Statistical Toolbox: Defense

facebooktwitterreddit

Credit: Brad Penner-USA TODAY Sports

After going over hitting statistics that I find useful last week, I promised to cover pitching this time around. But instead, I’m going to leave pitching for a future post. Today, and in a follow-up blog tomorrow, I address defense.

As opposed to hitting and pitching, defense is difficult to quantify. Every plate appearance for a batter produces a result. Did the batter get a hit, walk, or make an out? If the ball was hit into play, to which part of the field did it go, what type of batted ball was it, and what was the result? Likewise, every single major league pitch is a concrete data point with an identifiable result. What type of pitch was it? Was it called a ball or a strike? Was it hit into play? All of this data is stored away, awaiting an analyst. Defense is different. Unlike pitching and hitting, defensive actions do not accumulate data in the same, predictable, repeatable, manner. Not only that, but defense is the only area of baseball where some of the data collected is subject to the judgment of the official scorer, as well as our aesthetic perception of the game. In this post, I’m going to cover just two defense statistics: one that everyone is probably familiar with and another possibly unfamiliar metric that attempts compensate for what the first does not answer.

The Question: how can we determine the defensive quality of a player based on the plays the player does and does not make?

The Statistic: Fielding Percentage (FP), measured by adding putouts and assists and dividing it by total chances, which includes putouts, assists, and errors.

What it does and does not tell us: the foundation of FP is the Error. The Error is the answer to the following question: how can we best shame players for the mistakes they make? As a character in Michael Chabon’s Summerland lamented, “they are a part of life, sometimes I feel like that’s all I do in life, keep track of my errors.” Errors tell us that a player failed to make a play that should have been made. The operative word here is should. First, there is a great deal of subjectivity when identifying an error, and thus also FP. Sometimes it’s easy. When a shortstop throws a ball into the dugout, that’s obviously an error. But if a shortstop bobbles a difficult play and the runner just beats out the throw at first, it could go either way. The result is that we don’t know as much as we want about the defense of a given player. The second caveat is even more telling, and it has to do with perception. Namely, not all defenders are created equal, and creating the chance to make error has more value than avoiding errors due to a lack of opportunity. This is best explained in context.

In Context: Baseball is, in many ways, an aesthetically pleasing game. For example, say a fly ball is hit into the gap between the center fielder and the right fielder. It is not quite a line drive, so it has some hang-time to it, but it still might land for a hit. The right fielder isn’t going to make the play, but the center fielder makes a diving catch to make the out. It wasn’t one of those snow-cone grabs that leaves the ball balancing at the edge of a glove, but the center fielder couldn’t have made the catch without diving. It was a play made by extended arms followed by a fall to earth rather than anything of the superman variety. Still, one might describe a play like that as beautiful and give credit to the center fielder for making an outstanding play in the outfield.

In another scenario with a different center fielder but the exact same context of where the ball is heading and where the fielder was positioned before the ball was hit, the center fielder glides to the ball and makes the catch. We call those routine. Both were outs, but because one was more aesthetic, we tend to place more value upon it. The diving effort was nice, but a better fielder—perhaps due to speed, instinct, or a better route to the ball—wouldn’t have to dive. As Graham MacAree puts it, we tend to overvalue easy plays that are made to look difficult, while we undervalue difficult plays that are made to look easy. Such plays also influence FP because better fielders have more opportunities to make an error. Let’s use shortstop for this example. One shortstop might get to more balls, thus giving himself more chances to make an error and lower his FP. A poorer shortstop might just see those same balls head into the outfield for singles without ever touching the ball. There is no error, and thus no change to FP. There is also no value in comparison or evaluation when it comes to defense. It is an incomplete statistic.

*          *          *

The Question: what is the total defensive value of a given player relative to others who play the same position?

The Statistic: Defensive Runs Saved (DRS), developed and calculated by dark artists at The Fielding Bible.

What it does and does not tell us: the concept of DRS is simple. It measures defense based on value in terms of runs saved or given up. DRS is a complete statistic for defense in the same way that wOBA is for hitting. It is a total metric built from specific statistics that measure anything from the ability for first and third basemen to field bunts to the quality of an outfielder’s arm. DRS is also park adjusted, so the expansiveness and accompanying difficulty of playing at places like Coors Field are taken into account.

The caveat for all defensive metrics is that they require larger samples than hitting statistics. FanGraphs recommends taking a three year view of DRS before drawing conclusions about a player’s defensive value. One year snapshots are still useful. Indeed, beginning last year, DRS became a component in awarding Gold Gloves. Another caveat: a lot of people simply don’t trust defensive metrics. The reasons are generally due to the difficulty in measuring defense, as well as the disagreement among the most popular statistics. For example, the FanGraphs analog to DRS is Ultimate Zone Rating (UZR). While UZR can be placed on the same scale as DRS, as indicated below, the numbers don’t always match. Think of it in terms of wOBA and OBP: wOBA is scaled to reflect OBP for the sake of clarity, but a player’s wOBA will not necessarily correlate to his OBP.

In Context: I am a fan of DRS because it is centered on the most important factor of baseball: runs. The most vital role of a defender is to prevent the opposition from scoring runs because that is the clearest path toward a team victory. On a scale from the good kind of obscene to the bad kind, a DRS of 15 generally sits is at the top, and a mark of -15 is at the bottom.

Defensive Runs Saved (DRS)

(good) Obscene

Exceptional

Distinguished

Ordinary

Inferior

Shoddy

(bad) Obscene

151050-5-10-15

It is important not to forget our hitting statistical toolbox when thinking about fielding metrics. Every position comes with a different value. It’s easier for a below-average hitting shortstop to remain employed if he provides quality defense. Andrelton Simmons of the Atlanta Braves in 2013 had a wOBA of .303 with a wRC+ of 91—yet his defense at shortstop was an absurd plus 41. He was one of the most valuable players in all of baseball last year. On the other hand, first basemen are never sought for their glove because the position’s value is tied more directly to the bat. Just like most things in baseball, it’s all relative.

Finishing off in Rockies speak, let’s first appreciate how incredible Nolan Arenado’s defense was last season—his DRS was 30. That means that his run prevention was doubly obscene. Of course, there are two ends of the spectrum. Without naming names, I’ll just note that the Rockies worst fielder in 2013 had a DRS of -16. His name sounds like Carmichael McDyer, and he gets whatever the opposite of a Gold Glove is. He’s also the subject of tomorrow’s post.