What Moneyball did for on-base percentage, the Rays' 2008 triumph may have done for defense—even if the book on the latter has yet to be written (although it's reportedly on its way). Of course, the importance of avoiding outs at the plate, and of accumulating them in the field, was as clear to Lane and Chadwick, respectively, as it is to Beane and Friedman; the rest of the class merely needed a little prodding to send it plunging past the tipping point. Unfortunately for those prematurely in the know, these watershed moments often mark the end of their salad days, as other prospectors make inroads on their fertile claims. The rubes are growing scarce: just ask Manny Ramirez, Adam Dunn, Bobby Abreu, and the other defensively challenged sluggers who failed to douse themselves with eau de Ibañez before seeking long-term relationships this winter.

An appreciation for on-base percentage could have yielded a competitive advantage at any point in the game's history, but until fairly recently, fielding skills remained relatively impenetrable, even to those with the inclination to evaluate them. However, as defensive metrics improve and become increasingly reliable (a process which the imminent arrival of the Hit-f/x system promises to accelerate), the leathery component of run prevention will assume an even greater significance in player evaluation and analysis (while remaining an area in which scouting insight can elucidate persistent quirks in the numbers). In order to determine just how large a slice of the run-prevention pie defense deserves to consume, we might take a quick look back at an earlier investigation.

Steven Goldman devoted a chapter of Baseball Between the Numbers to examining the relative importance of run scoring and run prevention in an attempt to answer the question, "Can a team have too much pitching?" The abridged version of his answer, as one might expect, was "no." His more nuanced conclusion, however, stated that "you can't have too much pitching—except when you don't have enough of everything else." For the purposes of his essay, "everything else" was limited to offense, but we can take some limited steps to extend the discussion to defense—in Steven's words, "the invisible hand that affects much of what we perceive as pitching"—despite the difficulty of separating the pitcher's work from that of the fielders behind him.

As part of our inquiry, we can assess the relative impacts of pitching and defense on run prevention, and, by extension, winning. A one-point swing in defensive efficiency, in either direction, constitutes a difference of roughly four hits over a full season. Assuming a typical distribution of singles, doubles, and triples (and multiplying by their respective expected-run values), those four hits amount to approximately 2.2 runs. To analyze pitching's effects without muddying the waters with defensive contributions, we'll use QuikERA, a handy tool developed (and explained here) by Nate Silver, which estimates a pitcher's ERA based solely on his defense-independent strikeout rate, walk rate, and GB/FB ratio. A .01 addition or subtraction in QERA translates to a difference of approximately 1.77 runs allowed by a team over the course of a season, or almost exactly four-fifths that of a .001 fluctuation in defensive efficiency. The following graphs display the payoff in wins as a factor of improvements in defensive efficiency and QERA, according to Bill James' simple Pythagorean expectation formula:





Since fluctuations in both defensive efficiency and QERA affect only the "runs allowed" portion of the run-expectation formula, these graphs have little comparative value, but it's still helpful to visualize the gains which even modest improvements in either category can yield. The fewer the runs which one can hope to receive from one's own batters, the more valuable the suppression of the opposition's lineup becomes.

Teams which dramatically improved their defensive efficiency or QERA in a single year provide real-world examples of the benefits awaiting aspiring run preventers. Here are the top 20 teams in each category since 1954, along with their average improvements in winning percentage:

               Differences In
Year  Team      DefEff  WPct
2008  Rays       .054   .192
1980  Athletics  .048   .182
1981  Rangers    .046   .071
1981  Tigers     .035   .031
1991  Braves     .035   .179
1988  Brewers    .033  -.025
1971  Giants     .032   .025
2008  Marlins    .031   .084
1955  Cubs       .030   .050
1997  Tigers     .029   .161
1991  Angels     .028   .006
1971  Astros     .027  -.001
1965  Pirates    .027   .062
1985  Giants     .027  -.024
1992  Brewers    .027   .056
1998  Red Sox    .026   .087
2001  Twins      .026   .099
1968  Indians    .026   .071
1982  Padres     .025   .127
2005  Indians    .025   .080
Average                 .076

