Gold Glove process improving, but some choices still strange
The winners of the Rawlings Gold Glove awards, which are designed to recognize the best defender at each position in each league, were announced on Tuesday. Once the target of potshots—I coined the term “Fielding Grammys” nearly a decade ago—the awards have improved considerably in recent years thanks to the incorporation of advanced statistics into the process. Judging by the array of available defensive metrics, there's room to quibble with some of this year's winners but there were no egregious choices.
Established by the baseball glove manufacturer Rawlings in 1957, the Gold Gloves were voted upon by exclusively managers and coaches through 2012, with the caveat that voters could not select players from their own teams. Prior to the development of advanced defensive metrics, the award stood as the last word on glovework. Once a player won, he was often recognized year after year—sometimes for longer than a decade—on the basis of reputation rather than statistics, and at times it has seemed as though offensive numbers have seeped into the voters' minds. The introduction of better measures of defensive value made it easier to critique their choices, but sometimes even rudimentary numbers can suffice. In 1999, two-time winner Rafael Palmeiro of the Rangers added a third straight AL award at first base despite playing just 28 games there, evidence that plenty of voters had paid little attention.
Mindful of such criticism, in 2013 Rawlings began partnering with the Society for American Baseball Research to incorporate sabermetrics into the process via the SABR Defensive Index, which accounts for 25% of the vote. The votes of managers and coaches make up the rest, as before. The SDI aggregates two types of defensive metrics, those derived by batted ball type data (Baseball Info Solutions' Defensive Runs Saved, Mitchel Lichtman's Ultimate Zone Rating and Chris Dial's Runs Effectively Defended, a forerunner to later metrics) and those derived via play-by-play data (Michael Humphreys' Defensive Regression Analysis and Sean Smith's Total Zone Rating).
Batted ball data is richer than play-by-play data because it requires a trained viewer to distinguish line drives from fly balls and grounders. However that introduces observational biases into the process; even the height of the press box can influence how a ball is scored, particularly when it’s a hit instead of an out.
Between that and the extent to which methodologies of both types adjust for park and regress based upon sample sizes, the numbers vary from metric to metric in a given year, though they often tend to converge over sample sizes larger than a year. Defensive Runs Saved (DRS) is published at both baseball-reference.com and fangraphs.com, Total Zone Rating (TZR) only at B-Ref, Ultimate Zone Rating (UZR) only at FanGraphs, and Defensive Regression Analysis (DRA) is only at a site called The Baseball Gauge. Runs Effectively Defended (RED) is not publicly available, but Dial provided me with his spreadsheet upon request.
Below are the winners, with the number of Gold Gloves they've won in their careers in parentheses and an asterisk denoting if they're repeat winners from 2014. Note that only one player from this group, Yadier Molina, has won more than three times in his career, and that 11 of this year’s 18 winners are different from those who won last year. At the very least, that's a sign that the awards have moved beyond mere rubber-stamping by the voters.
For fun, see if you can spot the familiar word not mentioned anywhere in this story. (Answer at bottom.)
|First base||Eric Hosmer||Royals||3*|
|Second base||Jose Altuve||Astros||1|
|Third base||Manny Machado||Orioles||2|
|First base||Paul Goldschmidt||Diamondbacks||2|
|Second base||Dee Gordon||Marlins||1|
|Third base||Nolan Arenado||Rockies||3*|
Perhaps not surprisingly, it’s at this position where the metrics diverge the most from the hardware. UZR doesn't exist for catchers, but looking over the other individual metrics, the Nationals' Wilson Ramos stands out as a player with a legitimate gripe, as he had across-the-board advantages over Molina. Ramos led the league in DRA (+16.5 runs, compared to Molina's -3.4), Total Zone (+11 to Molina's +9) and Runs Effectively Defended (+6.9 to Molina's +5.6), and he tied with Buster Posey (who led the overall stat, the SDI) for the NL lead in DRS (+9, to Molina's +7).
Comparing Ramos and Molina via the traditional stats, Ramos had lower rates of stolen bases and missed pitches (wild pitches and passed balls) per nine innings, he caught a higher percentage of would-be base thieves and he even had a higher fielding percentage (a stat that’s best consigned to the dustbin of history because it makes no recognition of range).
Granted, none of those numbers exactly captures the impact a catcher has on his pitching staff, and it's worth noting that the Cardinals allowed an MLB-low 3.24 runs per game compared to the Nationals' 3.92, but some of that rests on the shoulders of the teams' other fielders, not to mention the pitchers. Washington's pitchers had a significantly better strikeout-to-walk ratio than did St. Louis's (3.7 to 2.8), even if Molina did have the better pitch framing numbers according to Baseball Prospectus (11.4 to 4.2), which itself is not part of the system. It's not surprising that Molina won another Gold Glove based on his reputation, but one can understand why Ramos, who struggled at the plate (64 OPS+ to Molina's 80), has a beef.
Meanwhile, in the AL, Perez's numbers (+1 DRS, +1.2 RED, +5 TZ and +9.2 DRA) don't make a tremendously strong case for him, though the consensus was more scattered; Tyler Flowers and Caleb Joseph were the top two in DRS, James McCann and Russell Martin in RED, McCann and Joseph in TZ, Carlos Perez and Brian McCann in DRA. The SDI rankings put Joseph (+6.2) and Jason Castro (+6.0) at the top, with Perez seventh (+1.0). Joseph probably has the strongest case here, though one can see how Perez’s edge in playing time (44% more innings), the Royals' success and its relationship to their emphasis on fielding could influence voters, as they had an MLB-high three winners.
