He’s got wheels: Rating the center fielders

Every couple of years, your team, usually in need of a center fielder at the major league level promotes/trades for/signs from another country/signs off the street a guy with “wheels”.  Inevitably, the really fast guy gets put into center field, and sensibly so.  The center fielder has to cover a lot of ground, particularly on fly balls, so it helps to be fast.  And every announcer is convinced that his team’s center fielder is really super fast, and therefore the best ever (in the sportscaster sense of the word) at tracking down flyballs.
Who does have the best range in center field?  Those who have been reading StatSpeak faithfully over the past month or so know that I’ve been working on my OPA! method for turning Retrosheet data files into a defensive rating system.  I’ve previously looked at infielders and their primary mode of contributing to the defense, the ground ball, specifically broken into component parts (range, fielding, arm, and catch).  Today, I look at the bread and butter of an outfielder’s job description, the fly ball.  Sadly, a fly ball doesn’t give us quite the rich information that a ground ball does, as converting this ground ball into an out consists of the outfielder getting to the ball (range) and then grabbing it out of the air (catch).  A fielder gets credit for getting to the ball, even if he drops it (and is thus given a fielding error for his efforts).  In that case, he is debited in the “catch” category.  If he makes the catch, he gets proper credit for completing the play.  There are some out there who will point out that an outfielder is sometimes called on to make a throw after a fly ball (sac fly?).  To be honest, I’m going to simply steal/plagiarize/copy John Walsh’s (of The Hardball Times) outfielder arm ratings method (perhaps with a few minor tweaks, but John deserves the credit on this one).  I won’t delve too much into detail on arms in this article, as John’s already catalogued the results fairly well.
Now, on to creating some ratings.  First off, whenever we get into the outfield, we need to deal with park effects.  All infields are shaped the same and most infielders have roughly the same assignments.  Outfields can come in different shapes and sizes and the really big or really small ones can play havoc with how many balls an outfielder will find his way to.  A park effect correction is needed.  OPA! is based on league averages, but it makes no sense to compare a flyball hit to an outfielder in Safeco to what we might expect from a flyball in a bunch of tiny little matchbox ballparks.  The other (continuing) problem is that since these are Retrosheet data, we are operating in the absence of knowing where about the ball went.  I can guess that if the ball eventually ended up in the left fielders glove that it was somewhere to the left of second base, but that’s not incredibly helpful, is it?  Anyway, we must persevere.
I took all flyballs fielded by an outfielder (which means that home runs were politely excused from the data set since no one can field those).  For each flyball, I calculated the expectation that the fielder would get to the ball based on what all the visiting team fielders had done that year in that ballpark.  Why just the visitors?  Consider Andruw Jones during his heyday in center field.  The saying went that “2/3 of the earth is covered by water.  The rest is covered by Andruw Jones.”  Think a sample composed 50% or so of Andruw Jones might have overly influenced the number?  Using only the visitors gives us a decent sample of many of the outfielders in the game so that we can say that roughly, we’ve captured the league average.  This approach has some obvious flaws given the unbalanced schedule and interleague play, but it’s about the best we’re going to get at this point.  I split up debits for balls that dropped in for either singles or doubles (and I looked at both separately because a ball that goes for a double is more likely to hit the wall where it can’t really be turned into an out) by using the 1993-1998 Retrosheet files, much in the same way I’ve done previously.
Among those who played more than 450 innings in center field (50 games) in 2007, the best range per flyball hit in their area (in which they had some responsibility) belonged to:

  1. Alfredo Amezaga: .065 outs above average per flyball
  2. Coco Crisp: .054
  3. Nook Logan: .051
  4. Curtis Granderson: .049
  5. Grady Sizemore: .040

The bottom five:

  1. Kenny Lofton: -.007 outs above average per fly ball
  2. Nate McLouth: -.008
  3. Nick Swisher: -.014
  4. Marlon Byrd: -.019
  5. Jim Edmonds: -.022