               Differences In
Year  Team       QERA   WPct
1963  Indians    -.88  -.006
1997  Brewers    -.86  -.010
2001  Athletics  -.85   .065
2001  Yankees    -.84   .054
1974  Rangers    -.83   .170
1997  Tigers     -.81   .161
2001  Cubs       -.78   .142
1984  Mets       -.76   .136
1963  Cubs       -.75   .142
1994  Expos      -.74   .069
1993  Mariners   -.72   .111
2001  Astros     -.72   .130
1997  Blue Jays  -.70   .012
1994  Padres     -.70   .025
1971  White Sox  -.69   .142
1958  Braves     -.69  -.021
1997  Mariners   -.69   .028
2004  Tigers     -.68   .179
1977  Dodgers    -.67   .037
1998  Padres     -.67   .136
AVG:                    .085

The teams which recorded historically large one-year boosts in QERA enjoyed slightly higher gains than those which tallied similarly pronounced increases in defensive efficiency. However, like most "before and after" tableaux, this juxtaposition omits a portion of the story: in a sample of this size, offensive disparities may account for some, or all, of the difference between the groups.

Setting offense aside for the moment, we might ask whether it's wiser for a team to invest its fortunes in one element of run prevention, in pursuit of a stunning reversal like those above, or to distribute its finite resources in a more even-handed manner. Again, history supplies us with a possible answer. One hundred and twenty-seven teams in baseball history have finished with the same league ranks in defensive efficiency and QERA, posting a collective winning percentage of .504. The 127 teams with the greatest disparity between their league ranks in defensive efficiency and QERA have registered a slightly higher .507 winning percentage. Of course, these numbers are misleading, because the list of the 127 most balanced teams also includes teams that were balanced in a bad way—near the bottom of the league in both categories. All of the 127 least-balanced teams, in contrast, were at or near the top of their leagues in one of the categories.

The worst average rank ((defensive efficiency rank + QERA rank)/2) of any of these least-balanced teams was 10.5 (the 2006 Washington Nationals, who ranked 16th in the NL in QERA and seventh in the NL in defensive efficiency). Only three had an average rank of 10th or worse. In contrast, two of the most-balanced teams finished with average ranks of 16, and twenty-three of them finished with average ranks of 11 or worse. If we eliminate these twenty-three well-balanced failures, we're left with 104 well-balanced teams with an average rank less than or equal to 10th, and a collective winning percentage of .518. The 104 least-balanced teams, on the other hand, amassed a collective winning percentage of .506. Again, offense plays a role here, but these findings suggest that it pays, albeit slightly, to devote equal care to both aspects of run prevention, rather than placing one's fate entirely in either the hands that throw the ball, or the hands that catch it.

This conclusion is born out by a survey of some of the last decade's most successful squads. Most of the 22 pennant-winners since the last round of expansion in 1998 have ridden balanced approaches to run prevention deep into October:

AL         Differences In Rank
Year  Team    DefEff   QERA
2008  Rays       1      6
2007  Red Sox    1      3
2006  Tigers     1      7
2005  White Sox  2      4
2004  Red Sox    5      1
2003  Yankees   13      1
2002  Angels     1      7
2001  Yankees    9      1
2000  Yankees    4      6
1999  Yankees    2      2
1998  Yankees    1      1
Average          4      4

NL         Differences In Rank
Year  Team     DefEff QERA
2008  Phillies   6      8
2007  Rockies    4     10
2006  Cardinals  6     10
2005  Astros     1      1
2004  Cardinals  2      3
2003  Marlins    9      5
2002  Giants     4     12
2001  D'backs    2      1
2000  Mets       4      4
1999  Braves     6      2
1998  Padres     7      2
Average          5      5

Since 1954, 111 teams (counting ties) have topped their leagues in defensive efficiency, racking up a collective .563 winning percentage along the way. As it happens, an identical number of squads have paced their circuits in QERA during the same period, finishing at .558. The moral of this story? Preventing opposition scoring is a winning strategy, no matter what route one takes to pursue it.

