Why Has Defence Been Undervalued?

This offseason has seen a sea change in most teams' approach to roster construction.  While there's been an outlier or two (Ibanez getting signed to play in the field really counts as two outliers), we're seeing front offices wise up to players' fielding value. Adam Dunn, one of the better sluggers in the game, is still unemployed at the end of January, and even when he signs it'll be for far below expected monies. In December the Mariners traded their closer in order to totally revamp their outfield, bringing two true centrefielders in to complement Ichiro in Safeco Field. All in all, teams are getting better at both evaluating how good a player is at defence and how that impacts their value to the franchise.

So, why was player defensive ability ever undervalued? What was it undervalued relative to?

The second question is by far the more straightforward, so it'll get answered first: Hitting. Combine the explosion in offence baseball has seen over the last 20 or so years with some fairly robust offensive measurement systems, and it's easy to see why. Teams have had a much better idea of how much a bat is worth, and built their rosters around hitting. Defence was at best a secondary concern - often as an excuse for offensive shortcomings. All glove, no bat players had some respect, but in a just world Adam Everett would have been earned a much larger sum of money over the course of his career.

Hitters are easy to compare to one another, too. They all have the same job, they all step up to the same plate, they are all judged in the same way. From batting average to wOBA, every hitter has been on the same scale. It is simplicity in itself to determine which players are good at hitting, which are bad, and the magnitude of the difference - whether that be in wins or points of batting average. And so we know that there's a massive gap between the hitting ability of a Joey Gathright and an Albert Pujols - we can see that Pujols wipes the floor with Gathright (don't feel bad, Joey, he kicks everyone else's ass too), and the numbers can tell us by how much. In this case it's by about 90 runs. Hitters have such a wide spread of talent that it's obvious glovework is a minor consideration compared to batting talent. Right?

Well, defence is nowhere near as easy to deal with. The advanced statistics which deal with it are in their infancy, and those which everyday fans are familiar with offer no obvious scale like batting statistics do. Errors? For any major league player worth his salt, fielding percentage should be close to one, making it difficult to see any real difference between players via a cursory look, and even when you dig deeper you're not going to be comparing players to all of their peers - you're comparing specialised fielders to one another. For a long time, there has been no reasonable scale to measure gloves.

Unless you've been living in a cave*, you'll know about UZR, PMR, +/-, and assorted other tools developed in order to obtain more accurate defensive measurements. While they're still not very good (or very predictive) compared to the stats we have to gauge offense, they're more advanced than fielding percentage and the like by an absolutely laughable margin. These are more than enough to give us a feel for the difference in defensive ability from one player to the next. Or, more accurately, between one player and the next provided they both play the same position. Oops.

The lack of a real scale is an obstacle, but not the only one. The real culprits here are the positions.

Yes, that is a bizarre assertion to make. Bear with me.

Imagine if each position in the batting order had to do something slightly different. The first spot might have a more difficult job - maybe they only get credit for triples, or some other arbitrary event. The order slowly gets easier and easier to bat in until the ninth spot where they get credited with runs and general excellence for standing around and looking sheepish. Do you think anyone would try to compare hitting talent across all of baseball? Or do you think that we'd just compare 3-hitters to 3-hitters, leadoff men to leadoff men?

It's a fairly clumsy analogy, but I'm sure you can see what I'm getting at. The very best hitters in the above scenario would play in the most challenging slot, and they'd get compared to one another. This will force the perceived difference in hitting talent to be smaller - if the best are competing against the best, there'll be far less separation. This will naturally have an influence on perception of hitting talent, because the spread in value shrinks by a huge amount when you have specialisation in tasks varying from the easy to the absurdly difficult.

Which is exactly what's happened with fielding, for those who haven't quite figured out what on earth my last two paragraphs have been about. We compare shortstops to shortstops and corner outfielders to corner outfielders, and in doing so neglect to properly account for the difference between a shortstop and a corner outfielder. It's an easy step from there to imagining that defence cannot account for very much - it gets measured on small scales while hitting is on a far bigger, universal one. This is why public perception of the impact of defence is so dismissive. They can see a huge gap between the best hitters and the worst, but gloves are much harder to distinguish - they're already segregated into talent groups. Compared to Manny Ramirez, Derek Jeter is a genius with the glove. Pitted against Cesar Izturis, he isn't. Compared to Izturis, Jeter is an outstanding hitter. Look at his bat against Manny's and you have a very different story.

Having positions which require more skill than others really screws up our off-the-cuff perceptions of talent. If we could distill fielding ability to numbers, I wouldn't be very surprised to see a major league spread about as large as that for batting. So let's give that a shot. We can use Tango's positional adjustments to correct for the difficulty in scaling fielders properly, and then use UZR to get an idea of the spread.

Hawpe (LF): -47.7 (UZR), -7.5 (Pos)

Cabrera (SS): +14.2 (UZR), +7.5 (Pos)

The difference between Hawpe and Orlando Cabrera ends up being on the order of 75 runs - not too far off our ML spread of hitting value. The distribution is a little different, but this shows that the run differential between the best and worst gloves in baseball is comparable to that of the best and worst bats.

And yet people are stunned that a player's defence can be meaningful in terms of wins and losses. Blame it on fielding positions.

*I like footnotes