There are some games that get me so frustrated, I want to toss my TV out the window. The hapless fifth starter not making it through the third inning. The one-out-only reliever who walks the only batter he faces, then gets yanked. The Punch and Judy utility infielder called in to pinch-hit who taps out weakly to second. The left-handed hitter with a huge platoon split predictably doing nothing against the southpaw.
Rarely does a game go by in which I don’t see a player thrust into a situation in which he’s overwhelmingly likely to fail. We’re told that the talent pool is shallower than it used to be, that players don’t have the same breadth of skills they used to have. Some say that for every strong major league player, there are three more on the roster who barely belong there, and there’s not a thing we can do about it.
I don’t buy it. There’s a better way to build a roster. It doesn’t require a $200 million payroll or an act from above. All it takes is some common sense and a willingness to try something a little different. If done right, it can create a competitive edge for teams willing to try it, and a better brand of baseball for us all to enjoy.
If you’ve read Baseball Prospectus long enough, you’ve seen us analyze roster construction in bits and pieces. Particularly notable was Rany Jazayerli’s three-part series assessing the merits of a five-man rotation vs. a switch back to the old four-man format.
You’ve likely read plenty of other thoughts on roster construction. Joe Sheehan, has written about the advantages that platooning can bring, both in simulated and real baseball games. Joe has also weighed in on the damage a bad bench can do to a team’s record, particularly in the playoffs. In Bill James’ most recent Historical Abstract, he argued against the use of one-out specialist relievers, be they lefties or righties. Not only do they help to destroy the rhythmic pace of a game, James argues, they also do little to nothing to help a team’s fortunes, especially when compared to other ways a roster spot can be filled.
All these folks, and plenty of others, have argued these points eloquently many times before. One point has been lost in all these discussions, however: To effectively make one change, you need to make them all.
Platooning requires multiple roster spots for players who’ll play against one side or the other. That puts a strain on manager fetishes for third catchers and multi-position players who may not be good anywhere. Going to a four-man rotation requires being cautious with pitcher usage, such that relievers who can throw multiple innings become a must-have. That, in turn, cuts into the ability to carry a specialist pitcher who’ll face no more than a batter or two a game.
To reshape a major-league roster into one that’s going to produce more wins and better baseball for all involved, it’s necessary to follow each and every one of these steps:
Platooning
Every team, even the richest in talent, is going to run out of star performers at some point on the roster. Sure, you want to let Alex Rodriguez and Albert Pujols and Todd Helton play every day, regardless of who’s pitching. But what about Carl Everett? Adam Kennedy? In his prime, Brian Jordan? There are scads of major-league players who, given several years to prove their mettle against both lefties and righties, have consistently crapped out against one of the two. Yet whether because of their perceived status as everyday players or a lack of will to try something different, those players have had their weaknesses exposed through managerial inertia.
To address this weakness, GMs should pursue platoon players more aggressively. The list of right-handed hitters who can mash lefties–even if they’re otherwise limited players on the wrong side of the defensive spectrum–is huge. Rather than chase down a no-field, no-hit wonder like Roger Cedeno for a roster spot, why not give one to Eduardo Perez or Joe Vitiello? If you have a lefty bat at a corner slot who’s not getting it done against southpaws, these lefty-mashers can turn decent production at a position into very good production. Great managers ranging from Bobby Cox to Earl Weaver to Casey Stengel to George Stallings have seized on the value of platoons in the past, and there’s no reason the managers of today can’t do the same.
However, a manager needs to be careful in who he opts to platoon. Young, promising hitters such as Hank Blalock and Hee Seop Choi have struggled early in their careers against lefties. They also haven’t racked up nearly enough at-bats to sentence them to a lifetime of part-time duty. Rest an up-and-coming talent against the occasional Randy Johnson, sure, but don’t hold back his development by overprotecting him. For years, the knock on Eric Chavez has been his inability to hit lefties. This season, he’s roasting them to the tune of .333/.434/.569.