Although Steven was referring to offense when he wrote that "in the absence of available pitching, making other changes will work just as well," the same statement holds true for defense. Still, while the evidence suggests that improving team defense can be an effective alternative to hoarding hurlers, whether a team should elect to invest its available funds in pitching or defense depends largely on the state of the talent market, and the unique circumstances of the club in question. The graphs above evince the benefits of improving defensive efficiency and QERA, but they can't tell us which of the two is easier to fortify.

A vote for pitching usually funnels aid through the proper channels without springing a leak elsewhere on the roster, while the addition of a defensive whiz often exacts a toll on the offensive side of the ball. However, even in these more enlightened times, a mediocre moundsman may command a higher price than a superlative defender, whose contributions might be more easily overlooked or underestimated. The window may be closing, but as long as this state of affairs persists, teams with average run production, pitching, and defense would be wise to channel their resources along the path of least resistance. Moreover, as Keith Woolner and Dayn Perry observed in a separate BBTN chapter, the pitcher that you get only rarely becomes the pitcher you paid for, "given the precision demanded by pitching, the ability of opposing hitters to crush weak pitches, and the risk of injury inherent in the physical act of throwing a baseball." The greater consistency offered by defensive acquisitions represents a compelling argument in favor of making them the foundation of a winter's retooling.

Furthermore, good defense makes pitchers look better, a phenomenon which has implications not only for fantasy baseball, but for the genuine article. Pitchers whose traditional stats are inflated by the human safety nets surrounding them may bring a greater return in trade than they deserve, especially from those who fail to examine their underlying performance closely. It's not difficult to imagine a scenario in which a superior defensive team could consistently mint superficially league-average starters, only to convert them into more valuable commodities by convincing gullible takers in more hostile environs to roll out the welcome mat; witness Greg Smith's inclusion as a nominal piece of the return in the deal which sent Matt Holliday from Colorado (14th in the NL in defensive efficiency) to Oakland (third in the AL).

It's also worth noting that while the presence of a superior pitching staff bestows little or no advantage on its supporting fielders (all anecdotal reports of efficient workers "keeping fielders on their toes" aside), the construction of a capable fielding unit can confer tangible benefits upon a team's valuable arms beyond those merely cosmetic perquisites mentioned in the preceding paragraph. As noted in the Rays team essay in BP2K8, a strong defense can hasten and enrich the development of a young pitching staff, "with effects on the order of 20 to 30 points of long-term ERA." In addition, adept glove work enables pitchers to work more efficiently, reducing the incidence of wasted pitches over the course of a season, and indirectly reducing the workload shouldered by a team's complement of relievers.

Last year's Rays vaulted to the top of the defensive efficiency differential leaderboard—and, despite posting a QERA nearly identical to their 2007 mark, the AL East—primarily by re-shuffling assets (B.J. Upton, Akinori Iwamura, and Evan Longoria) which were already under club control; the team's primary outside defensive acquisition, Jason Bartlett, failed to perform as expected in the field. Other organizations have taken note: the new Mariners regime's emphasis on outfield defense in the face of a largely inflexible starting rotation, smacks of a lesson well-learned, and the Yankees' fixation on high-strikeout starters, who minimize the damage done by porous defensive units, represents something more than a visceral reaction to an overdose of Sidney Ponson (landing Mark Teixeira and allowing Jason Giambi and Bobby Abreu to do what they do best—walk—should compound the benefits). The allure of the Rays' resounding success will continue to encourage copycats, especially if the defending AL champions manage to power through the plexiglass this season.

Next off-season's weak free-agent class might provide further inducement to pursue the stealth approach, as the teams which miss out on the few choice cuts available scramble for the most savory gristle. Since fielding contributions remain the most difficult to quantify, as well as the least easily observed with the naked eye, defensive specialists may find themselves riding home in a number of doggy bags—and, perhaps to the surprise of their new owners, feeding the whole family.

The Tampa Bay approach isn't suited to every roster, and most teams that adopt it will find the path from ignominy to respectability far smoother than the one leading thence to contention. Still, good gloves tend to sprout more readily on the farm than do MLB-ready arms, making it a sound strategy in the aggregate. Which is not to suggest that the Orioles may have stolen away with the annual AL East arms race by eschewing the likes of Sabathia, Burnett, and Smoltz in favor of Cesar Izturis—the big-ticket items usually do reserve admittance to the most desirable show. Nonetheless, as last year's high-QERA, low-defensive efficiency Yankees, Diamondbacks, and Reds discovered, procuring pitching without addressing underlying defensive woes is like using an HDTV to display SD sources: you simply won't end up with the pretty picture you thought you were paying for.

