With Spring Training right around the corner, undoubtedly we will hear tons of stories about how the Mets are going to concentrate on fundamentals. It’s one of the rites of Spring. Included in fundamentals is defensive play. Most people think that the Mets did not have a very good defense last year. However, some people have good defensive reputations and others have poor reputations that may not match up with the numbers.

People are still hesitant to trust defensive numbers. Some of that skepticism is warranted, most of it is not. Many will tell you that numbers cannot judge who is a good fielder –- that somehow defense is best measured by visual evidence. That type of subjective evidence is useful and worthwhile but anyone who thinks they can accurately judge how good someone is defensively just by watching is kidding themselves.

Would you be able to tell the difference between a .275 and a .300 hitter just by watching? Sure, you could tell that Albert Pujols is a better hitter than Omir Santos. But could you tell if Alex Cora or Cory Sullivan was a better hitter last year just by watching? Who was a better left fielder for the Mets last year —Daniel Murphy or Gary Sheffield? They were both bad, but we need numbers to determine just how bad.

The eminent scientist, Lord Kelvin, said it best when he stated, “When you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind.”

I believe that part of the problem that people have with defensive numbers is that they hold them up to an unfair standard. People want a single metric that will tell how good or bad a player is at defense. We don’t have that type of accuracy for offense. We think a .300 hitter is good, but if he has a .320 OBP than he’s not really helping the offense. We think a .370 OBP is good but it is another matter entirely if the player combines that with a .320 SLG. We think a .550 SLG is good but if it comes along with 30 GDPs and a bunch of caught stealing, then it takes on another form.

We have to look at defense the same way.

Jeff Francoeur has a rocket for an arm and looks very athletic so people assume he is a good defensive outfielder. But what kind of a jump does he get on balls? Does he take good routes to the ball? How far can he go to catch a ball? How reliable is his glove? Is he better at going in or going back on a ball? Can he climb a wall and take away a homer? Is he willing to dive for a ball to make a play or is he content to play it on a hop? All of these and more are questions we need to consider in determining if a player is good defensively.

The old standby defensive metrics did not attempt to answer these questions. The newer metrics are trying (they’re certainly not perfect) and come much closer to giving a broader understanding of who is good in what categories.

I want to look at projected Mets starters in the infield and outfield and see what three different defensive metrics say for each player. Here I will use UZR, taken from FanGraphs, Plus/Minus, taken from Bill James Online, and Defensive Runs Saved from Chris Dial. Click on the hyperlinks if you want a more detailed explanation of the various systems. They have each been rigorously tested, and while not perfect, they do tell you a lot more than fielding percentage, assists and putouts.

1B – Daniel Murphy
UZR: Double Plays 0.6, Range 6.3, Error Runs -2.2, UZR 4.7, UZR/150 7.6
Plus/Minus: Bunts +1, To His Right +7, Straight On +5, To His Left +1, GB +13, Air +1, Runs Saved 10
DRS: 4.74

All three agree Murphy was a good fielder at first base. Plus/Minus had him as the second-best defensive player overall at the position.

2B – Luis Castillo
UZR: Double Plays -2.7, Range -7.9, Error Runs 0.2, UZR -10.4, UZR/150 -12
Plus/Minus: GDP Runs -3, To His Right -5, Straight On -1, To His Left -3, GB -9, Air -2, Runs Saved -8
DRS: -6.76

All three agree that Castillo was a poor fielder. Dial and Dewan (Plus/Minus) had him ranked nearly identical while UZR had a worse opinion.

3B – David Wright
UZR: Double Plays -1.1, Range -9.3, Error Runs -0.1, UZR -10.4, UZR/150 -13.2
Plus/Minus: Bunts -2, To His Right -6, Straight On 0, To His Left -2, GB -8, Air -1, Runs Saved -11
DRS: -6.88
All three agree that Wright was a poor fielder. UZR and Dewan had him as terrible while Dial had him a bit better.