It’s worth noting that it’s easier to find a player who crushes only lefty pitching than one who destroys righties alone. Righty pitchers far outnumber lefties, so there’s a higher premium placed on righty-bashers. For every Matt Stairs, there are a number of Brian Buchanans.
OK, let’s have a look at some platoon data. Playing the role of data genius is BP’s Keith Woolner, to whom I owe my undying gratitude. (In the table headings, B = bats, T = throws)
2004 B T PA AB AVG OBP SLG B L 5326 4727 .259 .331 .405 B R 12879 11316 .266 .343 .418 L L 9085 8016 .257 .331 .404 L R 27402 23969 .276 .356 .455 R L 19121 16941 .268 .339 .440 R R 46627 41922 .263 .322 .414
2001-2003 B T PA AB AVG OBP SLG B L 20849 18632 .267 .332 .405 B R 60744 53757 .266 .338 .413 L L 39384 34944 .252 .323 .398 L R 134214 117691 .273 .352 .450 R L 81540 72208 .268 .339 .436 R R 224051 201113 .256 .317 .408
What jumps out right away is the gap between lefty hitters vs. left-handed and right-handed pitching. So far this season, lefty swingers have hit .257/.331/.404 vs. lefties, .276/.356/.455 vs. righties. From 2001 to 2003, those totals were .252/.323/.398, vs. .273/.352/.450. Put another way, this is roughly the difference between Jamey Carroll‘s career line, and Matt Lawton‘s.
Revamping the Bench
Platooning effectively does a lot of this. If three or four players are carried based largely on their platoon ability, that’s going to severely limit the possibilities for the kind of “versatile” bench Tony La Russa employs almost every year. The thesis here is that benches aren’t remotely carrying their weight on most clubs, and that no third catcher, pinch-runner or punchless utility player will match the impact that a good platoon player can bring, given 150 at-bats.
Here, we took each team dating back to 1972 and removed the batters with the most plate appearances, with the rest treated as the bench. This isn’t as precise as say, constructing exact bench profiles using play-by-play data from the last 33 years, but it can still provide a reasonable estimate of bench production. (Note: OPS_PLUS = BenchOBP/LgOBP + BenchSLG/LgSLG – 1; in other words the OPS_PLUS stat allows us to measure a bench’s contributions relative to league-wide performance)
LG LG LG LG BENCH BENCH BENCH BENCH BENCH
YEAR LG AVG OBP SLG OPS AVG OBP SLG OPS OPS_PLUS
---- -- ----- ----- ----- ------ ----- ----- ----- ------ --------
1972 AL .239 .302 .343 .645 .222 .284 .310 .594 .843
1972 NL .248 .311 .365 .676 .230 .297 .329 .625 .855
1973 AL .259 .326 .381 .707 .242 .305 .346 .651 .844
1973 NL .254 .318 .375 .693 .233 .300 .330 .630 .823
1974 AL .258 .319 .371 .691 .236 .293 .330 .622 .804
1974 NL .255 .321 .367 .688 .228 .295 .315 .610 .776
1975 AL .258 .324 .379 .703 .235 .298 .339 .637 .813
1975 NL .257 .322 .369 .691 .236 .310 .330 .639 .855
1976 AL .256 .317 .361 .677 .228 .289 .308 .596 .765
1976 NL .255 .316 .361 .677 .239 .306 .339 .645 .908
1977 AL .266 .327 .405 .732 .243 .300 .351 .652 .786
1977 NL .262 .324 .396 .721 .245 .309 .358 .667 .855
1978 AL .261 .322 .385 .707 .236 .295 .333 .628 .781
1978 NL .254 .316 .372 .688 .236 .305 .326 .631 .843
1979 AL .270 .331 .408 .739 .248 .305 .362 .667 .810
