Finding the Red Sox’s Optimal Lineup

It’s been a while since the Red Sox had a good spring training battle. The most recent I can think of was when Jackie Bradley made the club out of camp in 2013, but that was more about David Ortiz’s injury than anything else. Beyond that, teams like the Red Sox don’t often let spring training stats dictate all that much. Which is good and how it should be. The downside though is that there isn’t much drama during March. Again, that’s probably a good thing for the team, if a bit less so for ink-stained wretches like yours truly. 

This generalized pre-Spring Training roster stasis got me thinking though, the Sox lineup is pretty set. The team will be, on most days:

DH: David Ortiz
1B: Hanley Ramirez
2B: Dustin Pedroia
SS: Xander Bogaerts
3B: Pablo Sandoval
LF: Rusney Castillo
CF: Jackie Bradley
RF: Mookie Betts
C: Blake Swihart

That’s a pretty solid lineup, and in fact, it’s projected to be a productive group. Our system here at BP, PECOTA, projects the Sox to score 735 runs, which, by the projections, would be the third most scored in the AL. Only the Blue Jays and Astros would score more. 

When you look at the AL East though, the projected numbers are pretty interesting. The Orioles are the low team at 708 projected runs scored, then moving upwards, the Rays at 713, then the Yankees at 726, the Sox, and then the Jays at the top. That’s a pretty close nit group there with just 27 runs separating the Orioles from the Red Sox. What this tells us is that PECOTA thinks the teams are pretty equal offensively. Put another way, a few runs will make a big difference. 

Part of that difference is going to be sequencing. There’s not a whole lot the Red Sox can do about that in advance other than change their uniforms to blue and call themselves the Royals. Beyond that, the chips will fall and that will be that. What the Red Sox do have control over however is their batting order. 

Much has been written about batting orders over the years. Decades ago we thought they were very important. Then, when the Sabermetric revolution hit we decided, no, they are insignificant. Now though, we’re starting to see the nuance in between those two positions. Now we know that a few runs really can be important, and in an age when league-average players can cost tens of millions of dollars, optimizing the batting lineup can be an effective (and cheaper) way for teams to generate more runs.

The trouble is figuring out how to do that. There are lots of theories about the best way to set a lineup. Most prominently The Book: Playing the Percentages in Baseball by Tom Tango, Mitchel Litchman, and Andrew Dolphin seems to be the standard sabermetric go-to. They say, after intensive study, the team’s best hitter should bat first, followed by second, fourth, third, fifth, and then six, seventh, eighth, and ninth. In essence, teams should bat their hitters in order of quality with the exception of switching the third and fourth hitters.

In an age when league-average players can cost tens of millions of dollars, optimizing the batting lineup can be an effective (and cheaper) way for teams to generate more runs.

That’s fine, and I’m sure it’s mathematically correct, but you may notice one problem with it. That problem is determining beforehand who your best hitters are. Look at the Red Sox lineup. Who are the best four hitters in that lineup? Probably David Ortiz, Mookie Betts, Dustin Pedroia, and Xander Bogaerts? Can you imagine a batting order of Ortiz, Betts, Bogaerts, Pedroia? Also, can you believe this is the fourth question in a row? Me neither! But back to the (third) question at hand: there’s no way the Red Sox will bat Ortiz leadoff and Pedroia cleanup. It won’t happen. Maybe it should. Maybe in an optimized lineup it would, but in this, the world of 2016 baseball, it won’t. So that’s an issue.

But beyond real-world type problems, let’s look at the rest of the lineup. Because optimizing your lineup requires you to order your best hitters one-through-nine, we need to do that. But how do we do that? Who will be a better hitter in 2016, Rusney Castillo or Jackie Bradley? Would it surprise any of you if Pablo Sandoval was a decent hitting third baseman this year? How surprised would you be if he was terrible again? This complicates things. It complicates things because you can’t optimize your lineup if you can’t order the players by quality and you can’t order the players by quality if you don’t know if Sandoval is better than last year or the same or worse. You can’t optimize your lineup if you don’t know whether Jackie Bradley is better than Rusney Castillo. 

This leaves the Red Sox up a creek. But in a way it’s less of a problem for them, and for John Farrell. That’s because Farrell has last year to fall back on. He has all of the player’s careers to rely on for information, he has their personalities, and their preferences, and their ages. He can use these “soft” factors and come up with a batting order that maximizes not runs but player comfort. 

And he will do exactly that. But we don’t want player comfort, we want the most runs we can possibly get. So, instead, we’re going to call up the PECOTA player projections and run them through the batting order optimization tool available at Baseball Musings. This will tell us which lineups the tool thinks will score the most runs. I stress that this is just for fun, as ordering players by their projected on-base percentages and projected slugging percentages is hardly the stuff of rocket science. Still, it should give us some idea of what the optimized setup would be. 

So I plugged in the numbers and the best batting order is as follows:

1. Dustin Pedroia
2. David Ortiz
3. Pablo Sandoval
4. Mookie Betts
5. Hanley Ramirez
6. Blake Swihart
7. Xander Bogaerts
8. Rusty Castillo
9. Jackie Bradley

You can probably see some similarities with the lineup Farrell is expected to run out there on Opening Day. But you can probably pick up some differences as well. Sandoval won’t see an inning as the number three hitter unless the Red Sox start playing split squads during the regular season. That said, I really like hitting Swihart sixth and I love batting Ortiz second, even though it won’t happen. 

I mentioned a few paragraphs back that the Red Sox are forced to deal with real world preferences and realities, stuff that a simple ordering of the players by quality wouldn’t reflect. I also mentioned finding the middle ground between hardcore numbers and the men who play the game. I think there is a way to organize things so that perhaps the Red Sox can seek to optimize their lineup to increase run production in a competitive AL East division, but do so while keeping the realities and idiosyncrasies of the team and players in mind. 

Lineup optimizers are fun, but the trick is for teams to optimize on the ground, not just in a spreadsheet.

For instance, moving Ortiz from second to fourth is only moving him from second to third in terms of player quality, but puts him in a more comfortable spot. The same goes for moving Pedroia from first to second, as he doesn’t like to lead off. Move Betts from fourth up two spots to leadoff isn’t much of a compromise either and in fact, it wouldn’t surprise me if the projections were off and at the end of the year Betts was the team’s best hitter. And now we have Betts and Pedroia one-two and Ortiz in his familiar cleanup spot. 

Lineup optimizers are fun, but the trick is for teams to optimize on the ground, not just in a spreadsheet. The bigger trick this season though, whether for team or computer, is to figure out the actual order of player quality. If the Red Sox knew right now that Jackie Bradley was going to hit better than Rusney Castillo and Pablo Sandoval they might not move him out of the nine hole right away, but they’d certainly be more quick to pull the trigger a month into the season. 

And maybe that’s the lesson. The answer to a team’s lineup isn’t to plug numbers into an online tool, but then it isn’t to become static and reliant on player desires and past performance either. There’s a middle road to take and the sooner the team finds it, the more runs they’ll score.

Photo by USA Today Sports Images