M’s Defense And The Fielding Bible
You may have already seen Baker’s post about John Dewan ranking the Mariners as an above average defense in 2008, according to his Defensive Runs Saved metric introduced in the newly published Fielding Bible II.
I don’t have tons of time to walk everyone through why Dewan is wrong – and, I should point out that The Fielding Bible is a really good read with lots of great stuff, and that BIS has been instrumental in bringing advanced defensive measures to the forefront of baseball, so this isn’t in any way an attack on him or the the book – but there’s some problems with his Defensive Runs Saved metric.
But, for those of you who would prefer a short summary, here’s the basic problem – Dewan/James calculated that each play made in the OF is worth about .56 runs, while play made in the infield is about +.75 runs (numbers are from memory, I don’t have the book in front of me right now, but they’re close). I think that’s just demonstrably wrong. We know, thanks to linear weights, that the value of an out is about -0.27 runs, and the value of a single is +0.47 runs, so making a save on a play that would have otherwise gone for a single is worth ~+.71 runs. However, the run values of doubles and triples are .77 and 1.04, respectively, so turning potential extra base hits into outs is more valuable than turning singles into outs.
Obviously, a great majority of doubles and triples come on fly balls to the outfield, and because of that, there’s almost no way to realistically conclude that a play saved on the infield is worth more than a play saved in the outfield. But that’s the conclusion that Dewan/James came to, and part of why they have the Mariners rated as above average – infield defense is significantly more important in their system than outfield defense, and so by reducing the the value of the M’s weakest link (OF defense), it serves to make the M’s look better than they were.
There’s also positioning issues in how Dewan ended up calculating his +/- ratings, as can be seen best in his Chase Utley numbers. Because Utley shifts towards the first base bag to make up for Ryan Howard’s ridiculous lack of range, he gets to far more balls in the 1B/2B hole than most second baseman. Dewan doesn’t adjust for this, and gives Utley credit for a +32 rating to his left (compared to +6 straight on and +8 to his right). Because of how Dewan is measuring plays saved (based on normal positioning for all players across the league), Utley would look worse defensively by his system if the Phillies just got a better defensive first baseman and Utley shaded back towards the 2B bag to play a more normal second base. Giving Utley credit for Howard being awful isn’t really what we are trying to do.
There are some other issues (he’s not using context neutral numbers, James’ new Defensive Misplays is pretty subjective and all plays are weighted evenly), but those two are the big reasons why Dewan’s Defensive Runs Saved doesn’t line up with the other advanced defensive metrics. And, to me, they’re problems, not enhancements.
The UZR data that we publish at FanGraphs is based on the play by play data collected by Baseball Info Solutions. It’s the same data, but a somewhat different (in my opinion, better) methodology for converting that data into runs saved. As you know, UZR had the Mariners at -21.6 runs for 2008, thanks mostly to a -16.8 rating from their outfielders. This is, honestly, the more reliable number to use. It’s based on exactly the same data, and the discrepancies between the two systems lie mostly in some assumptions made by Dewan that I would consider fairly questionable.
So, don’t make too big a deal out of this, and don’t buy into the rhetoric of the M’s defensive upgrades only getting them from a 62 to 64 win team. Right now, this is a ~79 win team, and an improved outfield defense is a significant part of why the team will be a lot better this year than last.