Skewed Left: Spring St... (03/18)
the archives are now free.
21 comments have been left for this article. BP Comment Quick Links
| |
|
|
March 13, 2013 Skewed LeftSaberizing the Gold Gloves
So we won this weekend. At least I think we won. At least I think they told me we won. It was announced that the Gold Glove Awards will add a metric component to the traditional voting of major-league managers and coaches, a presumed victory for everyone who prefers the analytical and objective over the judgment of the human eye. So why no celebration in this virtual household, which stands for just that? First of all, the release didn’t give much information about the metric itself. Here’s a portion of the release from the Society for American Baseball Research:
In other words, they’re working on it. But this isn’t a critique of SABR’s motives, which are absolutely in the right place—taking a step toward greater likelihood of getting it “right” and spreading knowledge of defensive statistics. Nor is it really even about the uncertainty of what will come out of the SABR conclave. It’s about the certainty of the ugly process that will ensue. 1. There will be a fight when the metric comes out. Using data provided by Baseball Info Solutions, Mitchel Lichtman pioneered the Ultimate Zone Rating, which uses a fielder’s capacity inside and outside of a given zone to make plays. We at Baseball Prospectus use Fielding Runs Above Average as the defensive component of our value statistics, focused more on total plays made given conditions such as pitcher’s ground-ball rate, batter’s handedness, ballpark and base-out scenario. Both are exhaustively researched and justified, and yet the choice of using one or the other in Gold Glove voting would lead to totally different results. At the positions where UZR is calculated and available at Fangraphs.com, battery not included, here is the breakdown of the top players from 2012 who played the whole season in the same league.
Of 14 positions, UZR and FRAA agree on the Gold Glover in exactly three of them, which is a huge issue for the mainstream public appeal of this vote. (Not even mentioning what would happen in the discourse if a part-time player like Dyson or Bourjos were ranked first at a position.) It will be a very hard sell for the analysts out there to peddle this idea when it creates division within the sabermetric media and surely within SABR’s membership itself. But that will only be the first step. 2. There will be a fight when the first vote comes out. But instead of uniting them in some awkward arranged marriage, this has the potential to pit the traditionalists against the statistical analysts. When an award comes out where the vote doesn’t match the SDI, that will become a binary-outcome referendum on both parties. The coaches got it wrong or the numbers got it wrong. One of them had to get it wrong, and dammit, we need to know who it was. Instead of a celebration of the winner, it’s an examination of the process, which will become very tiresome very quickly. We don’t need any more “WAR, What Is It Good For” columns. Even if SDI has never been in a song lyric, let’s not take that chance over this. There are plenty of more worthy fights for the importance of analytical thinking. 3. We’ll argue over whether stats should be applied directly to more awards. When the statistical community puts its stamp on this award, it has to be prepared to stand behind it. The stats say Soriano, always thought to be a poor defender, was the most accomplished left fielder in the league last year. The stats say a part-time player was the most accomplished center fielder in the league last year. Is a statistic that research says can take three years to stabilize really the one we want imprinted on a single-season award? There’s an argument to be made that adding a WARP/WAR component to the Baseball Writers Association of America’s MVP awards—or one of the lesser-known MVP-type honors, or maybe even the Hall of Fame—would be a better step. But SABR doesn’t control any of the awards, and the BBWAA has not been looking to cede any control over its awards (disclosure: the author is a member of both organizations). So this isn’t a knock on SABR, which is doing what it can to advance the discussion, just an unfortunate case of who came calling and had some ground to make a deal. We hate the process of most awards, yet we haven’t come up with a much better one either here or in the case of the BBWAA awards. If coaches have proven to be the worst voters of any electorate—and it’s really neck-and-neck between them and fans—then change the electorate. Have the people in front offices paid to assess value hand out awards for defensive value. Perfecting the Gold Glove and other awards is a noble pursuit. This is a small step toward that goal that might be missed in the very predictable reactions to every part of the process.
Zachary Levine is an author of Baseball Prospectus. Follow @zacharylevine
|
Pre-Season Positional Rankings: Top 40 Fanta...
So from the defensive metric numbers from last year we are to conclude that Alfonso Soriano is a better defensive player than Carlos Gonzalez. And not just a little better--a whole lot better.
What a complete joke.
While there is some merit to what you all do with the defensive metrics. When a result like the above comes out--you need to take a serious look at what you are doing.
Or possibly you need to re-examine your own prejudices as well? It's always possible your eye is not the ultimate best way of determining defensive competence.
So what would you do when the metrics don't agree with your subjective observations? Veto the numbers? Have a subjective fudge factor? Invalidate all metrics that have outliers and idiosyncracies?
Nobody believes Darin Erstad was a .355 hitter, so while there's some merit in what people are trying to do with offensive metrics, when a result like that comes out someone needs to take a serious look at what they're doing.
The Erstad example illustrates the problem here.
Batting average is a very simple stat that measures exactly one thing. What it says about Erstad's merits as a player is up for debate, but it is a factual statement that for one season, 240 of his 676 of his at-bats ended with a hit.
Advanced fielding metrics incorporate a lot of things and reflect various weightings and interpretations. I can't compute UZR without a spreadsheet -- come to think of it, I can't compute it AT ALL because it includes proprietary data.
Bingo. If just one of these two methodologies said that Soriano was the best LF, I would have just thought it was some bizarre systemic anomaly and ignored it. But when both of them say it I have wonder what caused this to happen - I mean Alfonso Soriano...really? The problem is that I can't see the numbers, I can't see if there is a flaw they have in common, or if they are really meaningful. I just have to trust that these things work. I don't trust FRAA, so I am dubious that their inclusion in the Gold Glove (or Platinum Glove, whatever that is) voting will be useful.
That reliance on trust is the ultimate problem. And it ties in with Russell Carleton's piece from Monday. If Rafael Palmeiro 1999 is a damning indictment of the current voting structure, shouldn't we consider Alfonso Soriano 2012 a serious shortcoming for FRAA and UZR?
I don't think so. Palmaeiro won a popularity poll which he shouldn't have even qualified for; Soriano is based on statistical evidence, which carries a degree of objectivity.
Soraiano was the leader in those two statistical measures,that is just a fact. The question is how worthwhile a fact that is, which depends on how much faith you have in those respective statistics as a measure of defensive value.
By the way, the system in these comments needs fixing, Just because bluesman98's comment received a certain number of negative votes it was grayed out. Given that it was only a -4, that seems extreme to me, but worse, since it started the thread, graying it out took this whole thread with it. That is asinine, as this is a fairly interesting discussion. I voted bluesman back up, in an attempt to restore the thread. Please fix this.
Barry, the "hidden" comment floor is -4, but if it makes you feel any better, hidden comments are quite widely read at BP.
Or you can read the entire article:
"One-year defensive metrics are notoriously unstable—Alfonso Soriano’s 7.9 FRAA in 2012 came after a -6.4 in 2011 and a -8.0 in 2012."
...
"The stats say Soriano, always thought to be a poor defender, was the most accomplished left fielder in the league last year. The stats say a part-time player was the most accomplished center fielder in the league last year.
Is a statistic that research says can take three years to stabilize really the one we want imprinted on a single-season award?"