Comparing PADE to individual defensive metrics

Defense is a tricky, tricky thing. Cognitively, it's much easier for us to demonstrate that something has happened as opposed to demonstrating that something has not happened. It's easy to identify that Andrew McCutchen hit a home run, and that scored three runs because Starling Marte and Josh Harrison were on second and third.

It's a little bit tougher to try and identify McCutchen's skill when he's chasing down a ball in the gap -- especially when we take the world of context into account. Does his positioning have anything to do with his ability to make the play? How about his route? Is it his fault he can't get to something, because the pitcher hung his curve and Matt Holiday mauled that one into the gap? Did he not get to a ball because he shades towards right field due to Marte's range in left?

It's tremendously difficult to separate context (the other defenders, the pitcher, the hitter) from the end result. Goodness knows, people have tried. The new defensive metrics (UZR, DRS, FRAA, etc.) have done about as well as one could imagine, despite their imperfections -- and no matter what complaints you hear about them and their reliability, using them is WAY better than throwing up our hands and guessing.

But you are reading Beyond the Box Score, and you know stuff about baseball, so you already know this, don't you?

If context is so important to defense, and I believe it is -- aren't we better suited to using a more holistic model of defense to determine true defensive ability? Is there a chance that the best way to view defense is to look at the defense -- or the run prevention model as a whole -- is in the aggregate?

So let's take a moment to look at Baseball Prospectus's Park-Adjusted Defensive Efficiency (PADE) metric, as it looks at holistic team defense, rather than the sum of aggregate individual defensive ratings.

A Brief History of PADE

Developed by James Click (currently the Tampa Bay Rays' Director of Baseball Research and Development) back in 2004, PADE is a pretty interesting tool. By creating defensive park factors, Click was able to take Bill James's original Defensive Efficiency metric (which tells us the percentage of balls in play a defense turns into outs), and turn it into something that not just told you how the park factors effect that number, but what percentage above or below average a team's efficiency was.

For example, a PADE of, say, -0.99 -- like the 2014 Atlanta Braves had -- means that the team in question was about 1% worse than the league average at turning balls in play into outs.  A PADE of 2.49 -- like the 2014 Baltimore Orioles had -- means that the team was about 2.5% better than league average at turning balls in play into outs.

Pretty cool stuff. And by it's nature, it takes into account some contextual things -- positioning and shifting, for example -- that the individual metrics currently do not.

So, you can check out yearly PADE metrics over at Baseball Prospectus. And you should do that. Also, here's a brief data visualization of the 2014 numbers.

I'll cover a few things that I found most interesting from the 2014 PADE leaderboard -- and from the past couple of years -- right here, right now.

Year-over-year consistency at the top ...

In 2014, the second-best PADE (3.21) was registered by the Oakland Athletics. In 2013, they had the best PADE in baseball (3.27), and in 2012, they were second place as well (2.48). Yet if you look at defensive value (based on Ultimate Zone Rating) at FanGraphs, or if you look at raw Defensive Runs Saved (DRS), the A's are a middle-of-the-pack team, at best. Beyond that, the Coliseum in Oakland is considered a relatively easy park to play defense in, so the team doesn't get extra credit -- they lose it.

The Rays were a top-3 team in 2012 (2.51 PADE, No. 1 in baseball) and 2013 (2.22 PADE, No. 3), but slipped to No. 8 (1.49) in 2014. Cincinnati has also been very, very good over the past two seasons, according to PADE -- they marked 2.57 in 2013 and 2.89 in 2014.

... and at the bottom! We see you, Colorado!

Colorado gets dinged pretty heavily as a tough park to play defense in, but PADE is unrelenting. The Rockies have been the worst team in baseball by PADE for the last three seasons. While 2012 was a true nightmare (-5.61), 2013 and 2014 have just been terrible (-3.68 and -3.05, respectively).

The Rockies are interesting, though. According to the last published park factors (in 2004), Coors was considered one of the worst parks in MLB to play defense in. Coors has pretty sizeable dimensions, ostensibly to make up for, y'know, the dingers. And sure enough, the Rockies have been a very bad defensive team by UZR (-54.5) and DRS (-77) over the last three seasons. But that's hardly the worst team in baseball by those metrics -- Houston and Cleveland, everybody!

Perhaps the PADE park factors don't give the Rockies enough credit for playing in Coors? Or perhaps it is an issue endemic to the Rockies' sparsely-defended outfield? Think about it -- this is a team that now features three elite defensive infielders (Nolan Arenado, Troy Tulowitzki, and D.J. LeMaiheu). But their outfield over the past few seasons? Woof.

