Aim For The Head: Are Balanced Lineups Better?
by Keith Woolner
This week's question comes from J. M., who writes:
Based solely on offense, expected runs created - given the scenario that your
total starting lineup team OPS was fixed at a certain number. Would you be
better off building a team with a few superstars, balanced off with some truly
horrible players or a team of mostly mediocre players?
Would a lineup of five Shawn Greens (.947 OPS) and five Cesar Izturises (.547
OPS) score more than a lineup of 10 Carlos Guillens (.757 OPS)?
Thinking about how much of the Texas payroll goes to A-Rod and how bad they are
I was wondering.
As a side note, can we do away with the idea that the Rangers have been forced
to spend but a pittance on the rest of the team
because of the Alex Rodriguez
signing? Texas had an opening day payroll in excess of $105 million (3rd highest
in baseball). Taking $21 million away that A-Rod was paid this year (thanks to
http://www.bluemanc.demon.co.uk/baseball/mlbcontracts.htm for the contract
detail), that still leaves $84 million Texas spent on other players, which by
itself would have been the sixth highest payroll even if A-Rod had played for
free. The problem wasn't that they had no money because of Rodriguez, it was how
poorly they spent the other $84 million to build the rest of the roster beyond
the best player in the game.
But let's now turn to the question you asked -- whether "well
balanced" teams are more or less effective than "lopsided" teams
of the same average production. To investigate the question, I looked at teams
from 1961-2000 (strike-shortened years omitted), and looked at each team's OPS
and runs scored. To calculate the "spread" (the term I'll use
throughout this analysis to refer to the balance of the team's production)
around the average OPS, I took the square of the difference between the team's
OPS and the player's OPS, and weighted it by the number of plate appearances the
player had. I then computed the team's total spread around that.
e.g.
Assume the team has an 800 OPS, and the player has a 950 OPS in 600 PA.
a. (.950-.800) = .150
Now try it for a player with a 600 OPS in 400 PA.
a. (.600-.800) = -.200
We can see whether having the spread for each team increases the accuracy of our
expected runs beyond simple OPS. To do this, I ran a linear regression on total
runs, using OPS and spread as inputs. OPS, by far, dominated the regression, as
by itself it explains 91% of the variance in scoring among teams. However,
spread did barely clear the bar for being a significant factor in predicting
scoring as well (for the statistically inclined, a t stat of -2.37, and a
p-value of 0.017, or meaning that there's less than a 2% probability of seeing
the results if the results were purely random with respect to spread).
To some extent higher spread seems to produce lower scoring totals for the same
OPS. The magnitude of the effect at the extremes is about +/- 10 runs, so while
there's some effect, you can probably gain as much or more extra accuracy by
using a better model of run scoring that OPS alone.
For the curious...
The team with the lowest spread (most balanced lineup) was the 1976 Angels,
which featured the following roster:
b. .150^2 = .0225
c. .0225*600 = 13.5 points contributed to the team's spread
b. (-.200)^2 = .04
c. .04*400 = 16 points contributed to the team's spread
YEAR TEA LG NAME PA AVG OBP SLG OPS
---- --- -- -------------------- ---------- ------ ------ ------ ------
1976 CAL AL Remy,Jerry 540 .263 .315 .303 .618
1976 CAL AL Bochte,Bruce 532 .258 .350 .311 .661
1976 CAL AL Chalk,Dave 497 .217 .310 .253 .563
1976 CAL AL Jackson,Ron 447 .227 .291 .344 .635
1976 CAL AL Bonds,Bobby 422 .265 .341 .386 .727
1976 CAL AL Collins,Dave 405 .263 .336 .334 .670
