Optimizing the Nats Batting Order

Batting order is a tricky subject, how much time should really be given to it?  While we know that having some order to a lineup is beneficial, at the end of the day it can only do so much. With that said since the Nationals and Adam LaRoche finally agreed to a two-year deal we finally know who the starting eight will be for the Nats in 2013. And those eight provide some interesting batting order decisions, such as where to put Jayson Werth and Bryce Harper, and how to balance the lineup for optimal performance.

Right now there is a consensus three main rules for the Nats batting order:

  1. Ryan Zimmerman bats third.
  2. Denard Span bats leadoff.
  3. Try to split same handed hitters up.

If we follow those three rules we come to a lineup that looks something like this, along with what hand they bat with and their 2012 triple slash:

  1. Denard Span, L, .283/.342/.395
  2. Jayson Werth, R, .300/.387/.440
  3. Ryan Zimmerman,R, .282/.346/.478
  4. Bryce Harper,L, .270/.340/.477
  5. Adam LaRoche,L, .271/.343/.510
  6. Ian Desmond, R, .292/.335/.511
  7. Danny Espinosa, S, .247/.315/.402
  8. Kurt Suzuki, R, .267/.321/.404

Now if you wanted to truly alternate left-right-left-right then you could switch Zimmerman and Harper, still giving you an excellent lineup, just without the familiarity of Zim in the three hole. But how much of a benefit would truly
come from going to the alternating order instead of the familiar order?

To answer that we can look at a fun little online tool over at Baseball Musings that does lineup analysis using on-base percentage and slugging percentage.  The tool runs those averages against historical data to determine how well a certain lineup will perform. While it isn’t a perfect system and it isn’t easy to determine how relevant the
historical data is to baseball now, it still can give us a rough idea of how a certain batting order would perform.

When the first lineup is entered into the tool, it needs nine players so I added Jordan Zimmermann and his .246 OBP and .281 slugging % to represent the pitchers, it gave a runs per game average of 4.935. Compared to the 4.51 they scored last season that is a marked improvement, however it’s difficult to say just how accurate that number is. The alternating lineup gave a runs per game average of 4.936, an extremely slight increase over the first lineup, and with the added benefit of having no same-handed hitter consecutively.

Now the Baseball Musings tool doesn’t just analyze a lineup it also provides what it believes is the most optimal lineup. For the Nats the top lineup produced 4.986 runs per game on average and looked like this:

  1. Jayson Werth
  2. Adam LaRoche
  3. Bryce Harper
  4. Ryan Zimmerman
  5. Ian Desmond
  6. Kurt Suzuki
  7. Denard Span
  8. Danny Espinosa
  9. Jordan Zimmermann

That’s weird, and definitely is an example that using only numbers might not always lead to the best solution.  While I doubt the Nationals are ready to bat Adam LaRoche second, there are still some interesting ideas there.  For one, the Nationals weren’t so far off in batting Werth leadoff last season, and keeping him near the top is probably a good idea. Also here again we see Harper third and Zimmerman fourth, so perhaps giving in to convention and batting Zim third isn’t the best decision for the Nats.

But that isn’t the only lineup optimization idea. In the excellent and directly named “The Book” there is some discussion of lineup optimization.  Sky Kalkman at Beyond the Box Score sums it up nicely:

Another way to look at things is to order the batting slots by the leveraged value of the out. In plain English (sort of), we want to know how costly making an out is by each lineup position, based on the base-out situations they most often find themselves in, and then weighted by how often each lineup spot comes to the plate. Here’s how the lineup spots rank in the importance of avoiding outs: #1, #4, #2, #5, #3, #6, #7, #8, #9

Now if we follow this idea and apply it to the Nats lineup we get something that looks like this:

  1. Jayson Werth
  2. Adam LaRoche
  3. Bryce Harper
  4. Ryan Zimmerman
  5. Denard Span
  6. Ian Desmond
  7. Kurt Suzuki
  8. Danny Espinosa
  9. Jordan Zimmermann

Again this looks weird with Adam LaRoche batting second, and even more so with Denard Span batting fifth.
However, when run through the analyzer it produced 4.956 runs per game, which is better than the first two lineups we came up with. Perhaps the baseball world isn’t quite ready for the radical ideas of lineup optimization, or perhaps
more factors need to be considered that will give results we’re more used to. Regardless we can still use this information to make an informed decision on how to order the Nats lineup.

