What, exactly, has happened in the eight years that have passed since the publication of Moneyball? Billy Beane, Thor to Bill James’ Odin, the hero of the book, rejected the Red Sox, only to see them co-opt his front-office philosophies and turn them into two world championships. The Moneyball A’s retreated to midtable obscurity after some of Beane’s ideas flopped. Purists and casual fans soured on the entire idea of sabermetrics, classifying it as a pseudo-science created by a cabal of nerds who had declared jihad on the National Pastime. And yet, despite some high-profile resistance and some even higher-profile flops1 and the popularizing of the book’s lexicon, Michael Lewis’ book was, and largely remains, misunderstood. Moneyball was about statistics and on-base percentage in the same way that the ’77 punk revolution was about looking like Richard Hell; what was relevant and counterculture then is mainstream and comfortable now. Babies get mohawks now, and they come out of the womb knowing that Jeff Francoeur sucks. So what’s a small-market team to do now? Where are the market inefficiencies for them to exploit in 2011? What’s the new Moneyball?
Beane laughed at the idea of taking high school pitchers Zack Greinke and Cole Hamels after cheap, high-OBP college guys such as Jeremy Brown and Brant “Titties” Colamarino. Good news: It was still a good draft. His seven first-rounders in that 2002 draft have produced 26 WAR so far; the guys taken in the same spots in the 2001 and 2003 drafts produced an average of 17.4 WAR. Thanks, Nick Swisher!
First, we’ve got to figure out who’s getting what for cheap. The late researcher Doug Pappas developed Marginal Payroll/Marginal Wins (MP/MW) for Baseball Prospectus in 2003, which measures a team’s ability to turn spending into wins. It assumes that a team hoping to spend the league minimum on a 28-man roster (25 players and three injury replacements) could win 30 percent of their games, producing a 49-113 record in a 162-game season. Since both leagues adopted the 162-game season in 1962, only the 1962 Mets and 2003 Tigers have failed to win 30 percent of their games. The formula2 for Marginal Payroll/Marginal Wins measures how much a team spent beyond the bare minimum in team salary (about $11.2 million) to get beyond that .300 winning percentage floor.
That formula is: (Team Payroll – (28 x League Minimum Salary) / ((Team Winning Percentage – .3) x 162). The minimum salary for a player in Major League Baseball last season was $400,000.
The following table shows the marginal cost of a marginal win for each team during the 2010 season.
Not many teams could pull off what the San Diego Padres did last season. With just $26.5 million sitting between their team payroll and the league’s payroll floor, the Padres came within one win of the playoffs. That means San Diego and general manager Jed Hoyer generated 41 “marginal” wins from about 80 percent of Alex Rodriguez’s yearly compensation. Although the Padres weren’t able to produce a run to the playoffs for pennies on the dollar, three of the other teams in the top six in MP/MW were able to, including the American League champion Texas Rangers. While teams like the Yankees and Phillies were undoubtedly happy their hefty expenditures resulted in October baseball, 2010 made it very clear that teams can succeed in spite of a small budget. And by identifying what these thrifty teams have in common, we can figure out which aspects of team performance currently cost less than they should.
As it turns out, these penurious franchises shop in some of the same aisles. While competing with the big markets wasn’t as simple as the Moneyball model of merely applying objective analysis, the teams at the top of our chart overtook the rich kids in 2010 primarily by exploiting two undervalued markets. And while those richer teams will catch up, the bargain shoppers of 2011 have mostly continued to stock up on the same items that produced savings last season.
It was clear even at the time Moneyball was written that fielding would become a battleground for teams looking for an edge, but the sort of data that would be required to do robust analysis on a player’s fielding ability really didn’t exist. Traditional fielding statistics, such as errors, didn’t do the actual act of fielding justice. Fielders with great range rack up errors on plays that mediocre fielders would never have a prayer of getting to. That makes no sense.
In 2011, that’s no longer the case. Systems like Ultimate Zone Rating (UZR)3 and plus/minus4 usually disagree on the exact level of a player’s glove upon team performance, but they represent a dramatic increase in the level of knowledge and accuracy in fielding analysis that’s available to the public. Both UZR and plus/minus attempt to determine how many plays a fielder should have made versus a league-average player at the same position.
