clock menu more-arrow no yes

Filed under:

Andrelton Simmons probably should have won the Gold Glove

New, 23 comments

The Gold Glove voters did to Andrelton Simmons what Simmons usually does to hitters: deny them something they'd otherwise probably get.

Lance Iversen-USA TODAY Sports

Last night, sports equipment manufacturing company Rawlings announced its set of 2015 Gold Glove winners. An ostensibly obvious candidate for the award, Andrelton Simmons, was not among the list of recipients. Instead, San Francisco Giants shortstop Brandon Crawford was selected in his stead.

If you're reading this, you probably care at least a little bit about Simmons getting continued recognition for his defensive work, and are possibly incensed (to some degree) about last night's snub in this regard. Personally, I'm not particularly "offended" or aggrieved, but I do think this is another case where the "voters," nameless/faceless as they may be, are denigrating the meaning and prestige of the award by whiffing on who gets it. So, the below is a brief synopsis on why I think Andrelton Simmons should have (again) won the National League Gold Glove for Shortstops.

The first key assumption that I'm making, that I'm not sure the Gold Glove voters are, is that the award should be for defense and nothing else. Playtime considerations should factor in (a player who is consistently good but not great over a longer inning count could warrant consideration over a defensive wizard who only managed to play half a season at a given position), but offensive production, team's success, team/name recognition, and so on, are all irrelevant. I suspect that as far as these latter criteria go, that wasn't really the case with the voters. (Cases in point: Eric Hosmer, who was basically super-average this season, and has been below average for his career; Alcides Escobar, who won because chaos reigns, I guess.)

With that said, there are basically two key inputs into a player's defensive value:

  1. How hard is the player's position to play (relative to other positions)?
  2. How well does the player play that position (relative to his peers at the position)?
As the Gold Gloves are on a per-position basis, the first question is irrelevant. Fortunately, there are a few tools to help us gauge the answer to the latter. In fact, even the current Gold Glove process admits that there's something to be said for quantifiable objectivity in the selection process: the awards process uses something called the SABR Defensive Index (SDI) to inject something measurable into the process.

So, what in the world is SDI? Well, see if you can make effective heads or tails of this:
The SDI is built from two types of defensive metrics — those that come from batted ball location-based data, and those which originate from the play-by-play records of games. We gave more weight (70%) to the batted ball location-based metrics, which evaluate the degree to which a fielder makes plays in specific zones on the diamond. The player's performance is measured in comparison to his peers. The play-by-play based metrics (30% of the SDI) are important in that they approach defensive measurement from an alternative vantage point — a more generalized approach that estimates the number of batted balls hit into a fielder's area.
For those curious (I'll get into this a bit more later), that first play-by-play group includes Defensive Runs Saved (DRS) and Ultimate Zone Rating (UZR), both available on Fangraphs (UZR is Fangraphs' semi-preferred defensive metric, and goes into their calculations of player value), as well as something called Runs Effectively Defended (RED). The first two items have methodologies which make intuitive sense (to me, anyway) and are fairly public in both process and results, so they can be reviewed and critiqued. The latter, well... it doesn't appear to be publicly available or even commonly used, so I have no idea what it is.

The second group, meanwhile, includes Total Zone Rating (TZ) and something called Defensive Regression Analysis (DRA). The reason why this group differs from the first group, as best I can tell, is because the first group uses what I guess would currently be considered the best practice of giving each batted ball a run value (based on how many runs that batted ball would result in, on average, if it were not caught or converted into an out) and then assigning fielders part of that run value (or taking it away from them) based on how many other fielders at the same position would/would not make that play. Meanwhile, this group uses generic play-by-play data to assign value. The difference is basically that the first group looks at each individual batted ball as a point of giving/taking away credit, while the latter looks at game logs and how many total balls were fielded effectively (or flubbed) by each position in the course of a game/season/etc. in order to assign value. I'm partial to the former because the latter seems to be subject to sample size issues that may not shake out in the course of a single season, and it appears that the creators of SDI are as well, since they give the first group of metrics the bulk of the weight.

(Side note about Total Zone: Total Zone is a pretty tricky thing, because as best I can tell, what it measures changes over time based on the quality of defensive data available. I'm not entirely sure how the methodology of Total Zone currently differs from DRS or UZR, but it must do so, as their results are pretty different from DRS/UZR results in current years. From what I can tell, I'm not sure that I actually agree with SABR that Total Zone is a true game-log-based system rather than a batted ball system, as it seems pretty batted ball-based to me, but if it isn't a batted ball-based system, that seems like a mark against it rather than for it.)

