Gold Gloves and Coco Crisp's Terrible 2014 Defense

By Jason Wojciechowski on October 23, 2014 at 9:37 PM

Aaron Gleeman has helpfully compiled the list of Gold Glove "finalists" -- I still haven't figured out what "finalist" means in this context except that we know the list at each position contains the eventual winner. Just one A's player appears, Josh Donaldson, though Yoenis Cespedes is also a left field finalist and most of his defensive work (i.e. most of his splendiferous throws) was done for the A's.

Nor do the A's really have anyone who was slighted. Sam Fuld is a very good outfielder, but his 517 innings in center were the most he played at one position. (162 games x 9 innings = 1,458 innings, to give you a sense of scale.) Eric Sogard is above-average at second base but only played half a year of innings there. Craig Gentry is a part-timer. Josh Reddick missed 50 games.

And Coco Crisp ... well, that's an interesting one. He's been known as a good defender for a while, or at least as a good flycatcher with a limp noodle arm, but his numbers took an absolute dive this season: -13 FRAA, -17 DRS, -14 UZR, -7 Total Zone. Those aren't the be-all end-all, but when FRAA and Total Zone agree with the ball-in-play-based metrics, I tend to give the whole package a little more credence.

It's not hard to see where the drop comes from: Crisp caught way way fewer balls this year than he has in the past:

Year Putouts + Assists per game
2010 2.5
2011 2.4
2012 2.4
2013 2.8
2014 1.8


The Inside Edge numbers, which break chances down into probability bands (0%, 1-10%, 10-40%, 40-60%, 60-90%, 90-100%), show Crisp declining in every band. The numbers only go back to 2012, but, most precipitously, and also most intriguingly, check out the 1-10% band, which Inside Edge calls "Remote":

Year "Remote" plays made "Remote" chances %
2012 2 6 33%
2013 1 5 20%
2014 1 19 5.3%

Okay so there are some things to think about here. One of them is scorer bias. In 2013, Crisp had 47 chances labeled "impossible" and in basically the same number of innings in 2014, that declined to 30. So it's entirely possible that Crisp had something like 10 borderline balls switch ratings.

On the other hand! Inside Edge and BIS (which provides the data that forms the basis of DRS and UZR) presumably have different people watching/classifying the balls in play, and the drop in balls caught registers in the BIS-based metrics, so that might be an indication that it's not just error. (Unless, and let's here take some lessons from Colin Wyers, the classification systems are so alike that error is likely to be replicated across those systems.)

So I don't know. Coco caught fewer balls this year. That part is indubitable because we register one entirely objective statistic: putouts. He caught balls. He didn't catch balls. (That, incidentally, forms a large basis of FRAA.) On the other hand, how many air balls did the A's allow?

Year TBF In-play % In-play Air% Air balls IFFB% IFFB OFFB
2013 6069 70% 4248 59.9% 2544 16% 407 2137
2014 5971 69% 4120 53.8% 2215 14% 310 1905

(These stats are from Baseball Reference. If I knew off the top of my head where to get the raw stats, I'd just use those. Instead I have to work backward from the percentages.)

That's not an insubstantial difference. It's about 1.4 fewer outfield chances per game, which means 140 fewer chances to Crisp and his outfield mates in his 900 innings in center field. Breaking down the league range factors (there's probably a better way to do this), about 40 percent of those chances would go to center, so Crisp probably saw something like 55 fewer chances over the course of the year than he did in 2013.

Now, the statistics, especially the ball-in-play ones, account for this in the sense that they're only dealing in the first place with balls in Crisp's zone. He's not penalized for not catching those 55 balls that weren't hit to him. But in decreasing the sample, you're increasing the possibility that Crisp got bad luck on a weirdly skewed sample of difficult baseballs to catch. Which means that it becomes harder, despite the magnitude of the apparent defensive cratering we saw this year, at least in the numbers, to say "Crisp isn't what he was on defense" (and to therefore bemoan his contract).

There's also good reason to think Crisp's neck injury, which you'll recall began all the way back in May, had something to do with his defense. (Note as well his drop in steal attempts compared to 2011 and 2012, though double note that his rate was about the same as 2013.) On the other hand, he's always hurt. Always has been and, more importantly, always will be. If his injury was the reason he was -15 in the field this year, then we should probably count on that being the case until he hangs up the spikes unless there's some reason to think this injury was somehow special.

Maybe it was! I don't really know. It's all heuristics and guesswork and back-of-the-envelope math. There's only so far you can go with those. How far? Right about ... here.