A side conversation in the Andy Pettitte thread discussed whether one would rather have the Red Sox or Yankee outfields in 2011. There was some attempt to compare them using OPS and OPS+, which is problematic given the fact that at least three of the six outfielders rely heavily on speed and baserunning as part of their offensive value. So let's compare these players' offense a little differently.
OPS+ is good for a lot of things, including thumbnail sketches of players' offense. It's adjusted by park and league, which means you can compare the raw offensive value of a player without having the question of, "Well he hits in Fenway, so that's inflated," etc.
But OPS+ is not so helpful in three ways, and it's important to know what they are in case you're dealing with players who fall into these categories:
- It weights OBP and SLG the same, even though OBP is nearly twice as valuable. That doesn't come up very much, but there are high-OBP, low-SLG players who would be unfairly disadvantaged by a straight OPS/OPS+ comparison.
- It doesn't take into account baserunning, which is an aspect of a player's offensive skill. So a fast, highly successful baserunner could be severely penalized by looking only at his OPS/OPS+.
- It doesn't adjust for position. A 100 OPS+ is still league average overall, but at catcher or shortstop that's actually extremely valuable, whereas at first base or DH it's a huge negative.
Baseball-Reference splits WAR into its offensive and defensive components. If you don't like/trust the defensive data that's out there, you can use the purely linear weights-based offensive data, which is parks and league adjusted, properly accounts for the importance of reaching base and further includes baserunning as part of the data. It also includes a playing-time component, so a better offensive player who misses time will be dinged for it (or, rather, won't accumulate as much).
Using oWAR, here's what we get for Ells/Grandy/Gardner, the three obvious speed guys:
3.2 oWAR (2009, career best)
1.3 oWAR (2008, rookie season)
2.6 oWAR (2010, career best)
1.5 oWAR (2009, rookie season in 284 PA)
5.9 oWAR (2007, career best)
4.6 oWAR (2008)
2.9 oWAR (2009)
2.5 oWAR (2010)
The components of oWAR are:
- Reaching on errors
- Grounding into double plays
All these are expressed in runs above (or below) average based on linear weights.
[[Linear weights are derived by calculating the expected runs in any given situation, looking at what the hitter did, then calculating how many runs should be expected in the new situation and crediting the hitter with the difference, positive or negative. For example, if J.D. Drew comes in with the bases loaded and no outs, let's say the average number of runs we should expect to be scored that inning is 1.8 (randomly made up number). Drew then hits a bases-clearing double. He gets credit for three runs plus he's now a runner on second with no outs. Let's say we should expect another 0.8 runs to be scored that inning based on that situation (another made up number). So he gets credit 3.8 runs of credit. Subtract from the expected 1.8, and Drew gets 2.0 batting runs for that plate appearance.]]
It then includes a position adjustment and adds runs "above replacement" depending on how much playing time the player has.
Granderson was worth three wins (30 runs) with his bat alone in 2007 and two (20 runs) with the bat in 2008. He's about two or three runs above average in straight baserunning and doesn't reach on errors any more or less than average. He's generally avoided double plays between two and four runs better than average, but that fell to zero last year, and his injuries cost him a couple runs in 2010. He actually graded out as a better hitter in 2010 than in 2009, but went from four runs above average at avoiding double plays to zero, which pretty much wiped out the advantage. Except for 2007-08, Granderson's entire career is basically average to slightly above with the bat with slightly above average baserunning and good defense (though that fell from +4 to -4 runs above average last year; could be a fluke). I wouldn't expect much more at this point.
Give Gardner the nine runs of playing time that he missed in 2009 that he got in 2010, and he would have had 2.4 oWAR, and his 2010 would have only been a marginal improvement. How? Well, first his bat improved by basically a full win, from -3 to +6. His running was basically the same, very good in both seasons (+4 to +5), but he, too, went from being good at avoiding double plays (+3 runs in 2009) to merely average (0 in 2010). Also, spending his time in left field means that his offense isn't nearly as valuable as it was when he was in center. Position adjustment costs him five runs of value. So he gained 10 runs in batting and baserunning but lost eight in GDP and positioning. Gardner's real value comes in his defense, where he was 14 runs above average in 2010, up from 1 in 2009. Again, we'll see what 2011 says. Defensive metrics need three seasons of cumulative data before they have a large enough sample size to be trustworthy.
