Sunday, March 22, 2009

Optimizing the Nationals Batting Order

With a hat tip to Tom Tango, Mitchel Lichtman, Andy Dolphin, and Sky Kalkman, this post looks at the Nationals batting order and projects an ideal lineup to maximize offensive production.

First, a little context. Tango/Lichtman/Dolphin in The Book wrote about batting orders and blew up some long-held assumptions about the best way to build a line-up. Some of the key points:
  1. Batting order actually doesn't matter that much. Perfectly optimizing a line-up will typically deliver at best one or two additional wins over the course of a season. That's not nothing, but it's a lot less important than you'd think given the amount of oxygen, ink, and pixels spilled on the topic.
  2. One of the most important factors to consider is simply the increased number of at bats hitters get near the top of the lineup. This factor makes the second spot as important as the clean up slot.
  3. In the lead-off spot, OBP is overwhelmingly the most important factor, and speed is really not that important. You don't want clods, but you don't need big base-stealers. In fact, assuming you're bunching up your best hitters near the top of the line-up, you're not going to want to risk outs by running the lead-off man anyway. The best base-stealers are better used down in the line-up in front of slappy singles hitters.
  4. The third batter is much less important than typically assumed, because this hitter comes to the plate with two outs and no one on an unusually high percentage of the time (in the first inning).
Sky gives a very good quick summary of in this post:
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.

So, you want your best three hitters to hit in the #1, #4, and #2 spots. Distribute them so OBP is higher in the order and SLG is lower. Then place your fourth and fifth best hitters in the #5 and #3 holes, with the #5 spot usually seeing the better hitter, unless he's a high-homerun guy. Then place your four remaining hitters in decreasing order of overall hitting ability, with base-stealers ahead of singles hitters.
Sky has been applying these principles to various teams over at Beyond the Boxscore (sorry if I'm scooping you on the Nationals, Sky--I'm guessing you aren't going to bother with the worst teams in the league on this series...), and I thought it would be fun to do the same with our hometown nine. Here's my ideal Nationals line-up, assuming the starters currently listed atop the Nationals depth chart.

Here it is, the official FJB ideal batting order for the 2009 Washington Nationals:

1. Nick Johnson
It's really strange to me that fans and the team apparently have spent so much time fretting over the lead-off spot over the last four years when we have damn near the ideal lead-off man in Nick Johnson. He's an OBP god, and, although he's not a big speed guy, as noted above that isn't all that important here anyway. And still, it's not like he's a big base-clogger. In fact, last season, despite the injuries, he was fifth on the team in EqBRR (Equivalent Base Running Runs, a BP stat that measures the number of runs added on the basepaths from stolen bases, taking the extra base on singles, etc.) with 0.46 runs added. (Cristian Guzman, the presumed lead-off man, by the way, was next-to-last on the team with -2.99 EqBRR.)

2. Elijah Dukes
I'm really tempted to put Dunn here, because just getting him the 30-40 additional PAs he'd get in the #2 hole is tempting. The conventional wisdom that you don't want big strikeout guys in the #2 hole because you want to be able to move the runner over if an out is made. But this is actually exactly wrong: you'd rather have the #2 hitter, who comes to base with no outs and a runner on first a lot, make strikeouts than contact outs to avoid DPs.

All that said, I want to avoid bunching up our lefties to provide a measure of protection
from LOOGYs for Johnson and Dunn. Dukes is certainly the next best option here, and we don't lose too much in those extra 30-40 ABs with him versus Dunn.

3. Lastings Milledge
Milledge is the team's fifth best hitter, so he goes here in the third spot, the fifth most important spot in the line-up. Again, all those two-out, bases-empty first inning PAs really erode the importance of this spot vis-a-vis the #1, 2, and 4 spots.

4. Adam Dunn
If Dunn gets on in the first inning, by definition it will mean there's at least one runner on. Beyond that, he benefits from hitting in close proximity to two of our best three hitters (Johnson and Dukes).

5. Ryan Zimmerman
Clearly the best option here, Zimmerman gives you the combination of solid power and on-base skills to drive in Dunn when he doesn't clear the bases himself.

