The Mets’ new and unorthodox lineup caused quite a stir in the first week of the season. In addition to batting David Wright—arguably the team’s best hitter—second, manager Terry Collins dropped contact hitter Daniel Murphy to fifth, potential basestealer and leadoff candidate Juan Lagares to sixth, moved slugger Lucas Duda up to third, and even hit pitcher Jacob deGrom eighth.
These moves stand in stark contrast to Collins’s reputation as a baseball traditionalist. They even led some in the media to question whether the Mets’ skipper controls his own lineup card. Both Collins and GM Sandy Alderson maintain that he does.
However these lineup changes came to be, this much is clear: First, that they reflect some of the basic sabermetric tenets of lineup optimization. And second, that Collins seems to be on board with these changes, and deserves credit for that.
In The Book, authors Tom Tango, Michael Lichtman, and Andrew Dolphin explore how lineups can be arranged to maximize run scoring. To do this, they run simulations based on historical game records and determine both how often and how many runners tend to be on base—and with how many outs—when each spot in the order comes up. Here’s what they find:
The three best hitters should bat first, second, and fourth. The leadoff hitter should skew toward better on-base skills and less power. This is because he gets the most plate appearances of any hitter in the lineup, but the fewest runners on base to drive in. The fourth hitter should skew toward power, as he bats with the most runners on base. The second hitter should have a combination of both, and ideally be the best hitter in the lineup. The two-hitter is uniquely situated to both drive in the leadoff hitter and set himself up for the bigger bats hitting behind him; he also comes to the plate more often than everyone except for the leadoff hitter.
The next-best hitters should bat third and fifth. While they can generally be of equal skill, the better hitter should actually be in the five-hole because he comes to the plate with more runners on base. The third hitter, meanwhile, hits with two outs more often than does anyone else in the lineup; he also has far more plate appearances with the bases empty than does the five-hitter. As a result, the third hitter should ideally get most of his value from the home run ball, which is how he can do the most damage with two outs and/or the bases empty. The fifth hitter, on the other hand, should be the better hitter overall, and one who can both drive in runs and keep a rally going. The six- through nine-hitters should be put in order of decreasing offensive ability.
Finally, batting the pitcher eighth actually produces more runs than does batting him ninth. This is because the benefit of having a better nine-hitter to set up the top of the order slightly outweighs the cost of giving the pitcher more plate appearances. In other words, the theory of the "second leadoff hitter" actually has some merit.
It’s hard to say what the Mets’ "optimal" lineup would be, given the relative uncertainty surrounding players like Lagares, d’Arnaud, Flores, even Granderson, Wright, and—everyone, really—at this point in their careers. For the sake of argument, let's use the Mets’ ZiPS projections for 2015. If you plug those projections into the Baseball Musings Lineup Analysis—a fun tool that, granted, has its flaws—you get a lineup that basically fits the criteria laid out above:
|David Wright||Lucas Duda||Curtis Granderson||Michael Cuddyer||Travis d'Arnaud||Juan Lagares||Daniel Murphy||Wilmer Flores||Pitcher|
The model puts Duda, who projects to be the Mets’ best hitter, second. Cuddyer, a top-three hitter on the team who skews toward power, bats fourth. Wright, another top-three hitter and one who skews toward on-base skills, leads off. d’Arnaud is projected to be the team’s next-best hitter and bats fifth, while Granderson—the best hitter after that and one who gets most of his offensive value from the home run ball—bats third. Lagares bats sixth, followed by Murphy, Flores, and the pitcher. (The projections for the pitcher’s spot are simply based on what the Mets got out of their nine-hole last year.)
Again, the model isn’t perfect. Given that Lagares is projected to be by far the Mets’ weakest-hitting position player, it’s odd that the model wouldn’t put him further down in the lineup; based on the principles laid out in The Book, the pitcher should probably bat eighth with Lagares ninth, despite his speed. Nonetheless, the lineup analysis tool gives us a rough idea of what an optimal Mets lineup would look like.