Ellsbury in 2009 was exactly average with the bat (0 runs above average), but was +6 on the basepaths, the best baserunning season of any of the three players (though a difference of one or two runs I can't really see as being predictive). He also was +2 on reaching base via error, gained two runs for playing center field and didn't miss any time. He improved in every facet of his game from the year before, his rookie season, when he was -12 with the bat, +6 on the paths, -1 on errors, +2 on double plays and lost two runs because of the time he spent in left and right fields. His defense was rated as awesome in 2008 (22 runs above average) and terrible in 2009 (-10). Another wait-and-see on that one.
Here is a straight-up comparison of Ellsbury's 2009 with Gardner's 2010:
- Ells 09: 0 Bat, +6 Bsrun, +2 ROE, 0 DP, +2 Pos, +22 Rep
- Gard10: 6 Bat, +5 Bsrun, 0 ROE, 0 DP, -4 Pos, +19 Rep
DP and ROE are sort of hybrids, focusing on the player's batting and running, but if we count them both as running ability (because no one really thinks of either as a positive hitting event), it breaks down like this:
- Ells 09: 0 Bat, +8 Bsrun, +2 Pos, +22 Rep
- Gard10: 6 Bat, +5 Bsrun, -4 Pos, +19 Rep
Gardner was overall three runs better than Ellsbury on offense, batting and baserunning included, but he did it while playing mostly left field, where the offensive bar is higher, and he batted eighth or ninth most of the season, so he could not compile as much value over your typical replacement player as Ellsbury did batting leadoff.
So at this pojnt, it's way too close to call based on raw production in their respective breakout seasons, but if the Yankees continue to put Gardner in left field and bat him at the bottom of the order, it looks like they'll be costing themselves basically a win of value, regardless of how well he performs at the plate. Perhaps they have that made up elsewhere — his defense in left could be a win better than if he were defending in center, for example, or they have eight better hitters in the lineup, in which case batting him ninth is the best use of his talent.
And here are the 2009-10 lines for all the outfielders:
- Drew: +6 Bat, 0 Bsrun, 0 ROE, 0 DP, -6 Pos, +18 Rep, 1.8 oWAR
- Swish: +23 Bat, -1 Bsrun, 0 ROE, 0 DP, -7 Pos, +22 Rep, 3.5 oWAR
- Craw: +26 Bat, +5 Bsrun, 0 ROE, 0 DP, -7 Pos, +22 Rep, 4.7 oWAR
- Gard: +6 Bat, +5 Bsrun, 0 ROE, 0 DP, -4 Pos, +19 Rep, 2.6 oWAR
- Grand: +4 Bat, +2 Bsrun, 0 ROE, 0 DP, +2 Pos, +18 Rep, 2.5 oWAR
- Drew: +23 Bat, -2 Bsrun, -1 ROE, +2 DP, -6 Pos, +19 Rep, 3.5 oWAR
- Swish: +21 Bat, -1 Bsrun, -1 ROE, -1 DP, -7 Pos, +21 Rep, 3.2 oWAR
- Craw: +11 Bat, +4 Bsrun, -1 ROE, +3 DP, -7 Pos, +22 Rep, 3.2 oWAR
- Ellsbur: 0 Bat, +6 Bsrun, +2 ROE, 0 DP, +2 Pos, +22 Rep, 3.2 oWAR
- Grand: -2 Bat, +2 Bsrun, -1 ROE, +4 DP, +3 Pos, +23 Rep, 2.9 oWAR
- JD Drew: +29 Bat, -2 Bsrun, -1 ROE, +2 DP, -12 Pos, +37 Rep, 5.3 oWAR
- Swisher: +44 Bat, -2 Bsrun, -1 ROE, -1 DP, -14 Pos, +43 Rep, 6.7 oWAR
- Crawford: +37 Bat, +9 Bsrun, -1 ROE, +3 DP, -14 Pos, +44 Rep, 6.9 oWAR
- Gardner: +12 Bat, +9 Bsrun, 0 ROE, +3 DP, -3 Pos, +29 Rep, 4.1 oWAR
- Ellsbury: -8 Bat, +7 Bsrun, +2 ROE, 0 DP, +2 Pos, +25 Rep, 2.8 oWAR
- Granders: +2 Bat, +4 Bsrun, -1 ROE, +4 DP, +5 Pos, +41 Rep, 5.4 oWAR
That 1.8 oWAR from Drew was his worst since 1999, fwiw. Ellsbury didn't play much in 2010, but when he did play, he was beyond terrible, eight runs below average with the bat alone.