Thanks to Bil Burke and others for research assistance.

Ben Lindbergh is an intern with Baseball Prospectus and a student at Georgetown University. You can contact Ben by clicking here.

You need to be logged in to comment. Login or Subscribe
Welcome, Ben!
I throw on some eau de Ibañez every morning for good measure
Very interesting article, but (and this extends throughout BP) the graphs tend to be difficult to read - please make them larger in the display for greater clarity. Thanks - and welcome
\"Nonetheless, as last year\'s high-QERA, low-defensive efficiency Yankees, Diamondbacks, and Reds discovered, procuring pitching without addressing underlying defensive woes is like using an HDTV to display SD sources: you simply won\'t end up with the pretty picture you thought you were paying for.\" Love this quote... but what pitching did the 08 Yankees \"procure?\" They went with incumbent vets and threw some rookies into the rotation. Are you guilty of looking forward to the 09 Yankees? They have spent a ton of money on pitchers and other than Teixeira have done nothing to improve a pretty awful defensive team(5th worst in MLB, 3rd worst in AL in 2008). In that case, you make a great and much overlooked point.
The 2008 Yanks developed a darn good no-name bullpen (aside from Mo), and then acquired Marte. The rotation was Mussina, Pettitte and a MASH unit, so you\'re right there. As for their upgrades, Teixiera is an upgrade at 1B and their RF defense (whether it\'s Nady, Swisher, or various combinations thereof) figures to be *much* better (check out Abreu\'s defensive stats - they\'re shockingly bad). They will probably decline at SS (Jeter was actually decent last year)... 2B is a bit of a ? mark, as is CF (depends on who plays - Melky or Gardner, or someone else... if it\'s Swisher, the D takes a hit). Overall, I expect mild defensive improvement. Combine that with better pitching and you should see a serious RA improvement.
You\'re right--although the Yankees\' QERA improved from 12th in the AL in 2007 to 3rd in the AL last year, most of those responsible for the improvement weren\'t recently hired help. I meant it in a more general sense, ie., if you have good pitching, you must have procured it at some point, one way or another. I am guilty as charged when it comes to looking forward, however--the Sabathia/Burnett signings partly inspired the article. As I mentioned above, though (and as Arrian notes below), between the arrival of Teixeira, and a couple cases of addition by subtraction, it\'s reasonable to expect a moderate defensive improvement. Judging by his recent defense-related comments, Brian Cashman isn\'t expecting miracles.
Meant to post that as a response to alskor\'s comment. You get the idea.
Darn talented intern there. Good job!
Ben\'s got skills.
Minor correction: QERA is based on k%, bb% and gb%. It treats all batted balls that are not ground balls (i.e. line drives and fly balls) the same. The formula is (2.69-3.4*k%+3.88*bb%-0.66*gb%)^2 (In the article you link, Nate himself misdescribes it.) Thanks for an interesting article. I feel like BP has fallen behind the pack in understanding fielding defense. I hope that this article is the beginning of a change in that respect. Can you guys please develop a metric other than FRAA for player evaluation? It\'s fine for looking at players in 1978, where we don\'t have the data necessary to do a better job, but it doesn\'t stack up agaist UZR or +/-. That top chart is interesting. Am I right in thinking that it shows that the marginal value of improving defense is greatest for teams that already have a good defense? That means, contrary to what you might think, the good fielding is most important for teams with good pitching.
I don\'t think that\'s quite right Phil (re: your last paragraph). The top chart is saying that the fewer runs you score, the more important it is (the more marginal wins you\'ll get) to prevent runs through an improved defense. This makes intuitive sense to me. If you\'re not a good offensive ballclub, you\'ll have difficulty winning UNLESS your defense is quite strong. So, the gains associated with improving your defense as a bad offensive ballclub are more pronounced. What I have difficulty understanding is why there shouldn\'t be a similar sort of effect for the bottom graph. Why wouldn\'t improvements in QERA have distinctly different marginal win values at all QERA levels (as in the top graph)? Help, Ben! Very interesting read, btw.
