Optimizing the Rockies lineup

by PioneerSkies on Mar 18, 2009 10:39 PM MDT

I recently read Optimizing Your Lineup By The Book on the SBN blog Beyond the Box Score. In brief, it contrasts traditional baseball notions and statistical driven approaches to the batting order. The article draws heavily on The Book, as it breaks down each spot in the lineup. Rather than attempting a more detailed synopsis, I'll just recommend you read the article itself, as the rest of my post will rely heavily (though not entirely) on it's analyses.

After reading the article, I decided to construct an idealized lineup for the Rockies. I started by looking at the OBP and SLG for each potential starting position player. Granted, this is a rather limited inquiry, but from a plain reading of the article, other statistics didn't seem very crucial. I looked at both career OBP and SLG as well as the 3 year averages for each player. For Ian Stewart and Carlos Gonzalez, I also looked up their career OBP and SLG in the minors, due to small major league samples. Dexter Fowler has put up crazy numbers in the minors (.394/.452), but his September callup (.185/.154) shows he needs some more development. Also, with Seth Smith and Matt Murton, I focused on their LHP/RHP splits, since I buy into a Meth Smirton platoon in LF. I ranked everybody by OBP in descending order and ended up with (3 year averages in parentheses):

Helton .428/.574 (.414/.464)
Smith .366/.466 (.363/.448) v. RHP .395/.514
Hawpe .375/.493 (.384/.515)
Murton .354/.438 (.346/.422) v. LHP .382/.484
Spilborghs .374/.466 (.375/.466)
Stewart .337/.443 (.339/.454) minors: .373/.524
Iannetta .368/.440 (.368/.440)
Atkins .360/.474 (.368/.498)
Tulowitzki .346/.435 (.346/.435)
Gonzalez .273/.361 (.267/.351) minors: .340/.473
Podsednik .337/.374 (.332/.356)
Baker .319/.468 (.316/.473)
Barmes .301/.405 (.287/.389)

With these numbers in mind, the three best everyday players--based on OBP--are Helton, Hawpe and Spilborghs. The other thing that is painfully clear is that Ian Stewart needs to be in the lineup. For my purposes, I put Stewie at 2B. Using these numbers alone--particularly SLG, Atkins might appear to be better than CDI, but I give the slight edge to Dreamy because Atkins power has been trending down the last three seasons (.556, .486, and .452). Based on my interpretation of the numbers and the framework from the article, I constructed these two lineups:

v. RHP
1) Spilborghs
2) Helton
3) Stewart
4) Hawpe
5) Smith
6) Iannetta
7) Atkins
8) Tulowitzki
9) Marquis

v. LHP
1) Spilborghs
2) Helton
3) Stewart
4) Hawpe
5) Murton
6) Iannetta
7) Atkins
8) Tulowitzki
9) Marquis

Yeah, I realize Jason Marquis can't pitch everyday, but I put his name down because these are idealized lineups and Marquis is the best hitting pitcher the Rockies have had since Jason Jennings.

These lineups look a little strange even to me. Hawpe hitting cleanup? Atkins and Tulo at the bottom of the order? Four lefties in a row? My own discomfort is probably based on two factors. First, I focused on OBP far more than SLG. Jeff Baker would argue that his .473 SLG more than makes up for a lower OBP. Second, I was raised by a baseball purist (but not elitist), and the traditional notions about the batting order still have some sway on my psyche.

Nevertheless, I enjoyed this meaningless little exercise. It's always fun to dig into baseball statistics, and it's made me more comfortable with a few things about the roster. Spilly will be a solid leadoff hitter. Smith and Murton will be a dynamite combo in LF. This is great news, not only for the Rockies offense, but also for the studs that are developing. Dex and CarGon (or G-Gon, as I've heard he wants to be called) will be able to get their work in at AAA without pressure to produce immediately at the major league level. These players might not enjoy being 'relegated' to the Springs, but in the long run, the additional development will pay off.

So what do you think? Am I way off on the Rockies ideal lineups? Did I use a flawed methodology? Does Jeff Baker deserve to be the starting 2B??? Sound off!!!

Eat. Drink. Be Merry. But the above FanPost does not necessarily reflect the attitudes, opinions, or views of Purple Row's staff (unless, of course, it's written by the staff [and even then, it still might not]).

Tweet Comment 27 comments | 0 recs |

Do you like this story?

Comments

Very good stuff

It’s good to see some motivated analysis going on…I had actually read that Beyond the Boxscore article earlier this evening and played with the idea of putting something like this together. Good job.

A couple comments…

1) We discussed this a bit earlier this offseason. This website generates an runs/game estimate for every possible lineup combination. Quite frankly, most people balk at the strange choices generated (like Helton leadoff) but it’s a fun toy. And it takes SLG and OBP into account. Then again, the method you cite that lists the 3-hole hitter as the 5th most important batter is blasphemous to those that strictly subscribe to the traditional ideology that the 3-hole hitter should be the best hitter.

