With pitchers and catchers reporting soon, we just about have all the major free-agent signings wrapped up. It looks like Yovani Gallardo is about to sign with the Orioles, and there are rumors that Dexter Fowler could soon join him now that Baltimore has apparently decided to give up their first-round draft pick. With so many good outfielders available this winter, as the music stops, the available chairs for Fowler seem less desirable than we might have thought at the end of last season. Fowler’s defensive numbers have been pretty bad over the last few seasons, but his UZR figure was close to average in his only year with the Chicago Cubs. Determining change in skill from small sample size can be difficult, but it does not seem likely Fowler greatly improved his defense last season.
Defensive statistics are much-maligned outside of the analytic community, and even among those who use advanced statistics, there’s a degree of doubt regarding their utility. Much of the criticism stems from a misunderstanding about how to use defensive statistics given the larger sample size necessary to draw meaningful conclusions. Looking at three seasons worth of UZR can seem like going too far into the past, leading to rationalizations about short-term spikes in defensive numbers.
Over the last four seasons, Fowler has put up a UZR numbers in center field of -13.6, -1.7, -21.8, and last year’s -1.9 mark. Taken on the whole, it would be reasonable to conclude that Fowler is a below-average center fielder. It might be easy to look at the last three years, point to two that are pretty close to average, and call the -21.8 an outlier. This isn’t advisable, however, as doing so completely ignores a full year of data and merely cherrypicks the good seasons. The graph below shows two lines: yearly UZR for Fowler over the past five seasons, and a three-year average of UZR over the past five seasons, which is more representative of Fowler’s defense.

Using three years of data smooths Fowler’s numbers considerably. The inconsistency of the yearly numbers largely evaporates and places Fowler’s defense somewhere between six and 12 runs below average at center field over the past few years. That inconsistency carried over to his WAR numbers as well, as the graph below shows.

Instead of a 1.4 WAR season followed a 3.2 WAR season, we see a pair of 2.5 WAR seasons the last two years. Not surprising for the soon-to-be 30-year-old, ZiPS sees Fowler as a 2.4 WAR player heading into next season.
But what about the possibility that Fowler’s defensive improvements are real? It does happen. Jhonny Peralta made himself a much better shortstop later into his career — and improved positioning could have helped Fowler last season, as could getting out of the ballparks in Colorado and Houston. That narrative was getting pushed early last season and it does have some validity.
Read the rest of this entry »