I\'m glad I asked. I was reading those graphs all wrong (as a function of runs allowed...) Don\'t even ask what I thought I was reading, because I doubt it makes any sense.
Fantastic article, and very well written. Great stuff.
The 2008 Rays are a copy of the 91 Braves.
I\'m admittedly new to all these defensive statistics, and I certainly agree that good defense improves pitching, but isn\'t it possible that good pitching improves defense in the same way? Couldn\'t the Rays\' acquisition of Matt Garza, and the revamped bullpen played a significant role in the team\'s defensive improvement? Looking at BP\'s numbers for the Rays\' individual players- Navarro improved from -5 in 2007 to +16 in 2008, Pena -5 to +5, Crawford -8 to +14, BJ Upton +2 to +6. I find it hard to believe that all of the Rays\' regulars got better on defense on their own in 2008. Was Dioner Navarro really 21 runs in 2008 than 2007? Good pitchers generate weaker contact than bad pitchers, thus good pitcher\'s fielders appear to be better than they really are. The Rays were better on defense in 2008 than 2009 with Longoria and Bartlett, but I think your article understates the fact that their pitchers were also better. I would be interested to see how many teams last year were given a high defensive rating, but had a high ERA.
Interesting idea Birdfan... it makes me wonder if the type of pitching also affects the quality of defense. Do pitchers that surpress BABIP make defenses \"look\" better? What about pitchers that surpress liner rate or induce more popups? Does Dioner Navarro look better because the pitching staff induced more pop-ups behind the plate, or reduced the amount of runners on base to such an extent that opposing managers had to use more 1-run strategies like bunting and stealing?
Now you guys are on to something - unless the fielders can\'t deal with popups and weak grounders, that is. Great job, Ben!
While there has been progress, defense is still undervalued. See Hudson, Orlando.
I disagree with the comment that \'balanced approaches have led teams to October\'. Looking at the AL data, every team but 1 is either 1st or 2nd in one category or the other, including a 1/13 and two 1/7 splits. don\'t let the data get in the way of the story!
Yeah, I saw this discussion of balanced vs. unbalanced as a poor read of the data. A team that is 1st/ 4th reads as balanced and that\'s more likely to be successful than a team that is unbalanced, where they are in the bottom half in one of the categories. Well, sure 1/4 is better than 1/12. Would probably be more accurate to say that teams that are better at run prevention are more likely to play in should at least compare that to the splits between QERA and DER. Equally fishy is the discussion of whether to include \"balanced\" teams that were really bad. There\'s no meaningful reason to exclude them, except to polish up the conclusion you want to make. But the article\'s discussion on the marginal value of runs prevented is solid. I\'m not sure I understand why the curves are so different at different RS levels...maybe we should be talking about runs prevented as a % of runs scored. Would that resolve the differences among the curves? Thanks for a thought-provoking piece. Tim
I think the comment makes more sense if you look at the teams that won the World Series that particular year... in 2008 the Rays were ranked 1 in DefEff and 6 in QERA, but lost to a team ranked 6 in DefEff and 8 in QERA. The 2003 Yankees were 1 in QERA and 13 in DefEff, but lost to a team that was 5 and 9, respectively. Similar events take place in \'06 with the Tigers and Cards. If you look at the range of ranks, the team with closer ranks seems to win. Of course this is not always the case, as we see in \'05 with the White Sox and Astros...
Great article!
ARTICLE IDEA: First of all, great piece, Ben....thank you. As you referenced in your comment about how a great defense makes pitchers look better, how about a follow up to Marc\'s article (the one linked in your comment) that takes a serious look at the top five defensive teams for 2009? Naturally this makes sense for general interest reading, but the fantasy implications are huge.
\"Pitchers whose traditional stats are inflated by the human safety nets surrounding them may bring a greater return in trade than they deserve, especially from those who fail to examine their underlying performance closely.\" Defense as the new Coors Field.