1976 CAL AL Melton,Bill 387 .208 .302 .328 .631
1976 CAL AL Torres,Rusty 300 .205 .300 .356 .656
1976 CAL AL Guerrero,Mario 278 .284 .309 .340 .649
1976 CAL AL Etchebarren,Andy 275 .227 .305 .271 .577
1976 CAL AL Briggs,Dan 262 .214 .256 .294 .550
1976 CAL AL Stanton,Leroy 256 .190 .270 .281 .551
1976 CAL AL Solaita,Tony 249 .270 .369 .437 .807
1976 CAL AL Davis,Tommy 235 .265 .315 .329 .644
1976 CAL AL Humphrey,Terry 214 .245 .308 .311 .620
1976 CAL AL Jones,Bob 181 .211 .276 .355 .632
1976 CAL AL Lahoud,Joe 116 .177 .319 .219 .538
1976 CAL AL Ramirez,Orlando 76 .200 .263 .214 .477
1976 CAL AL Easler,Mike 56 .241 .268 .296 .564
1976 CAL AL Garrett,Adrian 53 .125 .208 .188 .395
1976 CAL AL Herrmann,Ed 53 .174 .283 .370 .653
1976 CAL AL Alvarez,Orlando 42 .167 .167 .333 .500
1976 CAL AL Miley,Mike 42 .184 .262 .237 .499
1976 CAL AL Dade,Paul 12 .111 .333 .111 .444
1976 CAL AL Lopez,Carlos 12 .000 .167 .000 .167
1976 CAL AL Nordbrook,Tim 9 .000 .111 .000 .111
1976 CAL AL Smith,Billy 8 .375 .375 .375 .750
1976 CAL AL Hampton,Ike 2 .000 .000 .000 .000
Every player with 250+ PA had an OPS within 82 points of .645, and only two
other players exceed that level dropping the PA threshold to 100 PA.
The widest spread was the 2000 Astros, narrowly edging out the 1961 Yankees,
which both had OPS spreads about 9 times higher than the Angels.
YEAR TEA LG NAME PA AVG OBP SLG OPS
---- --- -- -------------------- ---------- ------ ------ ------ ------
2000 HOU NL Bagwell,Jeff 712 .310 .428 .615 1.044
2000 HOU NL Hidalgo,Richard 635 .314 .397 .636 1.033
2000 HOU NL Alou,Moises 508 .355 .423 .623 1.047
2000 HOU NL Lugo,Julio 461 .283 .347 .431 .778
2000 HOU NL Biggio,Craig 454 .268 .392 .393 .785
2000 HOU NL Berkman,Lance 410 .297 .395 .561 .956
2000 HOU NL Spiers,Bill 405 .301 .388 .392 .779
2000 HOU NL Meluskey,Mitch 396 .300 .404 .487 .891
2000 HOU NL Bogar,Tim 342 .207 .295 .319 .614
2000 HOU NL Cedeno,Roger 302 .282 .384 .398 .782
2000 HOU NL Ward,Daryle 279 .258 .297 .538 .835
2000 HOU NL Truby,Chris 273 .260 .300 .477 .777
2000 HOU NL Caminiti,Ken 251 .303 .422 .582 1.004
2000 HOU NL Eusebio,Tony 247 .280 .364 .459 .823
2000 HOU NL Mieske,Matt 89 .173 .247 .272 .519
2000 HOU NL Barker,Glen 75 .224 .307 .373 .680
2000 HOU NL Elarton,Scott 66 .159 .197 .190 .387
2000 HOU NL Holt,Chris 65 .100 .169 .117 .286
2000 HOU NL Lima,Jose 61 .167 .180 .200 .380
2000 HOU NL Johnson,Russ 47 .178 .213 .178 .391
2000 HOU NL Chavez,Raul 46 .256 .304 .372 .676
2000 HOU NL Reynolds,Shane 42 .225 .262 .325 .587
2000 HOU NL Miller,Wade 40 .100 .100 .125 .225
2000 HOU NL Dotel,Octavio 33 .031 .061 .031 .092
2000 HOU NL McKnight,Tony 13 .000 .000 .000 .000
2000 HOU NL Powell,Brian 10 .222 .300 .333 .633
2000 HOU NL Cromer,Tripp 9 .125 .222 .125 .347
2000 HOU NL Slusarski,Joe 9 .111 .111 .111 .222
2000 HOU NL Ginter,Keith 9 .250 .333 .625 .958
2000 HOU NL Charles,Frank 7 .429 .429 .571 1.000
2000 HOU NL Ensberg,Morgan 7 .286 .286 .286 .571
2000 HOU NL Zosky,Eddie 4 .000 .000 .000 .000
2000 HOU NL Valdes,Marc 3 .000 .000 .000 .000
2000 HOU NL Bako,Paul 2 .000 .000 .000 .000
2000 HOU NL Wagner,Billy 2 .000 .000 .000 .000
2000 HOU NL Franklin,Wayne 2 .000 .000 .000 .000
2000 HOU NL Maddux,Mike 2 .000 .000 .000 .000
2000 HOU NL Cabrera,Jose 1 .000 .000 .000 .000
2000 HOU NL Gooden,Dwight 1 .000 .000 .000 .000
2000 HOU NL Gross,Kip 1 .000 .000 .000 .000
2000 HOU NL Linebrink,Scott 1 1.000 1.000 1.000 2.000
2000 HOU NL Green,Jason 1 .000 .000 .000 .000
2000 HOU NL Powell,Jay 1 .000 .000 .000 .000
2000 HOU NL Perez,Yorkis 1 .000 .000 .000 .000
2000 HOU NL Henry,Doug 1 .000 .000 .000 .000
Six of the top seven players in PA cluster into two groups almost 300 points apart, and the next three after that aren't within 100 points of each other either.
