The Dodgers Are Leaving Runs on the Table
The Dodgers have been the talk of the baseball world this season, with the $2 billion sale of the organization and their league-best 17-8 record. However, as well as the team has played, they actually stand to improve their offense with a simple lineup tweak: Move Dee Gordon in front of the pitcher and bat A.J. Ellis near the top of the lineup.
By batting him eighth, the Dodgers are not optimally utilizing Ellis’s on-base proficiency.
Sure, some of his ability to draw walks so far is tied to his lineup spot. He’s a patient hitter regardless of where he bats — but batters in that position tend to get intentionally walked or unintentionally-intentionally walked more often than those in other spots. Even so, the difference in on base percentage between Ellis and Gordon is still substantial after accounting for those lineup effects.
The Dodgers are leaving runs on the table by batting an OBP-sinkhole in front of Matt Kemp and Andre Ethier. The situation is exacerbated by the fact that their eighth hitter has a .449 OBP and projects to finish at around .390.
While the overall effects of lineup optimization are negligible over the course of the season, effectively swapping Ellis and Gordon will help increase their scoring output and stave off some offensive dropoff when Kemp and Ethier regress.
First things first, an apology to Ellis is in order. I ranked the Dodgers dead last in our positional power rankings, and while the reasoning wasn’t entirely driven by Ellis, I had little confidence in his ability to sustain a .350+ OBP at the major-league level.
At the time, the Dodgers were seemingly going to split time between Ellis, Tim Federowicz and Matt Treanor. There was plenty of uncertainty surrounding Ellis and Federowicz, and Treanor is no longer an effective major leaguer.
I still consider my reasons somewhat valid, but the Dodgers definitely don’t have the worst catching situation, and would probably rank in the upper half if those rankings were done today. Ellis legitimately seems capable of posting a very high on base percentage in the major leagues this season, and ZiPS has taken notice.
The projection system forecasts a .372 OBP for Ellis the rest of the way, which would have him finish the season with an impressive .388 rate. Gordon, on the other hand, has a putrid .263 OBP right now and a .305 projection from here on out. It isn’t even close, and by batting Gordon at the top and Ellis at the bottom, Don Mattingly is reducing his team’s odds of having runners on base for the meat of the order.
Further, the case could be made that the eighth spot in a National League lineup is perfect for a player with Gordon’s attributes. If he manages to get on, he can utilize his speed by stealing second base. The pitcher can theoretically then sacrifice him over to third base. The Dodgers would then find themselves in a much better position to score a run, as Gordon could cross home plate on a grounder as well as on a sacrifice fly.
Ellis doesn’t offer the same advantage at the bottom of the order, as the difference in speed between he and Gordon is likely similar to the effects of their OBP differential. Not that Gordon’s speed is equally valuable but rather that the gap between the players in each category is similar. Ellis probably wouldn’t be able to steal second in this scenario — or even score from second on a single. While he clearly helps the Dodgers more at the top of the lineup, Gordon helps them more at the bottom, further justifying the swap.
The Dodgers have scored 103 runs, good for sixth in the National League. Just imagine how much better the offense would have looked in the early going with a .380 OBP to .400 OBP player in front of Kemp and Ethier, rather than one with a sub-.300 mark. Sure, the effects of lineup optimization are typically negligible — but in this case the effects could be greater, given the sheer difference in Ellis’s and Gordon’s rates of reaching base and the MVP pace on which Kemp is currently performing.
Eric is an accountant and statistical analyst from Philadelphia. He also covers the Phillies at Phillies Nation and can be found here on Twitter.
The notion that when Gordon gets on he will just steal second and then get bunted to third just doesn’t make any sense. You give the pitcher the red light, everyone knows Gordon is running, and then ask him to bunt with 1 or 2 strikes? Speed is best suited in front of singles hitters, it is almost negated in front of someone who gives themselves up. Having said all that, Gordon should be nowhere near the lead off spot with a .265 OBP, and I would argue that he shouldn’t be in the Major League.
That’s a good point. Furthermore, I assume the #8 hitter will get pitched around more frequently with one or two outs than he will with zero outs, and the SB/bunt/groundout sequence only results in a run if it begins with zero outs.
On the other hand, Gordon can’t help but walk more if he’s batting 8th, and regardless of the tactics utilized when the pitcher comes up (bunting vs. taking vs. swinging away), his speed is infinitely more valuable when he’s standing on first than when he’s walking back to the dugout.
Further, the # of times per season that Gordon leads off an inning from the 8 spot (~50), followed by getting on base (down to 15), and a successful sacrifice (say, 8), followed by a scoring grounder or sac fly (maybe 4), makes that a pretty negligible benefit.
