Optimal Lineup Construction (?) [Archive]

View Full Version : Optimal Lineup Construction (?)

incarnadine

03-31-04, 09:47 AM

This has come up a few times with regard to questions about the Yankees leadoff position (Viz a viz Jeter or Lofton. I come in on the Jeter side of that question, and have reservations about Lofton having a regular spot in the lineup anywhere, let alone getting the most plate appearances every game). So I thought I'd start a thread, and see if I can tap the knowledge and opinions of some of the statheads and assorted logical-type thinkers on here.

Why not just construct a lineup as follows:

Order by decreasing OPS (or OPS+ or 1.8*OBP+SLG, or whatever metric you believe measures the overall "quality" of a hitter), taking care that you don't have more than two batters in a row of the same handedness. (To frustrate late-inning bullpen mix-and-match by the opposition.)

Who cares if your leadoff man has basestealing ability? He's only guaranteed to lead off the 1st inning, and are you really going to be swiping bags in the 1st inning?

For the Yanks, why not get Sheff (or Giambi or ARod) a few dozen extra PA per year, and maybe an extra one per game? Get your best hitters up most often, and you should score more runs.

For Boston, why let a guy like Damon get to the plate that often? Slap Manny or Ortizzle up there. Let them give the pitcher a rude awakening as early as possible in the game.

AngelAstro

03-31-04, 10:20 AM

I'm just reposting from the other discussion where you posted this same idea.

I'm not sure going by descending OPS is necessarily the right way to go, but I like your idea. The reason I say OPS may not be the right way to go is because Jeter's career OPS is .851 consisting of an OBP of .389 and SLG of .462. In Jeter's case his "low" OPS diminishes his great ability to get on base. Plus, since he doesn't have a very high slugging, you wouldn't want to bat him behind the likes of Giambi and Sheffield and A-Rod.

In my opinion it's best to have high OBP guys up in front of the high slugging guys to maximize the potential for driving in runs.

nyybleachercreature

03-31-04, 12:51 PM

I agree with the last statement. I think a batter that gets a high OPS due to OBP would be better suited to be high in the lineup than a guy that gets the same OPS due to high Slugging. The guys with the high OBP would be on base all of the time, setting up more RBI opportunities for the guy with the high Slugging. I think the way the lineups are constructed now works well.. High OBP guys up front, followed by high slugging, followed by (on most teams except the Yankees) the terrible hitters of the bottom 1/3 of the lineup.

incarnadine

03-31-04, 01:06 PM

Yes, but that sort of thinking sort of assumes that you start an inning in a predictable place.

<ul>
<li>OBP Guy 1
<li>OBP Guy 2
<li>OBP Guy 3
<li>Mr. Slug
<li>Sluggy McSlug
<li>Arthur Slugworth (President, Slugworth Chocolates)
<li>Crappy Hitter 1
<li>Crappy Hitter 2
<li>Enrique Wilson
</ul>

If you start an inning in the first third of the order, all is well. But say Arthur starts the inning, and leads off with a sharp single. The bottom third may well fail to drive him home.

It seems like the only thing you can count on is that a leadoff man gets 20 or 30 more PA's per season than a cleanup hitter. Those plate appearances should go to the guy who's the most offensively productive.

incarnadine

03-31-04, 01:10 PM

This more properly belongs in Sabermetrics Symposium. Please move the thread.

AngelAstro

03-31-04, 01:20 PM

Hmmm. This is something to think about.

We know that the lineup is only truly important in the first inning, because this is the only time the lineup doesn't fluctuate at the top of the inning. So, I suppose that in the first inning you want to play to try to score the most runs. Therefore, you bat the players that have the best chance to get on base first followed by the players with the best chance to drive them in.

So I believe the best way to construct your lineup would be to have

OBP 1
OBP 2
OPS 1
OPS 2
SLG 1
SLG 2
SLG 3
Junk (if you ask me, someone with speed)
Enrique Wilson

(When I say OPS, I mean someone who has a high OBP and high SLG. While OBP 1-2 means someone with high OBP while SLG doesn't matter as much. SLG 1-3 means someone with high SLG while OBP matters less, though preferably SLG 1 will have a higher OBP than SLG 2, etc.)

This way, the first two guys could get on base, followed by guys who can get on base and/or can drive runners in (usually the number 3 hitter is the best hitter on the team). Then would come the guys who could slug, but may have a harder time getting on base. And then you have your all field/no hit players. Or in the Yankees case, Enrique Wilson.

Even if the inning begins with say SLG 1, chances are the inning will end at Junk or Enrique Wilson, so that the next inning OBP 1 will lead off anyway - just because the player with the highest chance of getting out (low OBP) is going to be the 8th or 9th hitter. Thus, if a majority of the time the 8th or 9th hitter will be the last out, the next inning you want somebody up who has a high chance of getting on base. Remember, the key to a good offense is to not make outs.

This is how I see it, I'm definitely open to discussion.

incarnadine

03-31-04, 02:56 PM

Angel:

That seems pretty close to what I'd think, but there is a pretty good corollation between high OBP guys and high SLG guys. Maybe good enough to just conflate the two and not worry about it.