UZR takes every ball in play and measures how likely it was to be turned into an out, using seven years of batted-ball data, incorporating data such as the speed of the ball, the game situation, and the nature of the particular stadium. (A fly ball to left field in Fenway is pretty clearly different from a fly ball to left in Wrigley.) UZR then credits the fielder(s) for a successful out, or docks them for not making a play and converts that figure into runs. For an extremely detailed explanation of Ultimate Zone Rating, you can read creator Mitchel Lichtman’s explanation. at Fangraphs.
The plus/minus fielding system developed by Baseball Info Solutions is a simpler, more qualitative system. It uses video to define certain characteristics of a ball in play (such as whether a ball travels slow, medium, or fast) and link a ball’s location on the real field to that same ball’s location on a virtual field. That ball in play is then compared to the most similar balls in play traveling to the most similar spots on the field to determine the frequency of that ball being converted into an out, much like UZR. After gathering all that data over the course of a full season, plus/minus produces a number that measures the number of plays a player made (or didn’t make) versus a league-average player given the same opportunities. So a shortstop with a plus/minus of minus-3 made three fewer plays than a league-average shortstop would have, given the same 2,500 plays in the same parks that the shortstop had the possibility of making.
The “public” part of that statement is particularly relevant because so much of the work that’s been done on fielding since Moneyball has been conducted inside the inscrutable walls of major league organizations, which have both the funding and motivation to sink significant sums of money into the sort of video analysis and quantitative research required to produce advanced fielding statistics. The Red Sox have their own system, likely built in part by director of baseball information services Tom Tippett, who was previously the creator and developer of the Diamond Mind baseball simulator. Unlike hitting and pitching, though, estimations of fielding performance are far more arbitrary; the model used by the Red Sox may very well be totally different from the one used by the Rays, Athletics, or Padres. That creates some interesting opportunities in the marketplace. Everyone knows Tony Gwynn Jr. can’t hit worth a lick, but one team’s defensive system might very well measure him as the most valuable defensive center fielder in baseball, while another might see him as a below-average fielder. Gwynn’s range might make him worth a few million bucks to Team A and worthless to Team B. If Team A is right about Gwynn’s ability and it’s bidding against Team B, it might be able to sign Gwynn for the league minimum and pick up a reasonably valuable asset; if it’s wrong, it might spend a million dollars on a player who brings nothing to the table.
The Artists Formerly Known as the Devil Rays
The Rays represent the first and best example of how quickly a team can turn around its record through improving its defense. In 2007, the then-Devil Rays turned just 66.9 percent of balls in play into outs, a statistic commonly referred to as Defensive Efficiency.5 That was the worst rate in the league, and the organization set out to fix it with the help of former Baseball Prospectus writer James Click, who developed a more complex version of Defensive Efficiency that accounted for the effects of a team’s ballpark upon its ability to turn balls in play into outs. (Disclaimer: I conducted some research for the Rays under Click as a consultant last season.)
The formula for Defensive Efficiency is 1 – ((H + ROE – HR) / (PA – BB – SO – HBP – HR)), where “ROE” stands for the number of times a player reached via an error.
In 2008, the newly christened Rays went from the bottom of the Defensive Efficiency charts to the very top, turning 72.3 percent of balls in play into outs. The plus/minus system used by John Dewan in The Fielding Bible suggested the change in the Rays’ fielding was responsible for an 85-run swing, an improvement worth about eight to nine wins.
The Mariners were the next team to publicly jump upon the fielding bandwagon. After the 2008 season, new general manager Jack Zduriencik dealt closer J.J. Putz to the Mets in a three-team trade that netted him outfielders Endy Chavez and Franklin Gutierrez. Although the Mariners had an average defense before making the trade, the move gave them the best outfield defense in baseball. After adding shortstop Jack Wilson, the Mariners were able to improve their defense by 94 runs, which allowed them to win 24 more games than they had in 2008 despite scoring 31 fewer runs. They went from 25th in Defensive Efficiency to first, converting 72.8 percent of balls in play into outs. They were third in Defensive Efficiency in 2010, but an offense that had already finished last in the American League in 2009 scored 127 fewer runs in 2010, producing a 510-run performance that was more than 100 runs worse than any other American League team. That’s how you win 61 games with a great defense and a Cy Young winner, and how you end up paying more than $6 million per marginal win.