Isn’t this is just a specific case of the flawed received wisdom that you should put speedy guys in front of power guys?
Better to put speedy guys ahead of singles hitters.
yep
What makes you think that Ellis is still going to have a .390 OBP in the leadoff spot?
Hitting leadoff is tough, and most players perform worse in that spot.
Who said anything about .390? Per the article, Ellis is currently projected to OBP .372 the rest of the way, which is fine for the top of the order. Even if he turns out worse than that, he’s virtually guaranteed to have a higher OBP than Gordon.
He’s going to be a hell of a lot better than Gordon if his OBP is .330. There’s really not an argument to be made that Gordon is a better lead off man.
This is what makes me think AJ will continue to have an extremely high OBP: http://www.fangraphs.com/leaders.aspx?pos=all&stats=bat&lg=all&qual=50&type=5&season=2012&month=0&season1=2012&ind=0&team=0&rost=0&age=0&players=0&sort=4,a
He swings at fewer pitches than anyone in baseball. He swings at fewer pitches outside of the zone than all but 3 players. And he makes very consistent contact, especially on those few occasions when he does get fooled and swing at pitches outside the zone. He’s never gonna be Matt Kemp, but dude can defintely get to first base.
For some extra AJ Ellis fun, check out http://ajellisfacts.tumblr.com/
When Ellis finally develops some power, should they move him to hit behind Kemp?
Not if their lead off man has a .260 OBP. The basic principle of line up construction should be “get the most at bats for your best hitters.” The concern about chaining hitters to complement weighs far too heavily when it defeats this principle.
A basic line up in descending wOBA order would be close to optimal. After that, there are a few tweaks worth making, but they won’t do you nearly as much good as getting a terrible hitter out of the lead-off spot (or second, as is fairly common.)
He jus turned 31… I think he probably has as much power as you’re ever going to see out of him. And if there is improvement, I wouldn’t expect it to really be of much significance. He is what he is at this point, it’s just nice to see him finally get a shot.
Per my own research here, it seems unlikely he’ll ever develop power http://www.fangraphs.com/blogs/index.php/will-a-j-ellis-develop-any-power/
This is fascinating. Thank you.
You got it
I’m about as big an AJ Ellis fan as you’ll find anywhere outside his immediate family, and even I don’t expect him to develop much power. I’m as shocked that he has 2 home runs already as I am that Pujols has none.
If Gordon gets on with two outs and the pitcher up, he’s not stealing. That means he’s not utilizing his best offensive attribute. Bat him 7th.
Dee Gordon is batting first
Ellis is battng 8th
Someone forward the same idea to the Scosciapath, who is batting Vernon Wells second since Trout’s call up.
he’s letting vernon wells clear the bases with double plays so that albert can come up with no one on, take some of the pressure off. Strategy!
Dodger fans want Gordon batting leadoff for a few reasons
1) because when Gordon is on base he “distracts the pitcher” leading to greater success for the batters at the plate. I would love to see this theory squashed or marginalized by some stats.
2) Gordon needs to as many ABs as possible to develop into the future stat he can be……..ugh
and well….that’s pretty much it. Once you can reason with a Dodger fan into agreeing that OBP is important for the top of the lineup, these two reasons are still thrown out there to justify batting Gordon leadoff.
Wow I goofed that second point. Scratch the “to” and replace “stat” with “star”
We knew what you meant 😉
In other news
Ethier should bat second
Kemp should bat 4th
and Gwynn and Rivera should be in a straight platoon.
Who’s gonna protect Kemp? The corpse of James Loney?
No. Kemp should not bat 4th. There is an argument that Kemp should bat 2nd. That said, 3rd is where he should bat traditionally.
Gordon or AJ leading off?
I used the ZIPs (U) for input projections for both Dee Gordon and AJ Ellis, giving Dee one of my top speed ratings and AJ one of my lowest. Batted them 1st/8th against LHP and RHP (4 sims) – with each sim ran 100K times.
Result was.
vs RHP, Dee(1), AJ(8) got 0.79 more wins per 162 games.
vs LHP, AJ(1), Dee(8) got 0.59 more wins per 162 games.
I think the ZIPs(U) favors Dee a little as they regress towards their pre-season projections pretty heavily. But these were the results nonetheless. Both players will regress, so you pretty much have to go with ZIPs(U) instead of actual empirical 2012 results as your data input.
“Batted them 1st/8th against LHP and RHP (4 sims) – with each sim ran 100K times.”
Could you better explain this, and your simulator in general?
Sure, I’ll try.
