clock menu more-arrow no yes mobile

Filed under:

On the Astros: Dexter Fowler's defense, or a Jump to Conclusions mat

How has Houston's center fielder done defensively this season?

Bob Levey

Have you checked Jordan Lyles out lately? Have you seen what he's done in Colorado? It's been all the rage. It's gotten people to declare a win for the Rockies after a month of the 2013 season.

Lyles has a 2.62 ERA in 44 innings over seven starts with 24 strikeouts and 10 walks. Meanwhile, Dexter Fowler is hitting .248/.320/.354 for the Astros with two home runs and three steals. He also has -4.5 Fielding Runs already and has missed time with an injury. In short, he's done exactly as badly as everyone predicted.

I won't defend or deny the deal in this article. I've come to bury defensive statistics, not to praise the deal.

Most of the writers on this site, myself included, use WAR all the time. I use it when deciding on players of the week. I use it when comparing which free agent signings were busts and which were successes. I use it all the time.

But, do I think about the early-season implications of that defensive component? Not nearly enough, I think.

WAR was derived to compare players across eras. It was supposed to be a way to look at a player's contribution at the end of a season, before a real-time database as powerful as FanGraphs or Baseball-Reference was able to churn out daily calculations for players.

That instant information is both a blessing and a curse. We can instantly look at one number and decide how well a player has done, without examining the underlying causes.

But, it goes deeper than that. We also know that defensive statistics take a long, long time to stabilize. We may understand this, but it still doesn't keep us from praising Jonathan Villar's defense (which I did) or dismissing the data on Dexter Fowler's defense (which I'm about to do).

We write about the Astros daily, so we're expected to analyze them constantly. But, with defensive metrics, that can be a problem.

Take Fowler's numbers, for instance. He's got a very negative Fielding Runs total. Only qualified center fielders in the majors have worse defensive numbers than him. Same goes for Defensive Runs Saved, which already has Fowler at minus-5, good for third-worst in the majors. Same goes for UZR, which has Fowler as the third-worst center fielder in the majors.

A funny thing happens, though, when we look at the all-new Inside Edge fielding data.

There, we see that Fowler has made every single one of the plays they classified as "routine." He had 22 chances, nailed them all. He had one chance at a play labeled "unlikely," which only get made 10-40 percent of the time, and he missed it. Then, he had seven plays labeled "impossible," plays that get made zero percent of the time. He made zero of these.

Do you know how many plays he's credited with being in his "zone" and how many he's made, according to the advanced fielding stats? He had 57 "in-zone" plays and 49 made plays. Eight misses, or the exact number of impossible and unlikely plays combined. Carlos Gomez already has 10 "impossible" plays, but only four misses on balls in his zone. Subjective, much?

What if we scrub out those "impossible" plays from his record? His RZR jumps to .980, or the highest among qualified center fielders in the majors. Think that will influence his UZR? Think that will influence his WAR?

There's a genuine flaw in how we perceive value. Using WAR as a catch-all this early in the season falls into that trap. We don't really know what a player's defensive value will be. Not even the metric systems do and they might not normalize until the last games of the season.

Will we know then whether Dexter Fowler was a good addition? Will we know if he was worth as much as Jordan Lyles? Probably not, but we'll be a sight better than trying to make that call right now.