SS – Jose Reyes
UZR: Double Plays -1.1, Range -0.6, Error Runs -0.1, UZR -1.9, UZR/150 -9.4
Plus/Minus: GDP Runs -1, To His Right +5, Straight On -1, To His Left -2, GB +2, Air -2, Runs Saved 0
DRS: 0.14

Dewan and Dial have him as basically even while UZR shows him below average in his brief stint last year.

LF – Jason Bay
UZR: Arm -0.7, Range -14.4, Error Runs 2.1, UZR -13, UZR/150 -11.2
Plus/Minus: Kills 8, Runs Saved 3, Shallow +1, Medium +6, Deep -15, Runs Saved -5
DRS: -21.51

All three see him as bad. Dewan thinks he’s pretty bad, UZR says really bad and Dial says awful.

CF – Carlos Beltran
UZR: Arm -0.4, Range -4.4, Error Runs 0.1, UZR -4.8, UZR/150 -8.5
Plus/Minus: Kills 1, Runs Saved 0, Shallow -7, Medium 0, Deep +11, Runs Saved 3
DRS: 2.31

Dewan and Dial see him as being an asset but UZR viewed his defensive season as a bust.

RF – Jeff Francoeur (combined ATL/NYM)

UZR: Arm 4.0, Range -11.6, Error Runs 1.5, UZR -6.1, UZR/150 -5.9
Plus/Minus: Kills 8, Runs Saved 8, Shallow -2, Medium +3, Deep -5, Runs Saved -1
DRS: -10.78

Dewan had him as a neutral fielder, UZR thought he was bad while Dial had him really bad.

*****

Murphy is the only player that all three systems agree was a good defensive player last year. He shows good range, does a very good job on grounders and has to work a little on his reliability with the glove. Meanwhile, Castillo, Wright and Bay all three systems say were poor fielders. And two of the three rate Francoeur that way. Beltran and Reyes have mixed profiles, which could turn positive with healthy seasons.

Of course, this was only one year of data. Just like a player can have a fluke offensive season, he can have a fluke defensive year, too. Beltran has usually been a very good defensive player. These numbers could have been a fluke or they could be an indication of how much his knees were hurting. Regardless, the bigger the sample, the better the conclusion.

But judging from these numbers, there really should be a lot of work put in this year on defensive fundamentals. The Mets do not figure to be a good defensive team in either the infield or outfield.

And to answer the question posed earlier in the article, Murphy had a -10.4 UZR/150 in left field last year while Sheffield posted a -35.4 UZR/150. Dewan had Sheffield with -8 Runs Saved and Murphy at +1. Dial had Sheffield at -5.4 in left field but did not have numbers for Murphy. I think it’s safe to say that Sheffield was much worse, although Murphy made some high-profile gaffes that made people think he was worse than he really was.

And that’s the reason you just can’t trust your eyes with defensive numbers.

2 comments on “The language of leather

  • JohnStrubel

    I am amazed at how the perception and statistical realities are polar opposites. Mets followers have labeled Daniel Murphy as a defensive liability, yet his is “the only player that all three systems agree was a good defensive player last year. He shows good range, does a very good job on grounders …” But Wright is widely considered a strength defensively (including two Gold Gloves). Reyes’ range will affect the Mets (and Castillo) up the middle. For better or worse? Time will tell.

  • Brian Joura

    I feel better about the possibility of Reyes being a plus defender next year than Wright. I do expect Wright to be better in 2010 but he really deserved those Gold Glove Awards as much as Jeter deserved his. He makes some spectacular plays but doesn’t have the consistency that you’d like to see.

Leave a Reply

Your email address will not be published. Required fields are marked *

The maximum upload file size: 100 MB. You can upload: image, audio, video, document, spreadsheet, interactive, text, archive, code, other. Links to YouTube, Facebook, Twitter and other services inserted in the comment text will be automatically embedded. Drop file here