1979 NL .261 .320 .385 .705 .249 .315 .343 .658 .873
1980 AL .269 .328 .399 .727 .244 .299 .353 .652 .797
1980 NL .259 .316 .374 .691 .251 .312 .355 .667 .935
1981 AL .256 .318 .373 .690 .231 .288 .329 .617 .791
1981 NL .255 .315 .364 .679 .237 .301 .331 .632 .866
1982 AL .264 .325 .402 .727 .240 .296 .348 .645 .778
1982 NL .258 .315 .373 .688 .244 .307 .339 .647 .886
1983 AL .266 .325 .401 .726 .245 .298 .351 .648 .791
1983 NL .255 .318 .376 .694 .243 .310 .351 .661 .910
1984 AL .264 .324 .398 .722 .238 .297 .354 .651 .808
1984 NL .255 .315 .369 .685 .249 .310 .350 .660 .930
1985 AL .261 .325 .406 .730 .233 .295 .350 .645 .773
1985 NL .252 .315 .374 .689 .241 .306 .344 .650 .893
1986 AL .262 .327 .408 .735 .237 .297 .350 .648 .768
1986 NL .253 .318 .380 .698 .247 .315 .363 .678 .947
1987 AL .265 .331 .425 .756 .243 .304 .370 .674 .789
1987 NL .261 .325 .404 .728 .245 .308 .363 .672 .850
1988 AL .259 .322 .391 .712 .230 .289 .332 .621 .749
1988 NL .248 .306 .363 .669 .238 .300 .329 .629 .886
1989 AL .261 .323 .384 .707 .237 .292 .339 .631 .786
1989 NL .246 .309 .365 .674 .233 .297 .335 .632 .880
1990 AL .259 .325 .388 .712 .237 .301 .341 .642 .806
1990 NL .256 .317 .383 .700 .242 .305 .349 .655 .874
1991 AL .260 .326 .395 .721 .236 .300 .342 .642 .786
1991 NL .250 .313 .373 .686 .231 .294 .332 .626 .830
1992 AL .259 .326 .385 .711 .235 .297 .328 .625 .765
1992 NL .252 .311 .368 .679 .240 .294 .345 .638 .881
1993 AL .267 .335 .408 .742 .241 .307 .352 .659 .781
1993 NL .264 .323 .399 .722 .247 .309 .376 .685 .899
1994 AL .273 .342 .434 .776 .246 .306 .374 .680 .756
1994 NL .267 .328 .415 .743 .257 .319 .393 .712 .920
1995 AL .270 .342 .427 .769 .247 .312 .371 .683 .783
1995 NL .263 .327 .408 .735 .246 .314 .371 .685 .870
1996 AL .277 .348 .445 .793 .250 .315 .380 .694 .757
1996 NL .262 .327 .408 .735 .242 .311 .359 .670 .833
1997 AL .271 .338 .428 .766 .252 .316 .379 .695 .821
1997 NL .263 .329 .410 .740 .253 .318 .382 .700 .897
1998 AL .271 .338 .432 .769 .250 .311 .385 .696 .813
1998 NL .262 .327 .410 .737 .245 .310 .367 .677 .843
1999 AL .275 .345 .439 .784 .247 .309 .368 .678 .735
1999 NL .268 .339 .429 .768 .256 .326 .386 .713 .864
2000 AL .276 .346 .443 .790 .255 .321 .386 .707 .797
2000 NL .266 .338 .432 .770 .249 .323 .392 .715 .865
2001 AL .267 .331 .428 .760 .241 .301 .376 .676 .785
2001 NL .261 .327 .425 .753 .246 .316 .387 .703 .874
2002 AL .264 .329 .424 .753 .241 .302 .367 .669 .782
2002 NL .259 .328 .410 .738 .246 .314 .386 .700 .900
2003 AL .267 .331 .428 .759 .247 .306 .375 .681 .800
2003 NL .261 .328 .417 .745 .243 .308 .375 .683 .839
2004 AL .270 .336 .433 .769 .243 .304 .380 .683 .782
2004 NL .263 .329 .424 .753 .248 .314 .391 .705 .875
AL bench production this year has dropped to 22% below league-wide levels. Note that those numbers tend to ebb and flow over time. Also note that implementing multiple platoons may be slightly easier in the AL, which needs fewer pinch-hitters to bat for pitchers. On the other hand, the DH position presents another possible platoon spot for AL rosters, and the lack of defensive requirements for the position make DH a prime candidate for platoons.