The top five give gives us some names we might expect, all with reputations as excellent fielders.  The bottom five includes men who didn’t spend much more than 50 games at the position (McLouth, Swisher, and Byrd all spent the equivalent of 53-55 games there), and Lofton, who by the end of the year was patrolling left field most of the time. 
Careful readers will note that while the top five are all in the neighborhood of .04 to .06 outs above average, the bottom five are only .01 to .02 below average.  In fact, most of the regulars rated as above average.  How can that be?  There were about 150 men who patroled CF.  Only 33 of them made my cutoff of 450 innings.  Most of the part-timers were actually below average.  Managers apparently don’t fool around with center field defense too often, and when they do, it suffers.  There may also be a bias in my park effects in that I’m using visiting team fielding data.  After all, a fielder in his home outfield knows all the sight lines and the lay of the land.  (Think about when you drive someone else’s car.  It takes you a little while to get the hang of it.)  The visitors might be slightly less likely to get to a ball and so I am under-estimating what the true probability of catching the ball is.
What of the “catch” ratings for the gentlemen who patrol center field?  The list isn’t all too informative, but here it is in all its glory.
Top 5:

  1. Josh Hamilton: 0.003 outs above average per fly ball
  2. Coco Crisp: 0.003
  3. B.J. Upton: 0.003
  4. Jim Edmonds: 0.003
  5. Melky Cabrera: 0.003

Bottom 5:

  1. Hunter Pence: -0.004
  2. Gary Matthews, Jr.: -0.004
  3. Marlon Byrd: -0.005
  4. Bill Hall: -0.009
  5. Alfredo Amezaga: -0.011

Now, an outfielder also has a duty even to a fly ball that drops.  He must cut it off, before it becomes a double.  Extra base prevention is the often-overlooked piece of outfield defense.  Again, I controlled for expectations using the results from visiting fielders in the current ballpark.  In 2007, your top 5 were:

  1. Jerry Owens: 0.197 extra bases prevented above average
  2. Coco Crisp: 0.106
  3. Grady Sizemore: 0.103
  4. Carlos Beltran: 0.081
  5. Hunter Pence: 0.070

And the bottom 5:

  1. Nook Logan: -0.122
  2. Dave Roberts: -0.123
  3. David DeJesus: -0.175
  4. B.J. Upton: -0.182
  5. Nick Swisher: -0.210

Now, we come to the matter of whether “wheels” actually help a center fielder.  Speed is the one major attribute a player has that could theoretically help on both offense and defense.  Since I had the file handy, I imported the speed scores that I created for each of the center fielders in question.  These are scores based on offensive events (stolen bases, infield hits), so it should be interesting to see whether speed helps a player in the field.  The correlation between basepath speed score and range among center fielders was .311.  On extra base prevention (cutting the ball off), it was .296.  Not really all that overwhelming, but nothing to sneeze at. 
I probed a little deeper on this and found some outliers.  In my speed score system, a player who has a score of 1.00 or better is generally a pretty fast guy, and of the top four in range, all were above 1.00.  (Sizemore has a score of .70).  However, there were a few guys who seemed to outperform their speed.  David DeJesus was 6th in range one place ahead of Dave Roberts, still one of the fastest men in the league, but DeJesus had a speed score of .16 (just a bit above average).  Andruw and Jacque Jones both had speed scores slightly below average, and yet finished 9th and 10th in flyball range.  Then there were the disappointments at the bottom of the list.  Ichiro (speed score of 1.44) came in 15th (middle of the pack).  A few other noted speedsters (guys with speed scores greater than 1.00) were actually in the bottom half of the list including Wily Taveras (17th), Juan Pierre (19th), Hunter Pence (26th), and Chris Duffy (28th).  Kenny Lofton who was the fifth worst at getting to fly balls in center field had a speed score of .96.  Seems that a good set of wheels isn’t sufficient to playing a good center field.


5 Responses to He’s got wheels: Rating the center fielders

  1. Sky says:

    Pizza, have you made any ratings from your system available yet? Or just the leader tables in the few articles? Are you waiting until you actual rate all positions and work out the kinks to release any data? Thanks.

  2. ekogan says:

    Do you weigh the opposing fielders’ numbers by the games played or equally for each man?
    Almost half of the 81 games played by opposing CF in a stadium will be played by the divisional opponents, so a simple average of outs per ball by opponents will mostly compare a CF to his divisional rivals.

  3. Pizza Cutter says:

    That’s a good point. I should probably take that into account. I’m convinced that park effects need an overhaul, but I have yet to figure out a good way how.

  4. Pizza Cutter says:

    Sky, I haven’t yet dumped that data into the public domain and I won’t until I’ve got everything in order. But I will eventually. Probably here linked to a Google doc. Not sure yet.

  5. dan says:

    Try checking these numbers against Tango’s scouting reports…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: