Captain Kirk's Glove

One of the more controversial prospects in the Mets system is Kirk Nieuwenhuis.  For those of you unfamiliar with Nieuwenhuis, he's a Center Fielder out of Azusa Pacific University, a small NAIA school.  Despite the low level of amateur competition, Nieuwenhuis's toolsy profile made him intriguing enough for the Mets to draft him 100th overall in the 2008 amateur draft.  He made his debut that year for Brooklyn Cyclones, the Mets Short Season-A affiliate, and managed a respectable .329 wOBA.  Moving up to the Hi-A Florida State League in 2009, one of the most pitcher friendly leagues in the minors, Nieuwenhuis broke out in a big way with a .359 wOBA (actually .374 according to his statcorner page, but since they don't have data for his 2008 season, I'm crunching the numbers myself using the method described by The Book) on the season and a torrid .396 wOBA from the beginning of July through his late season promotion to Double-A Binghamton. 

Though many consider Nieuwenhuis one of the better prospects in the Mets system, he was nowhere to be found on Baseball America's Top 10 Mets Prospects list this year.  To make things even more confusing, BA also named him the best defensive outfielder in the system.  Considering his offensive breakout, one would think that superior defense would earn him a Top-10 spot for sure.  Not so fast though, Sean Smith's TotalZone ratings--the most advanced defensive metric we have for minor leaguers--had Nieuwenhuis pegged at -12 in 106 games, which is pretty awful.  So what gives?  Is Nieuwenhuis a superior defender as suggested by BA, despite his omission from their Top 10 list?  Or his he inferior, as suggested by his TotalZone? 

First things first, a bit about TotalZone.  TotalZone is based largely on the three batted ball types--Fly Ball, Line Drive, and Ground Ball.  For Minor Leaguers, this data is provided by Jeff Sackman, owner of Minor League Splits.  This is a tremendous resource for minor league analysis, but it instantly raises a question:  With about six times as many Minor League stadiums in the U.S. as Major League stadiums, none of them nearly as well equipped or funded as a Major League stadium, how can we be sure these terms are defined consistently?  A ground ball is a pretty easy thing to name when you see it, but there's a lot of gray area when you talk about fly balls and line drives.  Fortunately, Sackman provides the data we need to check the results.  Consider these splits for the five hitters with the most plate appearances from the St. Lucie Roster:

 

Kirk Nieuwenhuis                                                                    Carlos Guzman

	GB%	LD%	FB%
Home	57.2%	7.2%	34.5%
Away	53.0%	18.5%	28.6%

	GB%	LD%	FB%
Home	43.8%	10.3%	45.9%
Away	43.2%	22.9%	33.9%

 

 

 

Brahiam Maldonado                                                                   Reese Havens

	GB%	LD%	FB%
Home	44.9%	6.8%	48.3%
Away	42.8%	18.1%	39.1%

	GB%	LD%	FB%
Home	38.2%	9.7%	52.1%
Away	40.7%	16.6%	42.1%

Francisco Pena

	GB%	LD%	FB%
Home	45.7%	7.3%	46.3%
Away	44.9%	14.3%	40.8%

 

Perhaps more relevant for Nieuwenhuis's defense, here are the same splits for the top five pitchers on the staff by Batters Faced:

 

Scott Shaw                                                                          Angel Calero

	GB%	LD%	FB%
Home	39.7%	7.4%	43.7%
Away	34.6%	19.8%	37.6%

	GB%	LD%	FB%
Home	32.8%	8.5%	50.0%
Away	44.4%	18.8%	33.1%

 

 

 

Nicolas Waechter                                                                   Scott Moviel

	GB%	LD%	FB%
Home	44.8%	8.0%	44.5%
Away	51.9%	10.4%	32.1%

	GB%	LD%	FB%
Home	41.5%	6.9%	44.6%
Away	43.3%	17.5%	22.7%

 

 

 

Timothy Stronach

	GB%	LD%	FB%
Home	50.5%	9.3%	32.0%
Away	44.3%	19.8%	32.1%

 

I'm not sure who the St. Lucie Mets have tracking Gameday data at Tradition Field, or what the term "Line Drive" ever did to them, but its apparently not a very liberally used phrase in Port St. Luce.  Every one of these guys had a higher LD% on the road than at home, and for seven the difference was greater than 10%.  The only player listed here, pitcher or hitter, who had a LD% greater than 10% at home was Carlos Guzman at 10.3%.  This just doesn't gel with what we know about park factors and the natural variance you expect to see in a stat like LD%.

How might we expect this to effect Nieuwenhuis's TotalZone?  Well, it means that there are probably lots of balls getting hit to him that would be called Line Drives in most ballparks but getting tagged Fly Balls at Tradition Field.  That means Kirk is being expected to field many of these "actual" Line Drives as if they were Fly Balls.  But we know that Line Drives are much less likely to be converted into outs than Fly Balls, so this is unduly hurting his score.  If more of these Line Drives were being called Line Drives, he would be expected to field less of them.  His failure to field the ones he missed would hurt his score less and his ability to field the ones he did would benefit his score more.  In addition, the balls being mistaken for Fly Balls aren't just any arbitrary Line Drives.  They're likely more of the "fliner" variety; that is, balls that bear more resemblance to Fly Balls than the average Line Drive.  That means the set of remaining Line Drives are also more difficult to field than we expect the average Line Drive to be, again hurting Nieuwenhuis's score. 