Four scenarios.
Lineup A: Gordon bats 1st, AJ Ellis bats 8th vs RHP
Lineup B: AJ Ellis bats 1st, Gordon bats 8th vs RHP
————-
Lineup C: Gordon bats 1st, AJ Ellis bats 8th vs LHP
Lineup D: AJ Ellis bats 1st, Gordon bats 8th vs LHP
rest of lineup stays the same 2-7, 9. Opposing team stays constant just the handedness of their pitcher changes. I had Dodgers be the Away team as to give them 9 gauranteed at bats per game, which makes the results come out cleaner.
Simulator plays actual games and takes in to consideration all the important things you can think of.
Interesting but I’m personally leery of running simulations on guys like this when there is so much unknown about their skill sets and abilities against different pitchers. Their current ZiPS(U)’s also clearly favor Ellis — unless we’re looking at something different, Ellis has a ZiPS(U) wOBA of .335 compared to Gordon’s .282. I don’t see any possible scenario in which Gordon is a better offensive player now or over the rest of the season, regardless of the input choice. Even looking at pre-season projections, Ellis was around 15 points better in wOBA.
In a case like this, I am more wont to trust the skills that have seemingly been established, which is that Ellis is an OBP machine and Gordon is the exact opposite. I wouldn’t lead Ellis off, but would probably bat him second to best utilize his primary skill.
Eric, I would be happy to run it with YOUR set of projections. Give me a per 600 AB projection for Gordon and A.Ellis with the following stats.
AB,H,2B,3B,HR,SO,BB
I don’t think it’s necessary to run simulations here, as I just said, because to the extent that the inputs are really telling, we don’t know enough about their tendencies. I don’t think you need a simulation to know that someone with an 80+ point OBP advantage over someone else, batting in front of someone as tremendous as Kemp, is beneficial to the team.
You need input data (projections) to even analyze the problem and you need a tool to quantify what the advantage is if any. Otherwise, you are just spitting in the wind here.
Of course projections are never perfect, so you build around a certain confidence level.
Xeifrank I’ve plugged in the ZIPS projections for Ellis and Gordon using other lineup simulators than yours and switching Ellis and Gordon in the order improved the Dodgers run scoring from 3.47 to 3.53 runs per game. or 6.5 runs over 162 games. I strongly believe your simulation system is inherently flawed. Earlier in the year MSTI posted a link to the optimized Dodger lineup you proposed. It was also full of errors.
Table, any method that does not take in to account handedness of opposing starting pitcher is badly flawed. Any method that does not take in to account the speed of the batters is badly flawed. Doesn’t sound like you have done either from your publishings.
Shhh! Mattingly might read this and actually do it.
Let sleeping Dodgers lie.
“While the overall effects of lineup optimization are negligible over the course of the season, effectively swapping Ellis and Gordon will help increase their scoring output…”
Huh?
Anyway, I’d rather see Gordon bat 6th, after the power guys and ahead of the Loneys and Ellis’s who aren’t going to get XBH – his speed will be most useful there, I think, and the detrimental effects of batting a low-OBP guy 6th vs. 8th are extremely small.
I was getting at the notion that, while lineup optimization doesn’t typically amount to much, this has the makings of an extreme case, where you would be replacing a sub-.300 OBP guy with a ~.375 OBP guy, batting in front of a guy having an MVP-type season.
How many extra runs per season would you get out of this?
I believe at the Book Blog they say that lineup optimization is 2-3 runs in an average case. If you call this “extreme”, pushing it to 5 runs could be justifiable as an upper bound (if Kemp is a monster, Ethier keeps hitting, etc).
So about half a win over 162 games.
Yeah one simulator I used said the change would produce 6.5 runs…so yeah a little more than 1/2 a win….
Putting Gordon at 8th might kick his butt into working harder. He’ll make a good leadoff hitter someday. But not as long as he thinks everyone is OK with his current stats, which is the feedback he is currently getting as the leadoff hitter.
The “taking off the catcher’s crap to bat” logistics occur at least once a game with Ellis leading off. That’s probably keeping him 8th. Not saying it’s right. Just saying it’s the reasoning.
Whether or not Mattingly subscribes to OBP, Gordon’s average will stay in the 230-250 range, which nobody can overlook, sabermetrically inclined or not. Eventually the move will be made, and with every passing game I scoff at the fact that AJ Ellis is batting 8th. At the very least he should be switched with his evil step-brother Mark Ellis in the 2-hole.
Not only should AJ Ellis be hitting first or second, but I think Gordon should be hitting Ninth – especially in games where Kershaw and Billingsley are playing.