Here's the AL top 10 OBP:

<ul>
<li>M. Ramirez
<li>C. Delgado
<li>Ja. Giambi
<li>E. Martinez
<li>J. Posada
<li>B. Mueller
<li>T. Nixon
<li>A. Rodriguez
<li>D. Mientkiewicz
</ul>

Here's the AL top 10 in SLG:

<ul>
<li>A. Rodriguez (9 in OBP)
<li>C. Delgado
<li>D. Ortiz (19 in OBP)
<li>M. Ramirez
<li>T. Nixon
<li>F. Thomas (12 in OBP)
<li>A. Huff (20 in OBP)
<li>V. Wells (27 in OBP)
<li>M. Ordonez (14 in OBP)
<li>G. Anderson (42 in OBP)
</ul>

Here are the batters who were top 10 in OBP but not SLG:
<ul>
<li>B. Mueller (11 in SLG)
<li>Ja. Giambi (14 in SLG)
<li>E. Martinez (26 in SLG)
<li>J. Posada (19 in SLG)
<li>D. Mientkiewicz (47 in SLG)
</ul>

So there are the rare pop-less wonders who can get on base (Minty from the Twins) and the occasional free-swinger who hits it hard but rarely walks (Garrett Anderson and Vernon Wells). But for the most part, the guys who are patient enough to get on base all the time, also see enough pitches to get one they can really sock.

AngelAstro

03-31-04, 03:00 PM

And most/all of those players are among the best hitters in the game, which is not surprising. Of course, if you could have all those players then I could see making a lineup from highest to lowest OPS. However, as close to perfect as the Yankees are, they aren't quite perfect. So you have to work with what you have.

incarnadine

03-31-04, 03:11 PM

I'm trying to keep the discussion general, but to use the Yanks as a case study, I'm going to put up the OBP/SLG/OPS of the nine presumptive starters and see how that lineup looks.

<ul>
<li>Sheffield (419/604/1023)
<li>Rodriguez (396/600/996)
<li>Giambi (412/527/939)
<li>Posada (405/518/922)
<li>Jeter (393/450/844)
<li>Lofton (352/450/801)
<li>Matsui (353/435/788)
<li>Williams (367/411/778)
<li>Wilson (276/363/639)
</ul>

Observations:
Sheff had the best 2003 OBP and best SLG.
A-Rod had the second best SLG, but fourth OBP.
Jorge (again, but it keeps striking me) had a great year for a catcher.
Bernie had a miserable season.
Enrique had a cringe-inducing season at the plate as a utility guy.
Lofton is a much worse hitter than OPS would say, due to a bad OBP.

How would you arrange these? Or would you slap in Clark in place of Lofton as per the DH/CF/1B merry-go-round?

AngelAstro

03-31-04, 03:18 PM

Plus, in the first inning, would you rather have Giambi at-bat first with no chance of knocking in anybody but himself? I think you take your chances with your high OBP guys because if they do get on and then Giambi blasts one you have a 2, 3 or 4 run lead rather than a 1 run lead.

With Torre, I'm surprised he doesn't lead off Giambi considering how much he loved that 1 run lead in the first inning, since he always batted Soriano lead off. With my lineup construction, the best lineup last year would have been:

Name OBP SLG
Jeter .393 .450
Johnson .422 .472
Giambi .412 .527
Posada .405 .518
Soriano .338 .525
Boone .327 .453
Matsui .353 .435
Williams .367 .411
Garcia .302 .422

This is using last years numbers. In this case Boone, Matsui and Bernie are probably interchangable.

Of course Bernie's numbers are far from his career .390/.492. So if Bernie was having a typical Bernie year, he should have bat 5th with Soriano 6th.

That's my opinion.

incarnadine

03-31-04, 03:21 PM

I am in fact quite worried about Posada not being high enough in the lineup. He needs to get his AB's at the expense of guys like Matsui.

AngelAstro

03-31-04, 03:23 PM

Originally posted by incarnadine
I'm trying to keep the discussion general, but to use the Yanks as a case study, I'm going to put up the OBP/SLG/OPS of the nine presumptive starters and see how that lineup looks.

<ul>
<li>Sheffield (419/604/1023)
<li>Rodriguez (396/600/996)
<li>Giambi (412/527/939)
<li>Posada (405/518/922)
<li>Jeter (393/450/844)
<li>Lofton (352/450/801)
<li>Matsui (353/435/788)
<li>Williams (367/411/778)
<li>Wilson (276/363/639)
</ul>

Observations:
Sheff had the best 2003 OBP and best SLG.
A-Rod had the second best SLG, but fourth OBP.
Jorge (again, but it keeps striking me) had a great year for a catcher.
Bernie had a miserable season.
Enrique had a cringe-inducing season at the plate as a utility guy.
Lofton is a much worse hitter than OPS would say, due to a bad OBP.

How would you arrange these? Or would you slap in Clark in place of Lofton as per the DH/CF/1B merry-go-round?