Taking all this into consideration I think the second lineup (Span, Werth, Harper, Zim, ALR)  would probably be the best lineup for the Nats.  It follows the third and fourth lineups fairly closely and takes into account speed and what side the batter hits from, which those lineups do not. Alternating left-right-left throughout the lineup will neutralize how opposing managers use their bullpen, and the entire lineup has balance between speed and power. While it breaks rule number one I think it will be beneficial to both Harper and Zimmerman to bat 3-4. However, there is no question that whatever way Davey Johnson decides to order his lineup, it will be potent from top to bottom.

6 comments

  1. It’s important to note that despite his growing reputation, based on ’12 stats and projections for ’13 Harper is the Nats FIFTH best hitter and would therefore slot in the #3 spot in the order. It’s also important to include this caveat, how guys have hit in the past in NOT independent of where they hit in the order previously, batters will see a different mix of pitches depending on where they hit, in the most extreme example Wilson Ramos has a great OBP (.350+), but would almost certainly NOT draw nearly as many walks with a power hitter batting behind him rather than the pitcher (almost all his ABs were batting 8th).

    Like

  2. Further caveat, and a good reason line-up optimization is still as much art as science, is deciding what stats to use? Just OBP, or OPS? Should OPS be weighted to favor OBP over SLG? Does SLG count more than OBP in some slots?Even when you decided upon one or forumula to use, then do you use their previous year stats or career stats or even projected stats (and then who’s?)?

    Like

    1. Both are excellent points and as you said it’s definitely a large gray area about how best to optimize a lineup. I think it would be really interesting to see a study of what pitches a hitter is likely to see in each spot in the lineup. And that would definitely affect where you would put certain players.

      Like

  3. The difference between the optimized line-up and the standard line-up logic tells us the Nats should put out is 808 to 799 or 19 runs or roughly two wins. In as close of a pennant race as the Nats are expected to be in those two wins could be crucial but there are two other factors to consider here. The first and most important is injury and days off. The Nats are not going to have either of those line-ups out there for 162 games. Even if they are extremely lucky and no one misses anytime there are going to be off days and while Tracy, Moore, Lombo, and Shark are all good bench players none of them are as good as the players they will be replacing in the line-up. The second and possibly more important thing is the human element. It is hard to calculate what this even means or matters, but baseball players are human and while moving Zimmerman to fourth isn’t a demotion in any way shape or form moving LaRoche to second would be seen that way as well as moving Span anywhere but lead-off. Both of them seem like team players and overall good guys but even if they don’t let it show on the surface that the demotion bothers them it could effect their performance in a sub-conscience manner. With no knowledge of how if in any way it would effect these players it is probably not worth the risk when it comes to the matter of what will amount to a difference of less than 19 extra runs when off-days are accounted for.

    Like

    1. As with my argument about the pitches they see, Ramos in #8 slot will see vastly different pitches than he would in the #5 slot as an extreme example, but the other factor is Batting Approach – simply looking at the statistics and saying that so-and-so should lead off because their OBP is high, does NOT mean they would have the same success batting leadoff as the did say batting 5th, if a player alters their approach to fit the role they perceive (rather than what we’d come optimum) it negates the move. I think "protection" is given far to much credence by old school types, but is also too easily discounted by statistical analysts, who is on base and who is in the batters box DOES affect a pitchers approach to the batter, it’s not a huge difference, but over 150+ games it will be significant.

      Like

    2. The human element is definitely important and disappointingly difficult to quantify. If only it were as easy as punching in different numbers in a vacuum, but then the game probably wouldn’t be nearly as fun.

      Like

Comments are closed.