2) A huge part of this analysis stems from the fundamentals of prediction. Which is best? 3-yr averages? Career numbers? In my opinion, recent past results (thus 3 yr averages) are generally good for veterans, but the small sample size introduces more variability for young players. This is often taken into account through several projection models…which you can find here. That might be good for the likes of Smith, Stewart, etc.

3) It sounds like you would be very interested in The Book. I do recommend it; just know beforehand that it will require trudging through a LOT of statistics. Still, there’s a lot of stimulating prose regarding this type of analysis. (And fyi, they do propose that a pitcher should hit 8th, and putting him 7th or 9th are about equal)

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 18, 2009 11:38 PM MDT reply actions

2)...

I like to take some lessons from the financial world…. The biggest disclaimer you will see is Past Performance does not Predict future Results. I think the numbers on Barmes are misleading due to the deer meat mis-hap. I think this year will be Barmes’ Renaissance year.

The other thing you’re not considering is defense. Tulo and Barmes up the middle is a great DP combo. I say make it a Math Smirtonart combo and also work Stewy in to the Toddfather’s off days.

I think it would be interesting to take a page out of Larussa’s book and bat Barmes in the 9 hole.

by jefe_13 on Mar 19, 2009 12:00 AM MDT up reply actions

Well…which method do you think has more accuracy? Barmes’ 3-yr average or “I think this will be Barmes’ Renaissance year?” It’s often difficult as a fan to remove optimism or selective memory and personal impressions from our projections. Thus, in baseball statistics, past performance is a HUGE part of projections.

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 19, 2009 12:07 AM MDT up reply actions

Not so much selective memory...

On the same lines of factoring in small sample size and minor league play for other players, you have to look at Barmes’ 2007 play in the minors, which was most of his AB’s (428 AAA vs 37 MLB). In the Springs, he had a .393 wOBA and in 2008 he was .370 wOBA. In his case, I say throw out the uncharacteristical 2006 and it looks like he’s headed for a good year.

by jefe_13 on Mar 19, 2009 12:29 AM MDT up reply actions

That is true

But you also have to adjust his AAA numbers down due to league difficulty

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 19, 2009 10:09 AM MDT up reply actions

Or just use MLB projections

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 19, 2009 11:00 AM MDT up reply actions

natürlich

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 20, 2009 11:39 AM MDT up reply actions

I think this year will be Barmes’ Renaissance year.

You just say that because I think 2009 will be a Renaissance season for Todd Helton.

Aut Vincam Aut Periam

by PioneerSkies on Mar 19, 2009 10:14 PM MDT up reply actions

Here we go again…

We are truly in the presense of greatness here…-- unnamedDBacksfan

by DbacksSkins on Mar 20, 2009 12:35 PM MDT up reply actions

Stats

I found Poseidon’s Fist article on wOBA intriguing.

With the help of some nifty excel formulas, based on your numbers above, here’ s the simple wOBA (2*OBP+SLG)/3 for each player…

Player Career (3 yr)
Helton .477 .431
Smith .399 .391 vs RHP .435
Hawpe .414 .428
Murton .382 .371 vs LHP .416
Spilborghs .405 .405
Stewart .372 .377 minors .423
Iannetta .392 .392
Atkins .398 .411
Tulowitzki .376 .376
Gonzalez .302 .295 minors .384
Podsednik .349 .340
Baker .369 .368
Barmes .336 .321

by jefe_13 on Mar 18, 2009 11:51 PM MDT reply actions

I'm glad you found it intriguing

That was the aim. Still, for players like CarGo, Smith and Stewart, there needs to be some manual manipulation aside from statistical career averages due to small sample size. Also note that wOBA does not involve any sort of park adjustment, so Podsednik, Murton and Gonzalez would figure to get a small boost due to playing at Coors.

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 19, 2009 12:10 AM MDT up reply actions

The book also says

it’s good to have someone who strikes out a lot batting second to reduce DP’s. But some of these lineup tweaks are so minor. I haven’t picked up The Book in a while but if memory serves me right, batting the pitcher 8th would lead to the team scoring two more runs per season on average…

by Resolution on Mar 19, 2009 7:08 AM MDT reply actions

That's exactly correct

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 19, 2009 10:09 AM MDT up reply actions

You might also consider

Using projections rather than using past numbers/minor league numbers.

I’d advise checking out CHONE and seeing how that stacks up.

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 19, 2009 9:00 AM MDT reply actions

I agree with using projections

http://baseballprojection.com/COL2009.htm

Good stuff, though.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 19, 2009 2:35 PM MDT up reply actions

a few more thoughts...

1) The Book says that hitting a pitcher 8th creates two more runs in a season. I deliberately chose not to do this with my idealized lineups because I couldn’t bear the thought of having Tulo bat 9th after the pitcher. Having him in the 8 hole is bad enough!!