Other teams were able to turn their superb fielding into successful seasons, though. Of those six leading teams in Marginal Payroll/Marginal Wins, five ranked in the top 10 in Defensive Efficiency, including each of the three playoff teams. The seventh-ranked Padres scored just 17 more runs in 2010 than they had in 2009, but they won 15 more games. The Rangers (fourth) and Rays (second, winning 12 more games while scoring one fewer run) were joined in the DE top five by the A’s, who actually led the league in Defensive Efficiency after finishing 24th in 2009.
As for those Moneyball A’s, they improved by six wins despite scoring nearly 100 fewer runs. Oakland had five starting pitchers who threw at least 90 innings for the A’s in both 2009 and 2010; each of them saw their ERA improve in 2010, and the group combined to drop their cumulative ERA from 4.59 to 3.33. While some of the ERA improvement has to do with the natural development curve of a young pitching staff, it’s pretty clear Oakland’s hurlers were helped by the superb gloves behind them. While the Oakland fielding has been about average so far this season, the Rays and Rangers rank first and second, respectively. The surprising Indians ran to the top of the American League with a defense that ranked second in the AL in Defensive Efficiency as recently as May 23, but a 4-14 slump in their ensuing 18 games was joined by a drastic decline in their defensive performance. Cleveland currently ranks 23rd in Defensive Efficiency, 11th in the American League.
While sabermetricians continue to work on developing the first great fielding metric, they’re faced with a second problem. Building a model that reflects a player’s defensive impact is difficult enough, but there’s also the question of how accurately a player’s defensive performance mirrors his true talent.
At this point, both analysts and fans alike still struggle to understand how fielding performance changes from year-to-year. We think of players such as Ozzie Smith and Rey Ordonez as guys who were consistently excellent fielders, but just like hitting or pitching, a player’s fielding performance can oscillate wildly from year to year. According to the Ultimate Zone Rating (UZR) metric, Coco Crisp saved 25 runs versus the average center fielder in 2007 for the Red Sox. In 2006 and 2008, with different outfielders around him, he combined to be nearly four runs worse than league average. Was his big season in 2007 a fluke, or a reflection of the impact from Trot Nixon to J.D. Drew? And in 2008, did Jacoby Ellsbury take balls away from Crisp in left field that Manny Ramirez never would have gotten to, artificially pushing Crisp’s statistics down? These are all questions that the next generation of fielding metrics will need to account for, but it isn’t just an issue with statistics, either. That huge season from Gutierrez in 2009 — the one that got him the “Death to Flying Things” nickname — led to unabashed praise from statheads and scouts alike, but he wasn’t given a Gold Glove. Instead, the award went to longtime Gold Glove recipient Torii Hunter, who was moved to right field in 2010. That was enough to get Gutierrez the Gold Glove he deserved the previous year.
And while the best work on defense might still be behind closed doors, there might be a dramatic increase in the amount of data available to the public very soon. The Sportvision company uses cameras in every major league ballpark to produce PITCHf/x, which accurately measures the speed and break of each pitch and releases that data to both Major League Baseball and the public. That has allowed folks to conduct worthwhile studies such as this Hardball Times piece on how changes in fastball velocity affect pitching performance, valuable pitching research that couldn’t have been conducted years ago.
Sportvision’s new tool is FIELDf/x, which should open up the fielding toolbox in the same way that PITCHf/x created a deeper understanding of pitching. Instead of simply measuring a player’s output (e.g., where he caught the ball), FIELDf/x should allow for a more accurate measure of a player’s fielding process. Who gets the best jumps on fly balls? Which team does the best job of defensive positioning, getting its players in the right place to make plays before a ball’s even hit? What’s the optimal shift to employ against a particular slugger? FIELDf/x won’t provide all the answers, but if the data compiled by the project gets released to the public, it will bring researchers much closer to them.