To improve the starting lineup through platoons while either improving or at least keeping static current bench production, teams will need to put some thought into their reserves. I’d envision a fair number of platoon players on a typical team, say roughly three. A backup catcher would of course be a must. I’d then attempt to round out the bench with:
- An infielder who can play multiple positions, provide a decent glove, and get on base. Think Jose Oquendo in his better years.
-
A solid bat who’d act as a ninth starter (or 10th in the AL). This player would have enough ability to replace an injured player for a few weeks at a time if needed and not cost the team much in the way of production. Otherwise, he’d be used to give regulars occasional days off. Likely playing time would equal roughly 200-400 at-bats, depending on starters’ injuries. Michael Cuddyer currently fills a similar role with the Twins. A player like Brendan Harris could be an option for 2005.
Like breaking a young pitching prospect in through regulated bullpen use, this method would allow a young hitter to help the club and make good use of his first year of service time, without the risk of exposing him through full-time play. A good candidate here would be a B+ prospect from the organization; impact prospects such as David Wright would still be slotted into everyday jobs as rookies.
This bench structure needn’t be etched in stone, particularly since different teams will have different platoon needs. But a thoughtful, rational approach to building a bench is a must. Scrambling for Doug Glanville as a 25th-man afterthought could make the one- or two-game difference that sometimes decides a close pennant race.
Four-Man Rotation
Rany Jazayerli explains the benefits of a four-man rotation in great detail in the articles linked above. In a nutshell, reducing the number of starting pitchers a team must employ means chopping your worst starter out of the mix. With the way in which some lament the demise of pitching in today’s game, no one would dispute that it’s easier to find four good starters than it is five. A four-man staff could thus, if deployed properly, produce better starting pitching.
Here’s where the issue of pitcher workload comes into play. Managers and pitching coaches have become increasingly cautious as to how they use their pitchers. The average length of a major-league start has shrunk dramatically over the last three decades. Managers may be even more cautious with starter usage in a four-man situation, potentially further limiting starters’ innings pitched and thus increasing the burden on the bullpen.
The causes and suggested steps for pitching injury prevention are another topic best left for others, Dr. Glenn Fleisig of the American Sports Medicine Institute and BP’s Will Carroll among them. What’s worth noting here is that the elimination of the weakest link, a staff’s fifth starter, also removes the pitcher responsible for the most short outings on a staff, as those pitchers typically lack either the ability or stamina to hang with the rotation’s top four. Thus the increased bullpen burden caused by a more careful hand with the top four starters can be largely offset by taking out the pitcher most likely to get knocked out in the third inning by a flurry of rockets up the gap.
Here are our data measuring the average length per start by starters one through five in a rotation, by five-year intervals, 1972 to 2004:
YEAR GSRANK GS IP IP_PER_START 1972-1974 1 36.8 272.1 7.368 1972-1974 2 34.1 237.3 6.936 1972-1974 3 29.0 188.9 6.499 1972-1974 4 22.5 142.3 6.293 1972-1974 5 15.8 96.4 6.012 1975-1979 1 35.2 251.5 7.119 1975-1979 2 32.2 218.6 6.760 1975-1979 3 27.9 182.2 6.508 1975-1979 4 23.0 141.6 6.102 1975-1979 5 16.7 101.1 6.040 1980-1984 1 32.0 219.8 6.849 1980-1984 2 29.7 196.3 6.574 1980-1984 3 26.6 169.5 6.343 1980-1984 4 22.2 136.2 6.100 1980-1984 5 16.2 96.0 5.869 1985-1989 1 34.0 233.0 6.840 1985-1989 2 31.5 206.2 6.526 1985-1989 3 28.3 175.4 6.157 1985-1989 4 22.5 136.7 6.027 1985-1989 5 16.9 99.7 5.840 1990-1994 1 31.6 213.2 6.744 1990-1994 2 29.0 186.6 6.405 1990-1994 3 25.9 161.7 6.214 1990-1994 4 21.6 128.2 5.878 1990-1994 5 15.9 92.3 5.741 1995-1999 1 32.5 211.7 6.492 1995-1999 2 30.1 190.2 6.306 1995-1999 3 27.0 163.5 6.031 1995-1999 4 22.0 130.5 5.890 1995-1999 5 16.5 93.5 5.643 2000-2004 1 30.7 198.5 6.443 2000-2004 2 28.5 177.2 6.193 2000-2004 3 25.1 150.0 5.943 2000-2004 4 20.8 121.5 5.788 2000-2004 5 16.3 92.7 5.664
