Position adjustments are a hot topic lately. Everyone agrees that first base is easier to fill than shortstop, but opinions on positions closer on the defensive spectrum aren't as unanimous. And the relative rankings aren't as important as the magnitude of the differences.
Of the folks who attempt to measure positional adjustments, there are two main schools of thought:
- Base the adjustments on the difficulty of switching from one position to another defensively.
- Base the adjustments on offensive averages by position
I don't want to re-hash the arguments for either approach, so I'll just mention that I favor defense-based adjustments. Right now I just want to show how the methods compare in practice.
Here are Tango's positional adjustments, based on relative defensive abilities. A positive number represents how many runs more valuable a player at that position is per 700 PAs compared to the average position.
These numbers are based on a study of UZR ratings of players who played multiple positions and account for the fact that left-handed throwers are at a severe disadvantage at 2B, SS, and 3B. A weighted average (counting DH as half a position) puts the average position adjustment at -1 run. One would think the average should be zero, but more on that later.
Here are 2008 positional adjustments based on offense at each position per 700 PAs across MLB. Thanks to Devil_Fingers for doing the work:
The weighted average of these adjustments is -3.8 runs. That's right, they appear to be even worse than the defense-based numbers. But the reason that they don't average up to zero is that not all positions receive the same number of at-bats. First basemen and corner outfielders hit more often than short stops and catchers, because managers know they're better hitters. (And pitching is included, which it probably shouldn't be.)
So, before comparing the defensive-adjustments with the offensive-adjustments, we need to put them on the same scale. The correct scale is open for debate, but because I like symmetry, I'm going to average each set out to zero while keeping the absolute differences between positions the same:
Now that's pretty interesting. The systems are within two runs of each other for every position except two. In other words, while the theoretical discussion is intriguing, for practical purposes, both methods produce almost exactly the same results. However, among similar positions, there is some wider variation:
- The difference between the two systems in LF and RF has the opposite sign, meaning they disagree that the two positions are equally valuable by about three runs.
- The offensive adjustments put a much larger gap between shortstops and second basemen than the defensive adjustments do.
- The defensive adjustments have the infielders as four runs more valuable total than the offensive adjustments.
The differences at third base and designated hitter between the two systems shouldn't be ignored, either. Since they've been discussed thoroughly elsewhere, I'll simply summarize the arguments.
Designated hitters as a group don't hit nearly as well as you'd think. While David Ortiz, Aubrey Huff, and Jim Thome carried the position in 2008, you've also got Jose Vidro and AAAA-level platoons masquerading as DHs. Should we compare David Ortiz to those crappy hitters, who probably shouldn't be DHing? I don't think so, because the players who actually do DH are only part of the pool of players who could DH. Most players at first base and the corner outfield spots, positions that outhit DHs, would be upgrades at DH. So we should compare DHs to the best hitters without much defensive skill, not just players who fit that description and who also happen to be chosen by their teams to DH.
Third baseman are great hitters in today's game. And, via Tom Tango's research, their fielding abilities are on-par with second baseman. Given both of those statements, one is led to conclude that today's group of third basemen is more productive than the group of second basemen. But with an offensive-adjustment approach, third basemen are compared only to each other in both hitting and fielding, making them exactly as valuable as second basemen, by definition.
Finally, let me note that these offensive adjustments are based on 2008 numbers only. I'm sure they would be moderately different using even 2007 data, let alone 1998 data or 1958 data. It would be an interesting study to see how they change on a yearly basis, all normalized based on runs-per-win and to the same mean.