If you follow my logic that scoring each individual batted ball makes more sense than taking a broader, more watered-down view of defensive contributions, then you'll probably agree that it's not clear why the SDI mixes these "superior" measures with some "inferior" measures, given that they all attempt to do the same thing. If you had five weather forecasts and you knew that two of them were way worse than the other ones, would you still take them into account when trying to make decisions?

Anyway, we now know that Simmons did not win the 2015 Gold Glove based on the following:

  • 25% based on Brandon Crawford outpacing him in the SDI
  • 75% based on ???
Since I can't directly speak to the 75% portion, I'll try to make an objective case for Simmons using the publicly-available metrics baked into the SDI.

DRS

Defensive Runs Saved is one of the two most commonly-cited advanced fielding metrics. By DRS, there were three NL shortstops in the upper tier, defensively: Andrelton Simmons (25 DRS runs above average), Nick Ahmed (20), and Brandon Crawford (20). Simmons played more innings at short than either Ahmed or Crawford, but if you adjust their scores to put them on an even innings basis, Simmons still has a slight edge (26.4 to Ahmed's 25.9), while Crawford continues to lag behind (22.5). All of these scores are really good, but Simmons is meaningfully ahead of Crawford via this measure.

UZR

Simmons also tops the leaderboard in UZR, with 17.3 runs above average, trailed by Adeiny Hechavarria (15.8), Ahmed (11.3) and Crawford (10.9). When adjusting for innings played, Hechavarria actually takes a slight lead over Simmons (17.7 over 150 games for Hech versus 17.5 for Simmons), but Crawford (11.9) is again below Ahmed (16.1) and trails Simmons by a noticeable amount.

Across the two most commonly-used defensive metrics, there's not much justification for Crawford winning the award: he's behind Simmons by any formulation of UZR or DRS, and trails at least one other shortstop as well.

(Minor note: while the UZR/150 may suggest that Hechavarria should've been in serious consideration for the award, DRS was not as much of a fan [+9], and given that he's been a career negative defender by UZR, it's very possible that had he played the 159 additional innings to catch up to Simmons' innings total, he would have decreased his UZR rather than kept it going at the same rate he had accrued through 2015.)

Total Zone

I was not able to find a single-season Total Zone leaderboard for all players, but this is one area where Crawford has a definite edge. He's the only shortstop among the top 10, with +19 runs above average. Meanwhile, Simmons, Ahmed, and Hechavarria are at 15, 13, and 14, respectively.

Summing Up

So, here's my question. If we exclude Ahmed and Hechavarria and make this a horse race between Simmons and Crawford, here's what we know:

  • Simmons had about 17% more DRS, even on a pro-rated basis, than Crawford;
  • Simmons had 47% more UZR, even on a pro-rated basis, than Crawford; and
  • Crawford had, at best, a 36% advantage over Simmons in Total Zone.
We also don't know anything about their DRA and RED scores. But, we do know that the metric in which Crawford led is weighted less heavily within SDI, and also that Crawford has about a 20% lead in SDI, which means that Crawford's lead in DRA and/or RED over Simmons must have been huge to result in the final SDI standings that were published on the SDI website.

is that possible? Definitely. But I have a hard time envisioning a defensive system that gave Crawford such a large lead over Simmons. I dummied up a quick calculation that weighted DRS, UZR, and RED as about 23% of the metric each (a third of 70%, and TZ/DRA as 15% each), plugged in their values, and calculated a weighted average. Even assuming that Simmons is exactly average by RED and DRA (+0 runs), his score is 12.5, just slightly off of the 12.9 he has in the SDI rankings. Meanwhile, Crawford is just at +10.9, as compared to +15.4 in the SDI. Crawford needs to average about +12 runs in both RED and DRA while Simmons averages about +1 in each of these metrics in order to have them meet their SDI-reported totals. Given that, I feel comfortable saying that any statistic where Crawford has a +11-run lead over Simmons (that's over a full win just on defense) at this point in time is potentially suspect. I calculate that if Crawford is really +12 in both RED and DRA, he needs to have a 70% advantage over Simmons in these categories to still come in ahead of him by SDI, which also stretches credibility a bit.

In short, I'm pretty puzzled, and I think the existing publicly-available data make a stronger case for Simmons (or even Ahmed) over Crawford. I can't speak for the remaining 75%, in which Crawford's bat (117 wRC+) and team recognition (the Giants were non-terrible) may have given him some added punch. But even that aside, it still looks to me that Simmons was snubbed, potentially on the basis of metrics too currently opaque to feel much confidence in.

Overall, with a few exceptions, I thought the Gold Glove recipients made sense this year. It's just a shame that for some reason, one of the game's premier defensive players was somehow omitted from the honoree roll. Just like the Braves' 2015 season, well, maybe next year?