6. Cristian Guzman
With his poor on base skills, Guzman belongs nowhere near the lead-off spot. This is by far the biggest mistake Manny makes, and given his supposed propensity for statistical analysis, he should know better. Not only would Johnson be a better option, but Kearns, Willingham, and Dukes would be as well.

As for Guzman, for a contender, he'd really belong in the 7th or 8th spot. In this line-up,
Guzman is the next best hitter and can keep the L-R-L-R thing going. From here on out, we're just listing players in declining order of overall offensive value.

7. Jesus Flores
You remember his game-winners, but it's his .296 OBP that you should focus on. He's very young, and I like him as much as the next Nationals fan, but right now he's the worst of a weak crop in the NL East. He's another guy who wouldn't hit higher than 8th on a good team.

Hitting the pitcher 8th can in fact create a few extra runs a year by creating some space between him and the very best hitters in the order. You just don't want automatic outs in front of Johnson and Dukes. Jordan Zimmermann was a two-way star at UW-Stevens Point and some think he might be a competent hitter, but Lannan, Olsen, Cabrera, Balester, and Martis are as automatic as any outs in MLB. Really, we should be looking to pinch hit for them as aggressively as possible.

9. Anderson Hernandez
I cringe when I think the team might actually be considering batting AHern in the #1 or 2 holes. Almost all the projection systems expect him to be among the absolute worst offensive regulars in MLB. We should be doing everything possible to minimize his plate appearances.

Using CHONE projections and the lineup analysis tool from Baseball Musings, which was developed based on similar principles to those laid out in The Book, we find that the above line up would produce 4.825 runs a game.

For comparison, that's 0.113 more runs per game than would be produced by what I'm guessing is the most likely Nationals line-up to be used by Manny this year:
1. Guzman
2. Milledge
3. Johnson
4. Dunn
5. Zimmerman
6. Dukes
7. Flores
8. Hernandez
9. Pitcher
A gain of 0.113 runs per game might not sound like much, but that's 18 runs over 162 games, which is nearly 2 more wins. If you had to pay for two more wins on the free agent market, it would cost several million dollars at least at fair market value. If we can squeeze out two extra wins for nothing just by properly arranging our line-up, let's do it!


Positively Half St. said...

Some conventional wisdom is simply not going to fade. i would love to see Nick (if healthy) get many more at-bats. Nick in the lead-off spot will never happen, though.

e poc said...

I think about this every now and then. Dukes, Johnson, Zimmerman, Dunn, Milledge, Guzman, Flores, pitcher, AHern seems like a good, realistic lineup. If Belliard plays, I'd want to put him sixth and Guzman ninth, but I doubt Guzman would be pleased about hitting ninth.

Unknown said...

Interesting. If Belliard plays 2b, I would have him lead off then Johnson, Dukes, Dunn, Zimmerman, Milledge, Guzman, Flores, Pitcher.

Realistically they would switch Dukes and Zimmerman, but as you say, the 5th spot is more important than the 3rd...

Unknown said...

Looks like this is what we're getting:

Manny all but confirmed the top 4 today and I'm guessing it would be Dukes-Johnson 5-6 so it wouldn't be LLRR.


If we're gonna have Milledge leadoff, I'd like to see something like this:

Matt Bergfield said...

Off topic, but Jordan Zimmermann's nickname should be Commissioner Jordan. The Commish.

e poc said...

Milledge-Guzman is a horrible 1-2.

Steven said...

Milledge-Guzman is indeed really pretty dumb (but remember, we're talking about tiny numbers of actual runs at stake here). Milledge isn't as dumb as Guzman.

Still, it's like managers look at body type and write the lineup based on that. Skinny dude, plays CF or SS... he's a lead-off man!

Dave Nichols said...

Milledge did show some ability to get on base in the minors, so it's not completely far-fetched. but Guz, as i wrote when he was re-signed, is nothing more than empty batting average and should be at the bottom of any batting order.

i agree about 1/2 St's CW comment though. it's tough for a bunch of mathletes to to erase over 100 years of history, even though all parties are trying to accomplish the same objective: winning more baseball games.

Steven said...

c'mon Dave. Do you have to be a mathlete to know that you're better off getting more at bats for your better hitters and fewer to your worse hitters? pretty much every softball team on the mall figures it out.