by Keith Law
These metrics also can’t distinguish between a player who makes a lot of plays because he’s fast and a player who makes a lot of plays because he positions himself well before the ball is hit. Cal Ripken Jr. is considered one of the best defensive shortstops in history, something that is backed up by the rudimentary defensive statistics we have for his playing career, even though he was a well-below-average runner and in his era was the biggest man ever to play shortstop on a regular basis. (Ripken was listed at 6′4″, 200 pounds. At the time Ripken retired, no player at least that tall and weighing at least that much had ever played a single game at shortstop. Corey Seager and Carlos Correa, both of whom debuted in 2015, are both listed at 6′4″, 215.) Ripken was well known for his pre-pitch positioning based on who was on the mound, the pitch the catcher called for, and who was at the plate. According to his biography at SABR’s website, he would frequently attend the pre-series meetings pitchers and catchers would hold to review the opposing batters and discuss how to pitch to them. This observation and knowledge of where a hitter would be likely to hit a certain pitch allowed Ripken to position himself better before the ball was even put into play; he led the American League in assists seven times, and recorded the third-most assists in major-league history, thanks to his positioning acumen . . . and the fact that he didn’t miss a game for seventeen years.
In UZR, Lichtman only compares balls hit by left-handed batters to those hit by other left-handed batters, and balls hit by right-handed batters to those hit by other right-handed batters. He does make some small estimations based on certain batter archetypes, including very-low-power batters who might lead opponents to bring their outfielders in, and speedy hitters who face infielders playing slightly closer to home plate to increase their chances of throwing the runners out on groundballs. Those are necessary adjustments but also inexact.
The second problem is a very modern one, that of the defensive shift, perhaps the single most noticeable change in how baseball is played that we’ve seen in the last decade. (I refer specifically to “overshifts,” where a team might play three infielders on the right side of second base against a left-handed pull hitter.) Balls hit to shifted fielders end up in the trash bin; there’s no value in crediting a third baseman who was shifted to short right field with fielding a groundball there when it was 120 feet from his normal position. BIS does track teams’ usage of shifts and how often batters face them, but a ball hit into a shift doesn’t give us any useful information on the fielders involved except as one point of evidence that the shift was a good idea.
Once we’ve figured out how often a fielder in question makes a certain play, then we need to figure out what that play is worth. A groundball through the hole on either side of the infield is not as damaging as a hard-hit line drive up the right-center gap; the first is almost always going to be a single, and the latter will almost always go for extra bases. Although this sounds intuitive, it’s where advanced defensive metrics throw people off the most, because the value we’re talking about here isn’t tangible: we know what a hit or a home run is worth, more or less, but here we’re often talking about a hit that didn’t happen, a hypothetical value that I think makes a lot of fans uncomfortable because we’ve all grown up (myself included) judging players on tangible numbers describing discrete events we saw and can count.
Here’s Lichtman’s explanation, taken from his UZR Primer found on Fangraphs’ site.
Let’s say that that same batted ball in the example above was caught by the CF’er on the first play of a game. Since typically someone will catch that same ball only 25% of the time (see above), this particular CF’er will get credit for an extra .75 plays—100% minus 25%. We then convert .75 plays into runs by multiplying .75 by the difference between an average hit in that location and the average value of an air ball out. A typical outfield hit is worth around .56 runs and any batted ball out is worth around –.27 runs, so the difference between a hit and an out is worth around .83 runs. . . . Since our fielder gets credit for .75 extra plays, we give him credit for .75 times .83 runs, or +.6255 runs for that play.
This particular example is straightforward because it involves one fielder; when a ball is hit to a zone where multiple fielders might have some responsibility, UZR splits up the credit or blame based on how frequently we might expect each of those fielders to make the play in question. But the core idea to remember here, even if you’re never going to calculate UZR or dRS or anything like that on your own, is this: a defensive play has a value in runs prevented, and that same play, when not made, has a value in runs not prevented.
If you add up all of the values of the plays each defender made and should have made but didn’t, you get a total value for his defensive contributions . . . or, to be more precise about it, you get an estimated total value. Advanced measures of defense do not provide an exact accounting, but rather an estimate of runs prevented or permitted. If a player has a UZR or dRS for a season of +8 runs, that means we think his defensive performance in that season was worth about eight runs above average. It doesn’t mean that he’s an above-average defender, or that he saved exactly eight runs above an average fielder at the same position. It gives us a rough idea of what his glove was worth, and the more games or plays we can include in that sample, the more confidence we might have that this number reflects something like his actual talent level on defense.
But what do these numbers and situations actually look like in practice?