Compare that to the lineup Terry Collins has been running out for the majority of the young season:
|Curtis Granderson||David Wright||Lucas Duda||Michael Cuddyer||Daniel Murphy||Juan Lagares||Travis d'Arnaud||Wilmer Flores||Pitcher|
A few major similarities stand out. The first four hitters are the same, albeit in a different order. Both lineups have Cuddyer batting cleanup and Wright in one of the first two spots in the order. Like the optimized lineup, Collins divides his third and fifth spots by putting the better singles and doubles hitter (Murphy) fifth, and the bigger home run threat (Duda) third. The bottom five spots are identical, with the exception of Murphy and d’Arnaud being flipped.
The major difference is that Collins bats Granderson leadoff, which is a spot that should be reserved for a top-three hitter on the team. Although Granderson actually was the Mets’ third-best hitter by wRC+ last year, ZiPS projects him to be just the fifth-best hitter in 2015, behind Duda, Wright, Cuddyer, and d’Arnaud. Granted, it’s hard to see any manager in today’s game batting Wright, Duda, or Cuddyer—established power threats with mediocre-to-poor speed—leadoff. Assuming that the choice is realistically between Granderson, Lagares, and possibly Murphy, it’s reasonable to project Granderson as the best hitter and therefore best leadoff option of the three.
The other major difference between the two lineups is that Collins bats Duda third, while the lineup analysis tool bats him second. The difference between hitting second and third might not seem like a lot, but, by the principles of lineup optimization, it should actually be the difference between your best hitter and your fifth-best hitter. Duda is projected to be by far the Mets’ best hitter, meaning that the three-hole might not be the best spot for him.
Finally, compare those two lineups to how the Mets’ lineup would look by traditional, old-school conventions:
|Juan Lagares||Daniel Murphy||David Wright||Lucas Duda||Michael Cuddyer||Curtis Granderson||Travis d'Arnaud||Wilmer Flores||Pitcher|
The old-school lineup has some issues. First and foremost, this lineup would give two of the three premier offensive spots—the one- and two-holes—to two of the three weakest offensive position players, by ZiPS. Meanwhile, it would bat David Wright, one of the team’s elite offensive players, in a spot that should be reserved for the team’s fourth- or fifth-best hitter. Finally, it would push the team’s third-, fourth-, and fifth-best hitters to the fifth, sixth, and seventh spots in the order. That’s a poor strategy for getting the most value out of your lineup.
Overall, Terry Collins has been right on the more controversial lineup issues. The second spot in the order should be reserved for a great hitter like Wright, rather than a player like Murphy, who, despite being a good contact hitter, is probably not one of the best offensive players on this team. Collins’s instinct to hit the pitcher eighth is also correct; in fact, he should probably do it more often.
While batting Duda third instead of fourth or fifth is probably not optimal, it could be defended as a way to get him more plate appearances, and as a way to get a power hitter in the third spot. And, finally, among the major contenders to lead off, Granderson probably makes the most sense. He may no longer be as good of a basestealer as Lagares or Murphy are, but Granderson is projected to be the better hitter. Plus, while it may be the triumph of hope over experience, the Mets are paying Granderson to be a top offensive player, and if he lives up to that, batting him leadoff could actually work out for the best.
Some say that a team's batting order doesn’t matter very much in the grand scheme of things. They could be right. After all, by most estimates, the difference between a standard lineup and an optimized one would only be about five to fifteen runs over the course of a season.
But remember that, as a rule of thumb, ten runs translate to roughly one win. Therefore, an extra five to fifteen runs could mean as many as one or two extra wins in a given year. For a team, like the Mets, that could be fighting for a playoff spot—and even for those that won't be—why not order your lineup in a way that will likely result in more runs and more wins?
Smart teams gain a competitive advantage by staying ahead of the curve and doing effective, innovative things that other teams don’t. Lineup optimization seems to be an area in which most teams have not fully caught up to where the evidence suggests they should be. If Terry Collins is willing risk blowback from the press and the fan base by running out an unorthodox lineup that will likely make his team better, more power to him.