If an MLB team really wants to get effective pitching for cheap without waiting five years to rebuild its organization, the only real way to do it is by going through the bullpen. In 2010, three of the top six teams in Marginal Payroll/Marginal Wins also produced the three best bullpens in baseball. In order, the Padres, Rays, and Rangers got more out of their relievers than anybody else. Furthermore, the A’s and Reds had top-five bullpens in 2009, only to slip in 2010. (The Toronto Blue Jays had neither great fielding nor a great bullpen in 2009 or 2010 and don’t fit our analysis in anyway, which goes to show one thing: Jose Bautista contains multitudes.) And in 2011, the Indians’ resurgence has been coupled with the arrival of an elite bullpen. The Chris Perez-led unit has been the league’s
fourth-best group of relievers through Monday’s games.
To figure out who got the most out of their relievers, we’ll use the Win Probability Added (WPA) metric, which does the best job of accounting for the wide variety of game situations that a reliever can enter into. WPA is used to measure a player’s impact on Win Efficiency; here, we’re considering how a team’s win efficiency changed during a reliever’s stint in a particular game. As an example, take Darren Oliver’s performance in the Rangers’ 5-4 loss to the Athletics on May 2. Oliver entered the game in the bottom of the 10th inning, with nobody on and nobody out, in a 4-4 game. Using the historical likelihood of the road team winning the game in exactly this situation, we can estimate that the Rangers had a 36 percent chance of winning at the moment Oliver entered the game. Unfortunately for Oliver, Hideki Matsui hit his first pitch to right for a game-winning home run. Obviously, the Rangers now had a zero percent chance of winning the game. Oliver had lowered his team’s chances of winning by 36 percent, which WPA expresses as a decimal: -0.36. Calculate that for every relief appearance by every relief pitcher on each team over the course of a season, and we can figure out the impact each bullpen had on their team’s chances of winning.
How does it apply to 2010 cheapskates?
There are different ways to build an effective bullpen on the dole, but one thing is patently obvious: Cheap teams clearly can’t shell out the big bucks needed to acquire a proven veteran closer, so they have to be crafty. Take the Padres, whose bullpen produced nearly 8.5 wins last season, more than two wins better than the production of the second-placed Rays. The top eight relievers in the San Diego ‘pen pitched 492 2/3 innings with a 2.39 ERA. That’s roughly the combined level of production that the two Cy Young winners put together: Roy Halladay and Felix Hernandez threw 500 1/3 innings and had a 2.36 ERA, yielding 8.7 WPA. The Padres can’t trade for Hernandez or afford Halladay, but that bullpen came close to matching their numbers for just $7.5 million. That’s exactly what the Brewers paid former Padres closer Trevor Hoffman in 2010, and Hoffman gave them 47 1/3 innings with an ERA approaching 6. (It’s not cherry-picking Hoffman, either; the other pitchers who made exactly that much in 2010 were Bobby Jenks, Jason Marquis, and Brad Penny).
Over the past several seasons, though, nobody’s churned their bullpen more than Tampa Bay. The final edition of the Devil Rays in 2007 had a particularly gruesome ‘pen that produced minus-7.8 WPA. Naturally, they took an ax to that unit and revamped things that offseason, signing Troy Percival to close and converting former starter J.P. Howell to relief. Journeyman Grant Balfour finally got a consistent spot in the majors and became a strikeout machine, and they supplemented the whole thing with former first-overall pick David Price, who came up at the end of the season and immediately contributed as a rookie. Throw in the improved fielding mentioned earlier, and the Rays went all the way to the top of the WPA charts in 2008, with their relievers producing 9.3 wins. Incredibly, that’s a swing of nearly 17 wins from their bullpen. That’s how you produce drastic turnarounds.6
For a comprehensive story on how the Rays have been able to compete despite possessing one of the league’s lowest payrolls, Jonah Keri’s 2011 book The Extra 2% is highly recommended.