Ozzie Smith ranks as the greatest defensive shortstop in MLB history, a beautiful case where what our eyes tell us (those backflips!) lines up with the hard data we have on performance. Smith recorded more assists than any other player in MLB history at any position but one, shortstop/second baseman Rabbit Maranville, who played from 1912 to 1935, when hitters put the ball in play substantially more often than they have in the modern era. Smith leads all shortstops in assists even though he played fewer games at the position than Omar Vizquel, Derek Jeter, or Luis Aparicio, and recorded the eighth-most putouts among all shortstops as well, behind seven guys who played before World War II. Baseball-Reference uses TotalZone, a metric that uses historical estimates on where balls where hit into play along with the recorded putouts and assists, to roughly value the defensive performances of players prior to the advent of play-by-play data, and Smith comes out as the most valuable defensive shortstop and fourth-most-valuable defender at any position in history, saving about 239 runs over the course of his nineteen-year career.
Smith’s reputation matched up with the defensive stats, and he was a productive hitter for a shortstop in an era where shortstops rarely hit at all, so his Hall of Fame case was extremely strong and he was elected by the BBWAA in his first year on the ballot in 2002, appearing on 91.7 percent of ballots cast, clearing the 75 percent threshold required for election. His enshrinement marked only the second time the writers had elected a player primarily known for his defensive abilities to the Hall, after the 1984 induction of third baseman Brooks Robinson, the most valuable defensive player at any position in MLB history.
There’s certainly something satisfying about having the best defenders at key positions—Smith at shortstop, Robinson at third, even Bill Mazeroski, selected by the Veterans Committee, at second base—in the Hall of Fame. The greatest eligible center fielder, Willie Mays, is in; the only center fielder above him in TotalZone’s ratings is Andruw Jones, who will hit the ballot after 2018. Jones’s case is complicated by the fact that he was effectively done as a big-league regular at age thirty and played sporadically and poorly for several years, eventually playing his last two seasons as a professional in Japan at ages thirty-six and thirty-seven. He was a far better player than Kirby Puckett—who sailed into the Hall of Fame on his first ballot but is one of the worst players the writers have ever elected—and yet Jones probably has a comparable case to Jim Edmonds, who failed to reach the 5 percent minimum in his first and thus only year on the BBWAA ballot.
Edmonds played fewer g
ames but held his value deeper into his career, and if you consider the inexact nature of defensive statistics it’s fair to think of Edmonds and Jones as more or less equivalent, at least in terms of a Hall argument. If you put Jones in, you probably should have Edmonds in, but since Edmonds received almost no support whatsoever on the ballot, it would be a surprise to see Jones fare substantially better.
Jones’s appearance on the ballot after the 2017 season won’t be the one to spark the big debate, however; that will fall to Omar Vizquel. As I’ve shown when discussing the flaws with fielding percentage, Vizquel was a very good player who has become comically overrated by fans, coaches, and media alike. Indeed this has progressed to the point where many simply say that the defensive metrics around him are “wrong” because, in fact, they don’t support the preconceived notion that Vizquel was a latter-day Ozzie Smith. Here’s the hard truth: he wasn’t. He was a good defender, but was maybe half of what the Wizard of Oz was with the glove. While old defensive stats like fielding percentage revealed Vizquel’s shortcomings, TotalZone and UZR make it even clearer:
So despite Vizquel playing a little more than a season’s worth of games more than Smith did, Ozzie made more plays than Omar did—a lot more, enough that TotalZone gives him an edge of over 100 runs prevented compared to Vizquel over the courses of their careers.
Neither player was any great shakes with the stick, either; Smith was a career .262/.337/.328 hitter, while Vizquel ended up at .272/.336/.352, playing most of his career in a higher-offense era than Smith did, so Smith’s offense ends up more valuable than Vizquel’s by more than 100 runs over their careers. And Smith was a far better base stealer, with 176 more stolen bases than Vizquel had but just 19 more times caught stealing.
But the crux of the arguments for Vizquel as a Hall of Famer is that he was close to Smith’s level on defense, when all of the available information says that he wasn’t. Vizquel may have been a joy to watch, and writers who covered him in Cleveland and San Francisco have nothing but praise for him as a player and as a person. His performances on the field, however, simply do not justify his reputation, and if elected—as I suspect will happen on his first or second year on the ballot—he’ll become one of the worst players enshrined in the Hall of Fame.
Speaking of first-ballot Hall of Famers, in July 2016 Cooperstown welcomed Ken Griffey Jr., whose appearance on 99.3 percent of ballots set a new record—and, by the way, none of the three voters who omitted him has had the courage to come forward. Junior was clearly a Hall of Famer by any standard, but the narrative of his career doesn’t quite match the facts, either—to wit, his insistence on continuing to play center field in the last ten years of his career meant that he stopped being a valuable big leaguer after age thirty, and his defense was so bad in his thirties that he had several seasons where he was actively hurting the team just by playing.
Griffey Jr.’s career through age thirty was one of the best in history, marred only by injuries that seemed to stem from how hard he played in the field. He was known for highlight catches in center, and up to age twenty-seven or so his defensive numbers were outstanding, with 88 runs saved above average by TotalZone through that season. After his age-thirty season in 2000, he had 438 homers, a .296/.380/.568 triple-slash line, and 76 Wins Above Replacement (per Baseball-Reference); had he quit right there, he’d have been worthy of enshrinement in the Hall, and he appeared to be on pace for one of the five or ten greatest careers of all time.