To put all this another way, we usually expect BABIP to be about .730 on Line Drives and .150 on Fly Balls (this is true in the Majors, those numbers might be different in the Minors, but we can still use them here for the sake of argument).  Because of the classification discrepancy at Tradition Field, these expected rates are probably both quite a bit higher if we're using the house definitions of Fly Ball and Line Drive.  Nieuwenhuis should have been expected to field fewer balls called "Line Drives" and fewer balls called "Fly Balls".  This is an interesting quirk of looking at BABIP by batted ball rate.  Changing a group of plays from one classification to another will often cause the BABIP of both rates to move in the same direction, not opposite ones, but that's a subject for another time.  Because all this only applies to his home games, we should also expect that Kirk's TotalZone rating would be much lower at home than on the road, and indeed it is.  Its listed as -11 at home and -1 on the road.

While this tells us that Nieuwenhuis probably wasn't nearly as bad as his -12 TotalZone suggests, it still doesn't tell us how much better he is.  To figure that out, we'd have to figure out exactly how much we should be regressing LD% and FB% for games at Tradition Field.  We can guess that he's at least a decent fielder, but beyond that, we don't really know.  Here are a few more things to consider when trying to resolve TotalZone with Baseball America's suggestion that he's the top defensive outfielder in the system.  From Sean Smith's article about translating TotalZone to the minors, a few choice paragraphs:

Are these ratings useful?

For the ratings to be useful, they need to correlate from one level to another. If we knew a player had a +15 rating one year, that would be of no use unless it told us that he was likely to continue to post good ratings in future seasons, at higher levels of the minors. Preliminary investigation shows that these ratings are more useful for infielders than outfielders. The correlation is much lower for outfielders, though at least they are (usually) positive. I’ve found that you can usually get a correlation of 0.50, meaning you regress 50 percent to the mean, at about 350 chances for infielders. This represents less than a full season of chances for second, third and short. For outfielders, you need about two full seasons of data to regress 50 percent, or about 1,000 chances.

How do they translate from one level to the next?

For infielders, players perform worse, relative to the league, as they move up in levels. In other words, major league third basemen are better fielders than Triple-A third basemen, who are in turn better than players in the lower minors. For outfielders, the picture is less clear. For center fielders, the relative performance is relatively flat. Give or take a run or two, the average center fielder in Low A is about as good as the average major league center fielder. For corner outfielders, it appears that the quality of fielding, by looking at players who move up in levels, is lower as you move up in level, with major leaguers being the worst!

This was a bit hard for me to believe. There are some reasons that it could be possible—players lose speed relatively quickly, and outfield range is highly dependent on speed. Also, the lower levels of the minors use the DH in every game, which probably cuts down on the Adam Dunn and Pat Burrell types in minor league outfields. Another thing to consider is that players like Dunn were not the same lumbering plodders in the minors. At age 21, he was likely a good bit lighter on the scale and faster in the field. Major league outfielders, especially at the corners, are selected more for their bats than their gloves, and while having a lot of bulk muscle helps you hit for power, it does not help you chase down fly balls.

There are two important takeaways here.  First, that a single season of defensive data isn't really enough to accurately apply TotalZone to outfield defense. Second, whatever Nieuwenhuis's "actual" defensive ability currently is, we would expect it to be pretty much the same regardless of level.  If he's an average Center Fielder in the Florida State League, he'd be about an average Center Fielder in the Majors.  This doesn't tell us how good he'll be a few years from now--its possible that he fills out and loses some speed and mobility, and also possible that he improves his technique and route running. But once we get a large enough sample size of accurate data, we should have a pretty good idea of how good of a defensive outfielder Nieuwenhuis is right now.

The other thing to consider is that the Mets just aren't that deep in outfielders.  In the Full Season leagues, aside from Nieuwenhuis, the only major prospect of note this year was Fernando Martinez, who only played part of the year in the minors before getting called up and eventually shut down due to injury.  There were some other players who played a corner outfield position pretty consistently, such as Carlos Guzman and Lucas Duda, and there are a few  interesting defensive talents in the lower levels.  None are clearly superior to Nieuwenhuis defensively though.  So really, by saying he's the best defensive outfielder in the system, all BA is saying is that Nieuwenhuis is a better defensive outfielder than Martinez.  Martinez is solid fielder in his own right, but hardly spectacular, so saying Nieuwenhuis is the best defensive outfielder in the system really isn't saying an awful lot.

So we started with two apparently opposing views on Nieuwenhuis's defense, the old "Stats vs. scouts myth" as James likes to call it.  As it turns out, put in proper context, the two perspectives line up much better than it appears at first glance.  Calling someone the best defensive outfielder in a system weak on Center Fielders isn't a very strong statement, and because of circumstances beyond its developer's control, TotalZone's unfavorable rating of Captain Kirk probably isn't accurate either.  Its a good example of how using Minor League batted ball rates can be misleading.  In some cases, there are workarounds for this problem.  Rather than using HR/FB for minor leaguers, try using HR/BIA, a statistic that combines the "Fly Ball" and "Line Drive" classifications into one far less ambiguous group.  Instead of tRA, especially for quick-checks, FIP is probably a better go-to DIPS type pitching statistic, even if the method is a bit less rich.  Unfortunately, there is no alternative like this for defensive ratings.  TotalZone is still the best method of quantifying minor league defense I've seen so far, and I encourage its use, but its important to either do the due diligence on it or take it with a grain of salt.