You beat me to it, I was going to do that next. I'll assume Lofton in center, Bernie at DH. I would have:

Jeter
Bernie (I'm assuming last year was because of injury)
Sheffield
Giambi
A-Rod
Posada
Matsui
Lofton
Wilson

Actually, no matter if it's Lofton or Clark or Lee, any should bat 8th in this line up.

As you mentioned, last year Posada was huge. I also think that it shows that last year Matsui was nothing special, even Lofton was better.

AngelAstro

03-31-04, 03:25 PM

Originally posted by incarnadine
I am in fact quite worried about Posada not being high enough in the lineup. He needs to get his AB's at the expense of guys like Matsui.

Absolutely. Torre does like to mix and match lefties and righties though, so Posada will most likely bat behind Matsui. I don't know if Torre is right or wrong about alternating lefties or righties. His premise makes sense to me, though in practice it probably makes no difference.

AngelAstro

03-31-04, 03:26 PM

Also, if Bernie struggles, it may be wise to move Posada to number 2 and move Bernie down to 6th or 7th.

nyybleachercreature

03-31-04, 07:47 PM

Originally posted by incarnadine
Yes, but that sort of thinking sort of assumes that you start an inning in a predictable place.

<ul>
<li>OBP Guy 1
<li>OBP Guy 2
<li>OBP Guy 3
<li>Mr. Slug
<li>Sluggy McSlug
<li>Arthur Slugworth (President, Slugworth Chocolates)
<li>Crappy Hitter 1
<li>Crappy Hitter 2
<li>Enrique Wilson
</ul>

If you start an inning in the first third of the order, all is well. But say Arthur starts the inning, and leads off with a sharp single. The bottom third may well fail to drive him home.
Also, sabermetrically, doesn't lineup not matter?

It seems like the only thing you can count on is that a leadoff man gets 20 or 30 more PA's per season than a cleanup hitter. Those plate appearances should go to the guy who's the most offensively productive.
Is 20-30 more ABs in an entire season reason to take away at least one RBI chance per game from your best OPS guy? If your highest OPS guy bats first, he is going to have to rely on the bottom 1/3 of the lineup (the lowest three OPS) to get on base in front of him to get RBIs. If the highest OPS guy were batting 3rd or 4th, he would average at least 1 good look at an RBI every game.
Also, sabermetrically, doesn't lineup not matter?

incarnadine

03-31-04, 07:57 PM

Originally posted by nyybleachercreature
Also, sabermetrically, doesn't lineup not matter?

Well, that's my question.

Not only do I not know what the conventional sabermetric answer to the lineup question is, but even if I did, I'd want to know why.

The question's are:

<ul>
<li>What is the expectation for number of PAs for each position in the lineup?
<li>Does position in the lineup correlate meaningfully with getting AB's with RISP?
<li>If so, how does X more "ducks on the pond" compare with another 20 or 30 ABs per season for a franchise slugger?
</ul>

I don't know the answers to any of these. (But I am in love with using HTML code in forum posts, especially unordered lists and italics. As you can tell.)

nyybleachercreature

03-31-04, 10:54 PM

I dont know why, but I do know Ive heard a bunch of sabermetrics people (Im pretty sure Neyer was one) say that lineup construction has nothing to do with how well someone will hit.
I think 20-30 ABs (what, like .055% of ABs, being very liberal) dont have as much of an effect as having runners in scoring position for a batter to knock in much more often (the difference between, say, two guys at the bottom of the lineup with OBPs of like .320 vs two guys at the top of the lineup with OBPs of high .300s-low .400s). I dont know if this is mathematically right at all, but it seems that, even if there were a man on base for all of the extra .055% of AB (for the decreasing OPS situation), it would be less than the difference in OBP of .07% (for the traditional style).

incarnadine

03-31-04, 11:08 PM

I've read that there is no statistical evidence of "protection" in the lineup. But that isn't the same thing as saying that any of the 362,880 (9!) permutations of a given lineup will all generate the same number of runs.

markp

04-02-04, 07:41 AM

actually, Bill James said it doesn't matter nearly as much as he thought, not that it didn't matter at all.
Since the first three batters get quite a few more ABs over a season than the last three, having your best hitters there certainly makes a difference over 162 games.
The argument that "a lead-off hitter might only lead off one inning in a game" ignores another important thing: that whoever hits in front of your sluggers does it every time the line-up turns over. If you have high OBA guys (Like Bernie and Derek) hitting 1-2 and big slg guys like Arod, Jason, and Sheff they'll come up before them several times a game whether or not they lead off innings to do it.

AngelAstro

04-02-04, 08:16 AM

Originally posted by markp
actually, Bill James said it doesn't matter nearly as much as he thought, not that it didn't matter at all.
Since the first three batters get quite a few more ABs over a season than the last three, having your best hitters there certainly makes a difference over 162 games.
The argument that "a lead-off hitter might only lead off one inning in a game" ignores another important thing: that whoever hits in front of your sluggers does it every time the line-up turns over. If you have high OBA guys (Like Bernie and Derek) hitting 1-2 and big slg guys like Arod, Jason, and Sheff they'll come up before them several times a game whether or not they lead off innings to do it.

That's basically the point I was trying to make. Thanks for making it clear because I was obviously having trouble doing so. :)