From 2012 to 2014, Michael Cuddyer was worth -26 DRS and -15.8 UZR in 1864 innings in right field. Dexter Fowler was worth -13 DRS and -15.5 UZR in 1947 innings in center field. Carlos Gonzalez was worth -9 DRS and -4.6 UZR in 2371 innings in left field. Charlie Blackmon? Average defender, maybe below-average. Corey Dickerson? Poor defender.

In addition, the Rockies get negative defensive value, and a lot of it, from backstop Wilin Rosario.

Could the Rockies be an example of a team where the individual defensive numbers belie just how important outfield defense is in Coors Field? Is it possible that PADE is trying to tell us that, even though the range metrics look kinda bad on the surface, the Rockies would be better served by trying up their range in the outfield and get better coverage -- because those individual numbers don't even tell the whole story?

The Flipside: Cincinnati

I mentioned Cincy before, and that's a team that tells me a little different story. Over the past two seasons, that's a team that has had very good defensive numbers -- both by PADE and DRS/UZR. That probably starts with the up-the-middle combo of Brandon Phillips and Zack Cozart, who've been strong net positives for the past 3700 innings.

Player	Position	2014 DRS	2014 UZR
Billy Hamilton	CF	14	20.1
Zack Cozart	SS	19	12.1
Brandon Phillips	2B	6	8.1
Todd Frazier	3B	7	6.5
Devin Mesoraco	C	2	N/A
Ryan Ludwick	LF	-6	-4.8
Jay Bruce	RF	-7	-6.1
Joey Votto	1B	5	0.9
Brayan Pena	1B		1.4

Interestingly, the Great American Ballpark -- due to its tight dimensions -- is also one of the easiest parks for defense, according to Click's defensive park factors. Aside from Billy Hamilton (and Drew Stubbs), the Reds have effectively punted defense in the corners, often rolling out talents like Jay Bruce and Ryan Ludwick, and that's without mentioning Shin-Soo Choo's time as "center fielder."

In 2013, the Reds effectively punted outfield defense -- the combo of Bruce, Choo, and a leftfield cast of thousands is hardly a recipe for fantastic outfield defense. But the team stacked the infield with good defenders (Todd Frazier was, and is, an underrated third-sacker), and lo-and-behold, they turned 2.5% more balls in play into outs than an average team.

Was the decision to punt outfield defense done on purpose, given the way the ballpark plays? Was the infield defense SO good that it made up for the park's outfield defense on it's own? All of these are questions requiring more investigation, but from the outside it looks as if Cincy played to their park's strengths, and drew defensive value from it accordingly.

The Curious Case of the Royals

There was a very powerful narrative surrounding the 2014 Royals: that their success was driven, in part, by phenomenal defensive performances from guys like Lorenzo Cain and Alex Gordon. And make no mistake, these outfielders are tremendous defensive players. Cain and Gordon certainly pass the eye test, and their range metrics give players on the Royals all the credit in the world. Over the past three years, Kansas City rules the UZR and DRS leaderboards.

Year	Team UZR	Team DRS	PADE
2014	61.1	40	0.58
2013	79.9	93	0.7
2012	-15.5	12	-2.58

Yet when it comes to PADE, Kansas City is a middle-of-the-pack team, playing in a middle-of-the-pack stadium for defensive park factors. In 2012, the Royals were one of the worst franchises by PADE (-2.78), but in the past two years they were seventh (0.70 in '13) and 12th (0.58 in '14).

How could Kansas City be an elite defensive team via the individual metrics, and pedestrian when it comes to PADE? It probably isn't due to massive changes thanks to the park factor of Kauffman? Do UZR and DRS overvalue the incredible efforts the outfielders (and Sal Perez ... and Mike Moustakas ... ) make? Or does PADE undervalue those contributions?

I know, I know. More questions than answers. Not really what you're looking for from a sabermetric analysis article.

Honestly, I'm not sure yet what to make of the PADE data, and how to integrate it into my overall view of defense. There are items that make intuitive sense (the Reds and Rockies!), and others that don't (the Royals!). If nothing else, it's another data point, and it can give those of us who believe in a theory of interconnectivity when it comes to defense a way to try and quantify team skill.

Where do you think we should start digging deeper in trying to understand team defense?

. . .

Special thanks to Rob McQuown at Baseball Prospectus for answering a couple of PADE questions. All statistics courtesy of FanGraphs, Baseball Prospectus, and Baseball Reference.

Bryan Grosnick is the Managing Editor of Beyond the Box Score. You can follow him on Twitter at @bgrosnick.