Paul Goldschmidt led the NL in DRS (+18) but lagged in other categories. The Giants' Brandon Belt, who led in RED (+10.7) and UZR (+8.6) and was a strong second in TZ (+13), edged him in the SDI (+10.8 to +9.8). Belt only made 117 starts to Goldschmidt's 155, so one can understand why the voters might want to reward that attendance.
In the AL, Eric Hosmer was nowhere near the top of any of the rankings, but no other first baseman in the league led more than one metric. Chris Davis, who led in RED (+7.9) and was strong in DRS, and UZR, might have been a more justifiable choice, but he made 29 starts in rightfield and another 22 at DH, so some voters may have been less exposed to his work this year. Mike Napoli led in SDI (+4.9), but he was traded from the Red Sox to the Rangers amid a subpar year at the plate, so it's understandable how he could be overlooked as well.
Dee Gordon finished third in SDI (+4.5) in the NL behind Danny Espinosa (+10.0) and Addison Russell (+7.8), but the Marlins' speedster led the league in UZR, DRA and DRS even while placing dead last in RED. Given that consensus among the more visible metrics, it's not even necessary to invoke his offense—which included a batting title, a league-high 205 hits and a stolen base crown—as an additional crutch for voters, but it probably didn't hurt.
In the AL, Ian Kinsler led in RED, DRS and UZR and was a close second in TZ; his +10.7 SDI trumps Altuve's +5.7. Perhaps the Astros making the playoffs while the Tigers finished last in the AL Central in 2015 was a factor.
The consensus in the AL was less clear, with Adrian Beltre leading in both DRS and UZR, and Manny Machado in DRA and SDI; RED put Beltre in the red, bumping him to third place overall.
The two choices here were quite defensible (pun intended). Alcides Escobar led the AL in SDI and RED while Indians midseason callup Francisco Lindor led the league in both TZ and DRS; the former played 51% more innings than the latter, so that result is understandable in terms of both exposure and sample size.
In the NL, Andrelton Simmons had the advantage on Brandon Crawford in DRS, UZR and DRA by varying degrees, with the reverse true for TZ; Crawford led in RED while Simmons was in the red, however, and the SDI put the Giants' whiz on top, one of just three positions in the Senior Circuit where the Index leader prevailed.
From 1961 to 2010, the outfield awards went to the top three vote recipients in each league, regardless of position, but in 2011, the awards were split back into position. Last year, an arcane rule involving innings minimums and percentage of playing time prevented Kansas City's Lorenzo Cain, who had ranked second in the league behind teammate Alex Gordon in both DRS and UZR, from getting his due, because his time in rightfield, where he was often moved in the late innings, dropped him below the levels necessary to qualify.
This year, NL outfield DRS leader Ender Inciarte of the Diamondbacks (+29) was similarly penalized for seeing time at all three positions. Though he played only 372 of his 1063 1/3 innings in leftfield, his +13.5 DRA led the league, his +12 DRS was one run behind leader Christian Yelich of the Marlins, and he was only a few runs off the pace in TZ and RED—but officially, he didn't qualify due to games and innings in left. Yelich had a slight advantage over the leftfield winner, Starling Marte, in SDI (+8.2 to +7.4), though the latter was off the charts in DRS (+24).
Meanwhile, Yoenis Cespedes won AL honors in left despite being traded to the Mets in late July; unlike Inciarte, he did meet the playing time thresholds while leading AL leftfielders in SDI, DRS, UZR and RED while finishing a close second in DRA.
Kevin Kiermaier made an across-the-board sweep in the AL; his 29.1 SDI led all major league fielders at any position, nearly doubling runner-up Crawford's 15.4 and more than doubling AL centerfield runner-up Kevin Pillar's 13.3.
In the NL, A.J. Pollock was a reasonable choice but not nearly a unanimous one; he led in DRS and TZ but had a negative score in RED, and third behind Odubel Herrera and Billy Hamilton in the SDI, though he had significantly more playing time than both despite Inciarte's dabbling at that spot as well.
There was little to quibble about here. Jason Heyward led the NL in DRS, UZR and SDI while scoring well in DRA. Kole Calhoun led the AL in UZR, TZ and RED, and had a massive lead over J.D. Martinez in the SDI (+10.5 to +1.5).
This position lacks both UZR and DRA components and is based on a much smaller sample size in terms of innings in the field, but both Zack Greinke and Dallas Keuchel led their respective leagues in SDI and DRS.
SDI leaders took home three Gold Gloves in the NL and six in the AL. Catcher, first base and second base were the only positions where at least one of the two SDI leaders didn't win the hardware, while pitcher, third base and rightfield aligned in both leagues.
It's at catcher where one might have the most criticism of this year’s outcome due to the intangible factors, team success and the impact of longer-term reputations. But if that’s the worst that can be said about this year’s awards, they’ve still come a long way from the Fielding Grammys.
*And we made it all this way without once mentioning errors.