The Optimal Red Sox Lineup … Or Maybe the Only Way to Kill the Sox Offense
Doing my daily baseball reading, I came across an optimal lineup generator that David Pinto of Baseball Musings had coded, based on a Perl script that Ken Arneson of Catfish Stew wrote, based on the work of Cyril Morong from Beyond the Box Score. Morong has taken Retrosheet data from ‘89-’02 and ran a regression analysis to determine the coefficients for OBP and SLG for each lineup slot.
Got all that?
Then it turns out that BoSoxBob over at HireMeTheo ran the lineup analysis using the ZiPS projections from Baseball Think Factory. He found that the best lineup had Manny Ramirez second and Coco Crisp third. I’m guessing that’s pretty optimistic projection for Crisp (.303/.353/.448).
Well, given that I dork around in Perl most of the day for a living, and given that Ken Arneson did all the work already anyway, I figured I’d toss some Red Sox projections into the script and see what it spit out. Given that I’m partial to the heavily regressed Marcel the Monkey projections, I decided to use those. In the 2006 Marcels, Crisp’s line is a slightly more palatable (for my tastes, anyway) .293/.345/.439.
So … without futher ado … the optimal 2006 Boston Red Sox batting order!
Mark Loretta (.299/.371/.422)
Manny Ramirez (.295/.400/.554)
Jason Varitek (.275/.365/.467)
David Ortiz (.290/.383/.563)
Trot Nixon (.282/.367/.479)
Mike Lowell (.264/.340/.439)
Coco Crisp (.293/.345/.439)
Alex Gonzalez (.251/.315/.406)
Kevin Youkilis (.268/.363/.426)
According to the projections and the lineup analyis, that batting order would score 944 runs. However, it’s not terribly likely that we’ll see Terry Francona trot out a lineup like that this year. Ortiz/Nixon back to back won’t happen very often, and I don’t think Terry will ever bat Manny second. Plus, it sounds like they’re leaning towards batting Crisp leadoff, for “traditional baseball reasons” (i.e. he’s fast and plays CF) and probably because it’s also better PR to do that given that he’s essentially replacing Johnny Damon.
The likely batting order will probably shakeout like this:
Crisp
Loretta
Oritz
Ramirez
Nixon
Varitek
Lowell
Youkilis
Gonzalez
That lineup, according to the analysis, would score 920 runs — 24 runs less than the optimal lineup. 24 sounds like a lot — and it is, given that 10 runs of offense generally improves your team by 1 win — but over the course of the season that might not be worth the headaches that one would deal with when using the optimal lineup. One anecdote: given that there are about 360k permutations of a 9 player lineup (9! or 9*8*7*6*5*4*3*2*1), there are approximately 140k lineup combinations that would score more runs than the Sox likely lineup. In most cases, it’s literally 1/100th of a run, but I just thought it was an interesting little nugget for you to use on days when the Sox aren’t scoring … “Hey Francona … there’s like a hundred thousand better lineups you could have trotted out today … jackass.”
Still, there’s got to a better lineup that won’t cost the Sox 2 wins, but also will be palatable to the baseball traditionalists. The best I could find:
Loretta
Youkilis
Ortiz
Ramirez
Nixon
Lowell
Varitek
Gonzalez
Crisp
This version would score 931 runs, so it’s about 1 win worse than optimal, but 1 win likely than the likely lineup. Given that the Sox finished in a dead heat with the Yankees last season, I’ve just found us an AL East title. Go me.
Of course, now let’s discuss the fun questions. How could Terry Francona get himself canned? What lineup is green kryptonite to the Sox offense? What lineup would Grady Little run out there?
Ok, the Grady Little shot was a bit much. But here’s the Sox worst possible lineup:
Gonzalez
Crisp
Nixon
Youkilis
Lowell
Loretta
Varitek
Ramirez
Ortiz
The Bad News Sox would still score 887 runs, but losing almost 60 runs off of your season total might very well be the difference between 1st and 3rd place in the AL East this season.
So there you have it. The optimal Red Sox lineup (using the 2006 Marcel projections). Of course, this is the sort of thing that just spawns more questions (How does PECOTA make the Sox lineup shakeout? What happens if you replace Coco Crisp with Johnny Damon? Randy Winn? What if Lowell is complete ass again this season?).
Well - those questions will need to wait for a future installment. However, I was curious enough to throw the 2006 Johnny Damon projection in and take a look at the production of the likely Red Sox lineup (Damon/Loretta/Ortiz/Ramirez/Nixon/Varitek/Lowell/Gonzalez/Youkilis).
Let’s just say that according to the Marcels, Coco Crisp better be at least 15 runs better than Damon defensively in CF.
February 26th, 2006 at 7:12 am
Nice piece. I am a firm believer that Mark Loretta should be our lead off hitter. Nothing will change that belief until the regular season starts. Then, other worrisome bits and pieces, I’m sure, will enter into my thought processes.
February 27th, 2006 at 4:33 pm
I was meaning to check what the lineup would look like using Marcels - thanks for doing the work for me. I think that the team run totals you’re getting are more reasonable than the ZiPS results because of the heavy regression in Marcels. The lowest run total I got with ZiPS was still 956 runs.
I disagree with you about Crisp though - if anything, I expect him to do better than the ZiPS projection. Check out this story at HireMeTheo.com in which I computed composite park factors for each game Crisp played last year. I then computed composite park factors for the Sox 2006 schedule. By scaling Crisp’s 2005 stats by the ratios of the park factors, I found that Crisp’s .796 OPS would translate to an .857 OPS at Fenway.
FYI - I was the one who got Pinto motivated to write his lineup analysis tool, after alerting him to my story on the Sox lineup.
March 21st, 2006 at 8:53 am
I don’t like Loretta starting off, but he should be at the top. You need a speed guy who can hit at your number 1 spot there, and Crisp is just what the doctor ordered.