston, where he transitioned into relief but posted just 1.1 WARP over three seasons.

As teams look at Andrew Miller’s impending free agency they’ll be asking themselves a very simple question: Why was 2014 a breakout season for Andrew Miller? I’ll save some intern in an MLB team’s front office the time and tell you.

This breakout was a few years in the making. It all goes back to his transition from starter to reliever, way back at the end of the 2011 season. Miller really forced the hand of the Red Sox as he scuffled mightily as a starter during the second half of the season. He made five relief appearances from August through the end of the season, including his final three outings that season. The immediate effect was an uptick in velocity for Miller, as you can see below:

That increase coincides with his final three outings, all as a reliever. For the bulk of the season Miller sat around 93 mph, but his switch to the bullpen would see his fastball velocity jump up to the 96-97 mph range. This was the first ingredient in Miller’s breakout season, and it came way back in 2011.

During the 2011-2012 offseason, the Red Sox set about implementing some mechanical changes that would simplify Miller’s motion to the plate. In fact, Miller took to his mechanical changes so well that the Bobbys (manager Valentine and pitching coach McClure) were raving about him going into the 2012 season. Tim Britton of the Providence Journal noted as much:

Now, Miller's potential has earned a bit of a boy-who-cries-wolf reputation. Each spring, there's talk of mechanical adjustments, of throwing strikes, of harnessing that pure stuff into tangible results. There's no definitive evidence that what McClure is trying to do with Miller will succeed where pitching coaches Curt Young and Randy St. Claire and Mark Wiley and Chuck Hernandez have all failed. (That McClure is the fourth pitching coach in the last four seasons for Miller certainly hasn't helped him find a consistent mechanical approach.)

But the early results have been promising, and McClure's excitement at watching how the tweaks translate to Miller's live batting practice session Friday is palpable. Miller, too, knows this may be the last adjustment he gets the chance to make.

2012 was a bounceback year, as he raised his strikeout rate by more than 4.5 batters per nine innings, lowered his walk rate by more than a batter per nine innings, and produced a FIP (and ERA) two runs lower than it had been the previous season. Further mechanical tweaks and continued practice with his new mechanics would lead Miller into the 2013 season with an air of respectability about him.

For the 2013 season Miller refined his plan by eliminating his two least effective pitches from his repertoire (a move that was under way back in 2012). Throughout 2013 Miller threw just two pitches: a fastball and a slider. Eliminating his other offerings allowed Miller to find some balance with his two pitches, a roughly 3:2 ratio of fastballs to sliders that allowed him to keep opposing hitters on their toes.

Below I’ve included some screenshots that show the changes in Miller’s mechanics. Marc Normandin also did an analysis of Miller back in 2012 when he first made the mechanical changes, and it’s worth a read if you want a look at the changes as they were first implemented.

The first screenshots below show Miller’s position at leg lift. In 2011 (on the left) he’s bringing his right leg up much higher than in the 2014 screenshot (on the right). He also has his hands set lower in his mechanics from 2011:

Miller’s mechanics are clearly simpler and more compact, with the goal of making them more repeatable. Miller is very tall, very long, but In simplifying his move to the plate he would theoretically have more control over where his pitches end up. Since 2011 the portion of the zone where the majority of Miller’s pitches have ended up has gotten smaller, highlighting the consistency with which he can hit his spots.

If you paid close enough attention to Miller’s new mechanics, you might have seen 2014 coming. Take this quote from the 2014 Baseball Prospectus Annual:

“Bobby Valentine left few positives behind from his year at the helm of the Sox, but the changes to Miller's mechanics were unarguably productive. Inconsistency in his delivery was the issue, so it was simplified: He pitched exclusively from the stretch, used a shortened stride to ease his leg through his motion and replaced his loopy curve with a tighter, devastating slider variant. The results were surprising and immediate, and he improved so much in 2013 that his season-ending ankle injury was considered what had been unthinkable roughly a year before: a bullpen-debilitating blow. In his 71 relief innings with these mechanics, Miller has struck out 99 batters, death-murder-killed nearly every lefty he's come up against, and even limited the production of right-handers to .209/.345/.330. With his career thus rescued, he should rejoin a stacked Boston bullpen in 2014.”

For a better idea of how Miller’s mechanics from 2011 compare to those of 2014 you can watch the clips I’ve taken screenshots of here: 2011 and 2014.

Below I’ve included a GIF showcasing Miller’s zone profile from 2011 through 2014. You can clearly see how Miller’s approach when it comes to pitch location has shifted as he’s honed his craft in the bullpen.

In 2011, Miller pitched on the plane from the lower left corner of the zone through the top right corner of the zone. The following season he worked to eliminate some of those pitches at the top and above the strike zone. The following year he continued to evolve, working down in the zone effectively, with pitches being located on either side of the plate. This includes a significant portion of pitches missing the zone on either side of the plate. This season Miller pitched almost exclusively to the lower left corner of the zone, effectively mixing fastballs and sliders in that location.

That location, the lower left corner of the zone, has proven to be very effective. The difference between Miller’s solid 2013 and his elite 2014 is his ability to command his pitches in the zone, and specifically hit the spot that he’s taken advantage of this season. Hitters walk up to the plate knowing where Miller is going to throw the ball, and with about 50 percent confidence what he’s going to throw, yet hard contact is rare against the O’s reliever. Time for the smut!

It’s unclear what impact the tweak in mechanics, improved command, extra velocity in a relief role, or change in pitch usage has had on Miller’s performance. It’s extraordinarily difficult to attribute Miller’s 2014 performance on any one or more of these factors, as they’ve built upon one another—and perhaps with other variables that have simply made his stuff nastier. To say that 2014 is a result of simple tweaks that he made during the offseason would be a misguided oversimplification. This season has been years in the making, quite literally.

In 2011, Miller had peripherals that were about as bad as they could possibly be without guaranteeing a swift demotion to the minor leagues. Going from 1.2 K/BB in 2011 to 6.1 K/BB this past season is pretty remarkable. Miller was able to more than double his strikeout rate while cutting his walk rate by more than half. The trip that Andrew Miller has taken from borderline replacement level to bullpen ace has been a long one. The end result though, is remarkable, as professional hitters flail at Miller’s offerings.

We might not be able to use Andrew Miller as a model for predicting when other relievers are going to have breakout seasons, but we can see how Miller’s own emergence is the culmination of several changes that happened right in front of us. What we can do is marvel in the spectacular and somewhat surprising season he’s had in 2014. When Andrew Miller inevitably comes in for the O’s during the late innings of the ALCS over the next week, take a moment to admire the journey he’s been on, before marveling at what he can do on the mound.