His defensive numbers started to slip in his late twenties, and he should have moved out of center field right around age thirty or thirty-one, his first two years in Cincinnati after he’d spent the first eleven years of his career in Seattle. From 2001 to 2006, his last six years as a regular center fielder, his defense was worth 73 runs below an average center fielder by TotalZone. We have UZR figures available from 2003 forward; between 2003 and 2006, UZR has him 66 runs below average (plus another two runs below average in 2008, when the White Sox traded for him and delusionally returned him to center). He destroyed significant value with his play in the field, and because no team was willing to force him out of center field until about five years too late. For most of those seasons his bat would have been below average for any position he might have been able to play. Combined with Griffey’s decline as a hitter to a .260/.350/.483 line from 2001 until he retired, he generated only about 7.5 Wins Above Replacement of value for the last half of his career, one-tenth of what he was worth in the first half. What could have been one of the greatest careers of all time stalled out after Griffey turned thirty.
While for some players and positions these advanced defensive metrics are useful for the academic debates around the Hall of Fame, when it comes to catchers, their use is much more immediate. Indeed, for catchers advanced metrics are becoming more and more vital in game-time decisions, and this comes despite the fact that UZR and other similar stats do not apply to catchers because catchers don’t have range or zones in the same way other players on the field do. Catcher defense is a sum of parts that include controlling the running game, receiving, framing pitches, and game-calling, only some of which is measurable with the data we have right now.
The defensive value of catchers has posed many specific problems over the years, given the complex nature of the catcher’s job: receiver, game-caller, and primary obstacle to base stealers. A catcher receives a putout on each strikeout, which is one of the most absurd statistical oddities in traditional stats and makes any stat based solely on putouts and assists useless for the position. Passed balls count against the catcher, but wild pitches do not, even though the difference between the two is a subjective decision by the official scorer.
Base-stealing value itself isn’t as easy as it may first appear, because of confounding factors around when opposing runners attempt to steal. Certain pitchers hold runners well, or have great pickoff moves, so runners might as well be nailed to the first-base bag. Other pitchers just don’t bother with the running game, like the Mets’ Noah Syndergaard in 2016. Opposing baserunners stole 48 bases in 57 attempts with Syndergaard on the mound that season, both of which were the highest in the major leagues by huge margins, because Syndergaard didn’t do anything to try to hold runners. The Mets may not have had Johnny Bench behind the plate, but their catchers’ caught-stealing rates were worse than they might have been if Syndergaard had worked to keep runners from stealing. (Even worse than Syndergaard was Yankee reliever Dellin Betances, who is the worst-fielding pitcher I’ve ever personally scouted; Betances threw 73 innings in 2016, and baserunners were a perfect 21 for 21 in base-stealing attempts with him on the mound.)
The throwing prowess of a catcher can also suppress base-stealing attempts: If you know a catcher can really throw, then you’re probably not going to take the chance on getting thrown out. Longtime Cardinals great Yadier Molina, who represents a sort of gold standard for catcher defense in fact and in reputation, has a career 42 percent caught-stealing rate—that is, runners have only succeeded in 58 percent of attempts—in his career in the majors, about 50 percent higher than the league average during those years. Defensive metrics will credit him with the plays he’s made, but public metrics don’t consider the plays not made—runners who just gave up rather than try to run against him. Molina has seen about two-thirds of the attempts that a league-average catcher saw in those years. Your advance scouting reports on the Cardinals will tell you to run on Molina at your own peril, and that should be to Molina’s credit somehow, even though any value assigned to him is ultimately an estimate of something that didn’t happen.
Then there’s the controversial but indisputably real subject of catcher framing, the process by which catchers can affect umpires’ ball or strike calls just by how they receive those pitches.
Measuring catcher framing is perhaps the newest catcher-specific defensive metric. The first public research on catcher framing appeared on the sabermetrics blog Beyond the Box Score in 2008, by Dan Turkenkopf, who subsequently
went on to run the analytics department for the Milwaukee Brewers. Pitch-framing was long one of those black-box topics in baseball, where everyone acknowledged that it existed, without actual proof of its existence or its degree, due mostly to our inability to measure it. By 2008, we were starting to get the data required to evaluate just how much value a good framing catcher could provide his team—or how much a bad framer could cost his team—and the results of such research have thoroughly altered the ways in which teams value and pay their backstops.
Framing is a euphemism for what is actually happening behind the plate: Catchers who frame well are stealing strikes. Good framers have the ability to take pitches that are, in nearly all cases, within two inches of the perimeter of the strike zone and get them called for strikes more often than the average catcher can; poor framers do the opposite. Home plate umpires are asked to do an impossible job, to call balls and strike from behind the catcher when their view is perforce obstructed by the catcher, and to determine where the ball was when it crossed the plane of home plate rather than where it was when the catcher caught it. Umpires make plenty of mistakes on ball/strike calls—even though, as a percentage of total pitches called, their error rates are surprisingly low, around 3 percent—and it turns out that good catchers can affect those borderline calls to their advantage.