Even subtracting the fifth starter, we’re still looking at an average of just over 5 2/3 innings pitched for the fourth starter, just over 6 1/3 for the staff ace. Whether or not managers opt to be more careful with a four-man rotation, the value of a reliever who can pitch multiple, effective innings against both lefties and righties is going to increase significantly. This brings us to…
The New, Old Bullpen
Once again, we can reserve a lot of the blame for a damaging trend to Tony La Russa. When the late-’80s/early-’90s A’s found success using Dennis Eckersley as a one-inning closer, other teams started emulating the practice. That Eckersley was simply a dominating reliever, and not necessarily dominant only in save situations, was a point lost on many within the game. From there, specialization spread to LOOGYs (Left-handed One-Out only GuYs, thanks, John Sickels), often suspect pitchers who’d try to retire the Barry Bondses of the league in key situations. Managers increasingly refused to use their talented closers in any situation other than one defined in the rule book as worthy of a save. La Russa himself helped keep the underwhelming Tony Fossas employed through his 40th birthday, to name one LOOGY love-in.
The result has been bullpen usage completely out of step with the trends of the modern game, as you’ll see here (data again taken from 1972 to 2004, using five-year intervals, with average innings pitched per relief appearance included):
YEAR AVG_RP_IP 1972-1974 1.70 1975-1979 1.74 1980-1984 1.67 1985-1989 1.52 1990-1994 1.31 1995-1999 1.19 2000-2004 1.14
You can see the problem here. The length of starts has dropped rapidly, and so too has the length of relief appearances. The easy way out is to sacrifice position players–and thus offense–for 11th, 12th, and occasionally 13th pitchers.
Of course we’re not interested in the easy way out; we want to field the best team possible. As such, I’m proposing the following for pitching staffs: four starters, one swingman, five relievers. That’s it.
We need room on the roster to accommodate platoons and general offensive optimization, and one-out relievers don’t deliver enough value to let them get in the way of that. The swingman can act as a sixth reliever on many days, doing long-relief duty when a starter gets knocked out early. He can slot into the rotation when a starter needs an extra day’s rest to recover from a minor ailment. The five relievers must show the ability to go multiple innings at a time. As much as possible, they should be able to handle both lefty and righty hitters. Situational usage will be largely abolished, with no set closer, set-up man or designated inning for any one pitcher.
At BP we often talk about the supposed abundance of available arms to anchor a bullpen. That theory stems from the idea that the majority of relievers are unpredictable from one year to the next, liable to put up a standout performance one year, a washout the next. With this new pitching structure, teams will need to be more aggressive in finding effective pitchers languishing in the minors or in low-leverage jobs on other teams. They’ll need to devote more scouting and statistical research to finding struggling starters around the league best suited to key relief roles. They’ll need to continue their efforts to pick the cream of the crop from foreign leagues, especially in Asia.
And yes, occasionally they’ll want to ante up some cash for the rare reliever who can deliver more consistent results. The money saved on fifth starters and eschewing one-sided name players in favor of cheaper platoons will make this happen.
The new, optimal roster will revolve around a strong offense with as few weaknesses as possible, a rotation with one less hole, a stronger bench and a better bullpen that can control the last few innings of a game, whatever the situation. Like any strategy, it’ll be up to the shrewd GM and the perceptive manager to find the right talent to make it happen and use those players in a more effective way.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now