After those same relievers fell to league-average in 2009, the Rays rebuilt again for 2010. The unit that finished second in WPA that year was built around three pitchers. Balfour had another productive season, but the other two relievers were new. Joaquin Benoit had missed the entire 2009 season with a torn rotator cuff. The Rays picked him up for the league minimum before the 2010 season. Finally healthy after years of injury issues, Benoit produced 75 strikeouts and a 1.34 ERA in 60 1/3 innings. Meanwhile, the Rays saw an opportunity to acquire an elite player on a short-term contract when Rafael Soriano unexpectedly accepted salary arbitration from the Braves, who had already signed relievers Billy Wagner and Takashi Saito to replace him. With no leverage, the Braves dealt Soriano to the Rays for the middling Jesse Chavez. Soriano, meanwhile, got his one-year deal for $7.25 million and promptly produced four wins all by himself, saving 45 games while putting up a 1.73 ERA. The Rays have no interest in committing to a high-risk, high-reward player like Soriano for the long term, so they let him go to the Yankees after the season and pocketed two first-round picks (worth about $8 million in surplus value beyond what the Rays will pay in signing bonuses) when Soriano, ironically, declined arbitration. They’ve fallen to 22nd in the league this season in relief WPA, but consider that they lost those three guys and just about every other part of their bullpen. Considering their pen is currently helmed by castoffs such as Kyle Farnsworth and Juan Cruz, merely staying competitive is impressive enough.
Teams are able to identify and acquire relievers at bargain-basement prices by recognizing their year-to-year performance varies a lot more than their actual talent levels; in other words, just like a well-run fantasy team, intelligent professional outfits buy low and sell high. Since relievers throw only 60 innings or so a year, there’s just not all that much time for good luck or bad luck to balance out over the course of a season in the way that it does for a starting pitcher.7 Take CC Sabathia’s 2008, for example. In his first 10 starts, Sabathia went 3-6 with a 5.14 ERA in 61 1/3 innings. That’s a full season for a reliever, and it would have been a poor one. Of course, Sabathia still had 191 more innings left in him that season, and he went 14-4 with a 1.91 ERA. A lot can go wrong in 61 innings, even if you’re a good pitcher.
Sixteen of the top 35 starters by WPA in 2009 remained in the top 35 in 2010. Just nine of the top 35 relievers, using the same statistic in 2009, stuck around in the following season.
The best buy-low candidates are pitchers who traded in bad luck for a season or more, or players with great peripheral statistics (a high strikeout rate, and low walk and home run rates) that don’t match their ERAs. Usually, they owe a healthy percentage of their inflated ERA to sheer luck, thanks to a high batting average on balls in play (BABIP). As Voros McCracken noted in his now-famous 2001 piece on pitching, hurlers really don’t have much control over their BABIP.8
Since McCracken’s research was published in 2001, researchers have found that the type of ball in play does matter, particularly a player’s line-drive rate (and particularly for hitters as opposed to pitchers). But McCracken’s general theory still holds.
If you’re looking for the perfect example of an elite reliever who was plucked off of a scrap heap he didn’t belong on, it’s Padres closer Heath Bell. The Padres acquired Bell from the Mets after the 2006 season in a four-player deal that brought the Metropolitans reliever Jon Adkins and utility outfielder Ben Johnson. Adkins had pitched in 55 games for the Padres with a 3.98 ERA, but he had middling peripherals, with below-average strikeout and walk rates that did not bode well for future success. He pitched a total of one inning for the Mets. Johnson played in nine games. It wasn’t exactly Scott Kazmir for Victor Zambrano, but it was close.
The Mets were willing to give up Bell because he looked to be a Quad-A player, a reliever with decent minor league numbers who didn’t have the stuff to make it at the major league level. In three years with the Mets, he’d put up a 4.92 ERA in 108 innings. Not pretty. The Padres looked past that, though, and saw a reliever who had put up great peripherals at both the minor and major league levels, but had experienced a rash of awful luck in 2005 and 2006. Although Bell had the strikeout, walk, and home run rates of an above-average pitcher, hitters had a .384 BABIP against him in the two years before his trade, the highest rate in baseball for a pitcher with 75 innings pitched or more.
