By the middle of the 2016 MLB playoffs, it was all but a foregone conclusion that Andrew Miller's usage pattern and dominance during the Cleveland Indians' postseason run would change the way that teams build and deploy their pitching staffs. With Terry Francona's rotation already decimated by late-season injuries, the Tribe skipper was quick to pull his starters at the first sign of trouble, and Miller was often the first man up in those situations. Half of Miller's playoff appearances came in the sixth inning or earlier, and all of them lasted more than an inning. This seemed like a dramatic shift in usage pattern that would likely affect roster construction and managerial decision-making for other teams as they might try to copy this strategy (albeit with much less effective arms than Miller).
However, as far as using relievers earlier in games, the leaguewide revolution had really already long since started. As writers deemed each successive year in the early 2010s the Year of the Pitcher, MLB-wide ERA was going down and -- as one might expect if pitchers are performing well -- the average length of starts was going up.
This trend came to an abrupt halt in 2015, when the average length of a start dropped from 6.0 innings to 5.8 and the number of relief appearances in the league as a whole jumped by almost 650 games. This could have been a copycat trend in response to the Kansas City Royals' World Series run in 2014, in which they rode a dominant and deep bullpen and came within one base of winning a title. Their championship in 2015 with the same core of players gave even more credibility to their roster construction, and the Indians' 2016 performance has solidified the new reality that deep bullpens can carry a team.
Regardless of what might have sparked this change in bullpen usage in the past few years, it is clear that teams are using their bullpens differently. But how is it different, and perhaps more importantly, when are they using their pens differently?
You can see that the overall trend is toward shorter starts, but the changes in direction from year to year are very much driven by the amount of runs being allowed. More run scoring, shorter outings -- it all adds up. So far this season, that trend appears to be bucked a bit, as scoring is down but the length of outings still continues to drop.
How does this manifest itself in day-to-day tactics? Starting pitchers start because they can maintain a high performance level for several innings at a time, but their performance declines once they start moving through the batting order a third or fourth time. Whether because of fatigue or the opponent having a longer look at him, at around that time, there's a point of inflection where it is often better to bring in a reliever instead of leaving the starter out there.
Unsurprisingly, we see a fairly consistent decline as a pitcher faces a lineup more times. There is a little bit of an improvement in triple-slash numbers by the fourth time around only because pitchers who are performing well get a chance to make it through the lineup three full times. However, their ability to miss bats is severely compromised at that point, so they are more at risk to allow big innings in big spots.
In the past couple of years, teams have taken to avoiding allowing their starter to make it to the fourth time through the lineup. Since 2014, the number of starters who have faced at least one lineup spot four times has been cut in half, decreasing by 4 or 5 percent every year.
Teams are bringing relievers in during the middle innings more frequently, and the pitchers they bring in are better-suited to handle a tough spot than a weary starter. Relievers who have come in for the fifth through seventh innings this season have allowed a .713 OPS and are striking out 22.8 percent of batters so far this year, which bests a starter in his third time through the order by more than 50 points of OPS and adds almost 50 percent more strikeouts.
Despite the above trends, teams are not trying to create their own Andrew Miller. Miller has been referred to as a "fireman" because he comes into tight spots when runners are on base or the heart of the order is up. This season, relievers entering in the middle innings have come in with an average of 0.65 runners on base, which is the second-lowest mark in the past 15 years, and have come in to face one of the batters hitting No. 2 through No. 5 in the order just 37 percent of the time, which is the lowest in that time frame. Perhaps Miller's part of the reliever usage revolution was more specific to the postseason, so we might see it show up more as the playoff hunt takes shape.
The best middle-relief teams
So, which teams are doing the best in this new world of going to the bullpen earlier than ever? I wanted to see which bullpens are doing the best in the fifth through seventh innings. To compare players and teams, I decided to include LOB percentage (strand rate) and FIP (fielding independent pitching) to measure effectiveness. Strand rate gives the percent of runners who are on base for a pitcher's appearance that fail to score, including those the pitcher inherited when he came in. FIP is a simple and valuable sabermetric tool, as it measures a pitcher's performance on only the results he has the most control over, namely walks, strikeouts, hit batters and home runs. It's scaled to match ERA, so a good FIP is around 3.50 or better, but as a metric, it's more useful than ERA in projecting future performance.
If you've followed the season closely, there are some obvious entries here, in particular the Astros and Indians, who are playing a big series this weekend in what might be a playoff preview. The Astros' Chris Devenski broke out in a big way in the early going this year. Up until a rough outing last Sunday, he was carrying a sub-2.00 ERA with almost 50 percent strikeouts while making eight appearances of more than one inning. The Indians' contingent obviously includes Miller, but it is Bryan Shaw, Zach McAllister and Nick Goody who are helping carry the load in the middle innings. We might not expect them to pitch this well all year because none of them has allowed a home run so far this season, and their 95 percent strand rate will be almost impossible to sustain over the next four-plus months.
As for the other teams on the list, the Dodgers are interesting because their rotation depth has helped bolster their long-relief options. Ross Stripling, he of the near-no-hitter in his MLB debut, has performed well in swingman duties, and Alex Wood did an excellent job there until he was summoned to the rotation to continue that performance. The White Sox have had breakouts from Tommy Kahnle and Anthony Swarzak to support a somewhat beleaguered bullpen. The Red Sox have done some mixing and matching in front of closer Craig Kimbrel to figure out the middle innings, and given the discrepancy between their FIP and their other stats, we might see their run prevention improve a little in the coming weeks..
Teams looking to improve in the middle innings
Taking a different tack, I wanted to see which teams have been getting really good performances from their back-end relievers that might be better served by using them a little earlier in the game. To analyze this, I decided to list the teams that had the most back-end relievers (those who have appeared in the eighth or ninth inning of a close game this year) with at least five innings in such situations and a FIP below 3.50, and see which of those teams are having trouble in the middle innings. That could include both poor middle-relief performance and also poor starter performance toward the ends of their starts.
Off the bat we notice that a few teams make both lists (Indians, Astros, Red Sox), with each team having formidable closers. All of those clubs are doing a fine job already in middle relief. Maybe they could afford to remove their starters a little earlier to take advantage of that strength -- and maybe they will when the season reaches its stretch run.
The Rockies added a couple of arms to their bullpen in the offseason in Mike Dunn and Greg Holland, and that has really improved the unit's depth. They don't have an outstanding performer, but that group has allowed them to pull their starters fairly early, and they could afford to be even more aggressive with that strategy. Both the Angels and the Blue Jays were forced by injury to stretch a reliever in their rotations (J.C. Ramirez in L.A., Joe Biagini in Toronto), so not only is their bullpen depth compromised, but their average starts tend to be short. In the Yankees' case, the injury to Aroldis Chapman limits their flexibility right now, but a middle relief option like Adam Warren could be stretched out to help them bridge the gap to get games to Dellin Betances.
Alex Vigderman is a research analyst for Baseball Info Solutions. He still gets goosebumps when he hears Harry Kalas' home run calls.