Monday, March 10, 2014

New fielding metrics provide surprising, positive results for Dodgers' fielders

Fielding metrics have advanced a lot in the last few years. Beyond simple fielding percentage and range factor, today's metrics are based on comparing how many plays a fielder made to how many an average fielder would have made given the same exact chances.

Based on the speed, angle and trajectory of the ball, they can say, for example, 75 percent of Major League second basemen make a particular play. So if you make it, you get credit for an extra quarter play, but if you don't, you get docked three quarters. That's the basic idea and then systems enhance it in different ways.

But most of the low-level play-by-play data has not been available, just a high level summaries (like the annual total per player per position). Well now, at FanGraphs, you can get Inside Edge data with just enough bucketing of results to actually be helpful.

The data say per player, per position, per year, how many chances did they convert on each level of play difficulty. So you can see if a player excels at making the routine play consistently or has trouble with that but can make some athletic plays that few players would make. The caveat is, 80 percent of plays are routine. That not only dominates any overall evaluation, it means you have few legitimate chances to make amazing plays.

For example, if you only get 10 chances a year at web gem-type stuff, can we really say anything meaningful based on the fact that you converted two instead of one of those 10 chances? But it's kind of like field goal kickers -- you can see one is 20-for-20 inside 30 yards, 8-for-10 between 30 and 50, and 1-for-4 at 50-plus yards, you get a sense of what kind of kicker you have.

Anyway, it's fun to look at and gives you some kind of insight, so let's look at how this year's Dodgers did last year. I rate each player's performance at the position for the type of play compared to the rest of the league at that position for that type of play, via the color coding.


Name
Pos
Inn
Remote
Unlikely
Even
Likely
Certain
Scott Van Slyke
LF
211
0/0
0/3
1/1
2/2
37/37
Matt Kemp
CF
576
1/3
0/5
3/4
3/5
126/126
Andre Ethier
CF
645
0/3
0/2
1/3
3/4
128/128
Andre Ethier
RF
443
0/3
0/2
1/3
3/3
101/101
Carl Crawford
LF
835
1/4
2/6
4/5
3/4
155/156
Yasiel Puig
RF
773
0/3
5/10
3/4
3/3
137/138
Juan Uribe
3B
900
0/7
3/5
9/12
17/19
249/254
Adrian Gonzalez
1B
1291
0/6
1/3
3/9
13/15
222/230
Hanley Ramirez
SS
651
0/2
1/6
4/9
20/23
221/231
A.J. Ellis
C
972
0/6
12/30
1/2
6/6
27/29
Dee Gordon
SS
206
0/5
1/3
1/1
3/5
74/80
Tim Federowicz
C
374
1/3
4/14
1/2
3/4
11/12

Dee Gordon has not done a good job at SS. A really bad job, actually. Juan Uribe and Yasiel Puig are legitimate plus defenders. Scott Van Slyke is actually quite reliable, although he will not make the highlight reel. Adrian Gonzalez does surprisingly poor here (compared to his reputation) while Hanley Ramirez does profile as just a tick below-average (certainly not bad). As most suspected, Andre Ethier was a serviceable backup CF in a pinch and OK in right. Carl Crawford can still go get 'em. Catchers don't get to make that many plays (outside of catching the pitch) but A.J. Ellis and Tim Federowicz are both OK or a tick below. Matt Kemp is not killing you out there, although he should have made a couple more plays than he did.

By the way, you know who's really good? Nick Punto, that's who.  Here's how he looked at shortstop at age 35 last year:

Nick Punto
SS
309
0/2
0/3
3/4
5/5
103/105


It's a cool new data stream, and will actually be updated nightly during the season. Although given the paucity of legitimate opportunities to do something spectacular over a full season, I don't know how much I'll look at in-season numbers. But at least it starts to shine a bit of a light in the black box of modern fielding metrics, which is always a good thing.

Photo credit: SD Dirk, Flickr

No comments:

Post a Comment

Post a Comment