YEAR TEA LG NAME PA AVG OBP SLG OPS
---- --- -- -------------------- ---------- ------ ------ ------ ------
1961 NYA AL Richardson,Bobby 694 .261 .295 .316 .611
1961 NYA AL Maris,Roger 691 .269 .376 .620 .997
1961 NYA AL Kubek,Tony 645 .276 .307 .395 .702
1961 NYA AL Mantle,Mickey 640 .317 .452 .687 1.138
1961 NYA AL Skowron,Bill 604 .267 .320 .472 .792
1961 NYA AL Boyer,Clete 569 .224 .313 .347 .660
1961 NYA AL Howard,Elston 477 .348 .390 .549 .939
1961 NYA AL Berra,Yogi 432 .271 .333 .466 .799
1961 NYA AL Blanchard,Johnny 274 .305 .383 .613 .996
1961 NYA AL Lopez,Hector 268 .222 .295 .305 .599
1961 NYA AL Cerv,Bob 131 .271 .344 .483 .827
1961 NYA AL Ford,Whitey 108 .177 .269 .188 .456
1961 NYA AL Gardner,Billy 108 .212 .278 .293 .571
1961 NYA AL Stafford,Bill 71 .179 .225 .239 .464
1961 NYA AL Terry,Ralph 67 .227 .239 .258 .496
1961 NYA AL Sheldon,Rollie 58 .125 .155 .125 .280
1961 NYA AL Daley,Bud 46 .133 .152 .156 .308
1961 NYA AL DeMaestri,Joe 41 .146 .146 .146 .293
1961 NYA AL Coates,Jim 37 .029 .081 .029 .110
1961 NYA AL Torgeson,Earl 26 .111 .385 .111 .496
1961 NYA AL Arroyo,Luis 25 .280 .280 .360 .640
1961 NYA AL Johnson,Deron 21 .105 .190 .105 .296
1961 NYA AL Turley,Bob 21 .095 .095 .095 .190
1961 NYA AL Ditmar,Art 19 .053 .053 .053 .105
1961 NYA AL Gonder,Jesse 15 .333 .467 .417 .883
1961 NYA AL Reed,Jack 14 .154 .214 .154 .368
1961 NYA AL Hale,Bob 13 .154 .154 .385 .538
1961 NYA AL Tresh,Tom 8 .250 .250 .250 .500
1961 NYA AL Reniff,Hal 5 .000 .000 .000 .000
1961 NYA AL Clevenger,Tex 4 .250 .250 .500 .750
1961 NYA AL McDevitt,Danny 2 .000 .500 .000 .500
1961 NYA AL Thomas,Lee 2 .500 .500 .500 1.000
1961 NYA AL Downing,Al 1 .000 .000 .000 .000
The 600+ PA players alone tell the story, with OPS marks of 611, 702, 792, 997,
and 1138. And the next eight players have about as many PA below 600 OPS as
above an 800 OPS (if we give Berra one extra point).
Click here
for the exciting conclusion.
Keith Woolner is an author of Baseball Prospectus. You can contact him by
clicking here.