"Terry Collins and Sparky Anderson, two silver-haired old-schoolers, were talking baseball in Collins' Anaheim office one afternoon in the late 1990s, when Joe Maddon charged in, his mind packed with data and his hands full of paper.
"Collins' bench coach had been studying spray charts of Ken Griffey Jr.'s hits, and noticed that the lefty slugger almost always sent the ball toward right field. So, Maddon suggested, why not shift all the fielders over to the right side? The strategy was not unprecedented - Cleveland manager Lou Boudreau had once employed it against Ted Williams - but it was highly quixotic for the contemporary game.
"Collins looked directly at Maddon and said, ‘Go for it. Do it. Why not? The information is there. Let's work at it.’"
Some fifteen years after Collins’s encounter with Maddon, the defensive shift has become a fixture of the game. In 2011, the shift was used fewer than 3,000 times, according to data compiled by Baseball Info Solutions (BIS); this year, BIS projects that number to be north of 14,000.
What explains this rapid sea change in defensive strategy? In short, the shift works. According to Steve Moyer of Inside Edge (IE),
"Shifts have saved a net of 390 hits this season through Monday [September 8]. If we were to add those 390 hits back into the grand total, the overall MLB batting average would rise to .254 from .252—a significant increase considering we're talking about 146,785 at-bats."
You might expect the Mets to be at the forefront of this successful and cutting edge experiment in defensive alignment, given the analytical approach of their front office and Collins’s early openness to shift. Surprisingly, this doesn’t seem to be the case. The defensive data suggest that, in fact, the Mets use the shift less frequently and less effectively than do most other teams.
According to IE data released last month, the Mets shifted 294 times in 2014 as of September 8. The Mets’ shift total ranked 28th in all of baseball and 13th among NL teams (the latter being a fairer basis of comparison given that AL lineups have an extra hitter).
The Mets also ranked low in shifting efficiency. The team saved just four net hits using the shift, good for 25th in the majors. They also ranked 25th in shifts per net hit saved, a useful stat that expresses these numbers in rate form. On average, the Mets saved a net hit every 74 times they shifted (whereas, for most teams, that number was less than 40).
Unfortunately, publicly available data on the shift are hard to come by. In February, the Hardball Times (HT) released some of its 2013 shift data, whose metrics and methodology differ slightly from those of IE. This is something to keep in mind when comparing the two datasets.
For example, IE’s shift totals include balls hit in and out of play, while HT measures only balls hit in play. Also, HT's numbers do not include net hits saved, although they do include component metrics from which net hits saved can be derived: Along with times shifted, HT provides each team’s opponent batting average on balls in play both with and without the shift on. As Chris Teeter of Beyond the Box Score explains, "Multiplying the BABIP difference score by the total [number of] times the team had the shift on provides the number of hits a team saved (or gave up) by shifting."
HT’s 2013 data suggest that the Mets were slightly more aggressive and effective shifters last year than they were in 2014, relative to the rest of the league. For example, the 2013 Mets shifted 155 times on balls in play, which ranked 17th in all of baseball and fifth in the National League. The Mets ranked 18th in the majors in each of the three efficiency metrics derived from HT’s data: shift advantage (.007, the difference between BABIP without the shift on and BABIP with the shift on), net hits saved (one), and shifts per net hit saved (143).
The two key takeaways from the available datasets are, first, that the Mets don’t shift all that often compared to other teams; and second, that, when they do shift, they don’t get a very impressive return. Moreover, the Mets' relatively weak shift outcomes are not a product of other teams shifting especially well against Mets hitters. In fact, the two Mets most often targeted by the shift—Lucas Duda and Curtis Granderson—seem to fare pretty well against it. Moyer reports that, despite facing more than 300 shifts this year, Duda neither gained nor lost any net hits as a result.
BIS data from early June (unfortunately the most recent numbers available on Granderson) show Granderson hitting .381 on grounders and short line drives against the shift in 2014. If anything, other teams seem to be unusually inept at shifting against Mets hitters, which should boost the Mets’ relative shift outcomes.
All that said, it wouldn’t necessarily be fair to blame the Mets’ shifting woes entirely on poor organizational and coaching strategy. As Moyer explains, shift outcomes can be heavily influenced by non-controllable factors (i.e., luck):
"Take a look at the Pittsburgh Pirates. They ranked No. 5 in MLB in total shifts attempted in our sample with 826. But the net result of all those shifts has been -2 hits, meaning the Pirates allowed more hits because of the shift than they would have if they'd been positioned normally.
"It is tempting to conclude that the Pirates are just doing it wrong, or that hitters are finding ways to ‘beat’ their shifts by hitting or bunting the ball where the defenders aren't. But neither scenario appears to be true. After watching hundreds of plays in which the shift failed to prevent a hit, we found that most times the ball just went to a place where it normally wouldn't—the result of a check swing, a broken bat or dumb luck."
The data seem to support Moyer’s argument. Three teams considered to be pioneers of the shift—the Rays, Pirates, and Astros—saw their level of shifting success vary wildly over the last two years. The Rays went from second in net hits saved last year to 26th this year; the Pirates went from ninth to 28th; and the Astros went from 17th to first. (The Mets, always the model of consistency, were in the bottom half of the league in both years.)
While there was a strong correlation between how often a team shifted in 2013 compared to how often it shifted in 2014, there was virtually no correlation year-to-year between its net hits saved or shifts per net hit saved. In other words, teams generally used the shift consistently from year to year, but experienced inconsistent and unpredictable returns.
Interestingly, within each of the last two years, there was a fairly strong correlation between the number of shifts a team employed and its net hits saved—but a weak correlation between shifts employed and shifts per net hit saved. This suggests that, generally, teams will be rewarded by playing the odds and making liberal use of the shift, even if the "bang for the buck" on each shift seems relatively small and hard to predict.
The logical conclusion, based on these findings, is that the Mets should probably use the shift more often than they do. Obviously it’s hard to criticize the organization too harshly for its defensive approach, considering that it has access to so much more defensive data than do we as fans. But it does seem fair to question why the Mets haven’t used the shift more frequently than they have, and what they’re doing to make the shifts they employ more successful.