2) Using the lineup analysis site, the batting order that I dreamed up would score 5.687 runs per game. To put this in context, 5.687 runs per game is 921 runs scored in a 162 game season (rounding down). In 2008, the Texas Rangers scored the most runs in the MLB with 901. While using this website, I plugged in the 3 year averages for each player, with some exceptions: I used career numbers with Jason Marquis (.227/.306), I used Stewart’s career minor league numbers as above, and I used the arithmetic mean of Smith and Murton’s splits (rudimentary, I know) to create Smirton (.388/.499). The lineup analysis site also tells me that the best lineup using these players would score 5.738 runs per game (or 929 runs in a season), and would be as follows: Helton, Hawpe, Atkins, Smirton, Stewart, Marquis, Spilborghs, Tulowitzki, Iannetta. Can you imagine the pitcher hitting sixth???

3) It’s true that

Past Performance does not Predict future Results

so probably the biggest flaw with my methodology is that I worked with historical statistics instead of projections. Personally, I don’t know which projection system is best or most accurate. I do know that the good people at FanGraphs list five different statistical projections for each player. I suppose my next task is to start over, average all five projections for each potential starter, and construct new idealized rosters from there. This isn’t an impossible task for me, but I don’t think I’ll be able to complete it until after I return from Spring Training.

Thanks for the input folks!

Aut Vincam Aut Periam

by PioneerSkies on Mar 19, 2009 11:05 PM MDT reply actions

Two answers, sort of:

Bill James is overly optimistic. Marcel is ok, but there’s better. CHONE‘s my favorite, but it’s not demonstrably better than MINER or ZiPS (or PECOTA).

“Tulo in the 8th hole is bad enough” comes from viewing each player’s value by their ability to drive in runners. If you instead view their contributions by the probability they’ll score (which isn’t perfect, either), then you instead want Tulo after the pitcher. And that’s where the value of batting the pitcher 8th comes from. Yes, the 6th and 7th hitters won’t get knocked in quite as often, but there will be more runners on base for the 1st and 2nd hitters (who should be really good hitters). We can’t measure the that trade-off intuitively, but simulations show it’s a small advantage. The ideal 9th hitter is OBP-heavy, as he’s a “second leadoff hitter.”

Looking through the CHONE projections, it looks like the Rockies lineup is full of a lot of similar hitters, guys with OBPs in the .350 to .370 range and SLG’s in the .440 to .480 range. Hawpe and Helton are outliers. I’d go…

Spilburgs
Helton
Atkins
Hawpe
Murton
Iannetta
Murton
Tulo
Pitcher
Stewart

But since CHONE predicts Atkins (3rd best hitter) to be only six runs better over 150 games than Stewart (worst hitter listed), there aren’t a lot of choices to be made.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 20, 2009 7:57 AM MDT up reply actions

who is spilburgs

is that a town

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 20, 2009 9:21 AM MDT up reply actions

Thanks for the appearance

Good advice….but even though Murton has impressed in ST, I don’t think I want him hitting twice in the lineup ;)

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 20, 2009 11:41 AM MDT up reply actions

Again, you really, really need to use projections

Saying Stewart is the same player in AAA as he is in MLB is a major flaw in these lineup constructions.

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 20, 2009 9:20 AM MDT up reply actions

to clarify

just pick one and go with it. Or maybe construct several lineups based on CHONE, ZIPS, MArcel, and Bill JAmes.

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 20, 2009 9:21 AM MDT up reply actions

I'd be interested to see that

We might end up with a 1000+ run team with BJ, but I think it would be worth looking at all of them…

"I have no special talents. I am only passionately curious." - Albert Einstein

by Andrew T. Fisher on Mar 20, 2009 11:42 AM MDT up reply actions

Yes.

If you just start throwing out names, people might disagree for two reasons: they disagree on how good each player is and they disagree on how the lineup should be ordered. By stating what stats you’re using, you can at least figure out where you disagree. You can favor one lineup with one set of performance assumptions and a different one with different assumptions.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Mar 21, 2009 11:46 AM MDT up reply actions

my current project for pitcher projections

which is totally unaffiliated with PR involves me literally combining PECOTA, CHONE, ZIPS, and Marcel. I broke everything down into component stats (K, BB, HR, IP), summing them all, and then finding the cumulative rate stats of all the projection systems. I figure that should be right.

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 21, 2009 12:26 PM MDT up reply actions

splitting hairs...

does any projection system predict a player’s lefty/righty splits?

Aut Vincam Aut Periam

by PioneerSkies on Mar 20, 2009 10:25 PM MDT up reply actions

No

But if you look at baseball reference/minor league splits and look at their career splits, you can probably make an educated guess on what they’d look like.

Follow me on Twitter! http://www.twitter.com/rockiesmagicnum

by Andrew Martin on Mar 21, 2009 12:24 PM MDT up reply actions