What happened next, of course, was sweet justice. Bell’s peripheral stats remained great, but his performance caught up to his actual talent. He put up two great seasons in relief with a BABIP that was actually below the league average, and benefited by moving from a league-average park to one of the league’s best stadiums for pitchers, Petco Park, which depressed scoring by about 12 percent. The unluckiest pitcher in baseball finally got lucky. In 2009, Bell took over the closer’s role from the departed Hoffman and became one of baseball’s best stoppers. Meanwhile, Jon Adkins is currently a scout for the Red Sox, and Ben Johnson is out of baseball. The lesson remains the same: Don’t become a Mets fan.
Since Bell is a free agent after the season and the Padres don’t appear to be contending, he’s likely to be traded or leave the team this offseason. The Padres could choose to replace him with setup man Mike Adams, who has similar peripherals to Bell since returning from shoulder surgery in 2009, but with a microscopic 1.35 ERA that comes (in part) thanks to a .226 BABIP.
Another method in building a bullpen for cheap, though, is one espoused by the great Joe Posnanski:9 What if teams deliberately avoided putting their best reliever in the closer spot? As Posnanski notes in his piece, the rate at which teams close out games with a lead in the ninth inning has not changed whatsoever since the arrival of the “closer” philosophy with Dennis Eckersley on Tony LaRussa’s A’s teams. Teams with a lead heading into the ninth won 95.5 percent of the time in 1952. They won exactly 95.5 percent of the time in 2010, too.
Using your best pitcher as a relief ace outside of the ninth inning isn’t anything new, as rubber-armed firemen such as Goose Gossage were de rigueur in the 1970s. Those guys picked up saves, though, and Posnanski’s model would deliberately avoid keeping these relief aces out of the save situations that produce big bucks in arbitration and free agency.
Organizations would end up deriving a significant financial advantage. Setup men are paid far less than closers with comparable statistics (the notable exception being Rafael Soriano’s three-year, $35 million deal), both in salary arbitration and on the free market. While teams would likely find resistance if they tried to move a closer in his prime into this new fireman role, it wouldn’t be all that difficult for them to start a player in that role and just keep him there. Posnanski notes the White Sox did this in 2010 with Matt Thornton (who they then promptly tried as a closer in 2011 and gave up on), but there are other teams that are better suited for it. The Red Sox, for one, might already be pursuing such a strategy with Daniel Bard. Although Bard was clearly the best reliever in the Red Sox bullpen last year, they were better off using him in tight spots and leaving the easier work at closer to Jonathan Papelbon. After this season, Papelbon will be a free agent, while Bard will hit arbitration for the first time. The Red Sox will save somewhere around $1.5 million this year by keeping Bard out of the closer’s role, and if they move Bobby Jenks to closer and keep Bard in his current role, that figure could rise to about $10 million over the course of three seasons. They don’t need the financial help, but it’s a smart way to optimize your player usage and save some money. It’s a plan other teams around the league might very well begin to follow.
There are dozens of other ways teams create opportunities for themselves in the post-Moneyball era. Rebuilding franchises such as the Royals, Nationals, and Pirates are spending more in the draft than the Yankees or Mets, finally understanding their success is inextricably linked to developing young talent. The Reds and Rangers have taken huge leaps of faith in the international free-agent market, shelling out record deals for developing players in the hopes that they’ll hit the lottery. The A’s even gave a 16-year-old pitcher from the Dominican Republic $4.25 million in 2008, hoping Michael Ynoa would develop into the next Felix Hernandez. (He suffered an elbow injury and was forced to undergo Tommy John surgery last August. It’s still a lottery ticket.) Teams are taking advantage of their unique ballparks more than ever before, as the Padres go out and acquire pitchers with high home run rates, only to watch them plummet in Petco Park. After a run as a statistically-inclined organization under J.P. Ricciardi, the Blue Jays are investing more in scouts than anyone else in the league. The list goes on.
In reality, not only is Moneyball still relevant, but the process of identifying undervalued markets and exploiting them is more important than it was when Moneyball first hit shelves. The rich teams are a lot smarter than they were in 2003; quite frankly, it’s a lot more difficult to succeed as a small-market team now than it was back during W’s first term. While low-budget teams are finding those opportunities in different places than they were back in 2003, the reality is the overarching concept behind Moneyball hasn’t changed: Get more for less.
Bill Barnwell is a contributing writer to Grantland.