Wins Above Replacement (WAR) is not a statistic.
A statistic is a piece of data from a portion of a population. That is not what WAR is. WAR is a measurement, one that has been concocted by sabermetricians in an attempt to quantify player value. It is imperfect (this is no mystery), and a large amount of the imperfections come conjointly with the the lack of universal, reputable defensive metric. The introduction of Statcast to the game of baseball should slowly help us get a better understanding of player defense and how to quantify it, but for now, we wait.
But WAR also attempts to adjust for player position in a way that rubs me the wrong way. Here is Fangraphs calculation of WAR:
WAR = (Batting Runs + Base Running Runs + Fielding Runs + Positional Adjustment + League Adjustment +Replacement Runs) / (Runs Per Win)
Baseball Reference and Baseball Prospectus have slightly different variations, but all are built roughly the same way and all contain some sort of positional adjustment. Here are the positional adjustments for each position per Fangraphs (all are per 162 defensive games):
Catcher: +12.5 runs
First Base: -12.5 runs
Second Base: +2.5 runs
Third Base: +2.5 runs
Shortstop: +7.5 runs
Left Field: -7.5 runs
Center Field: +2.5 runs
Right Field: -7.5 runs
Designated Hitter: -17.5 runs
In essence, a first baseman, for example, who played all 162 games at first base is charged -12.5 runs solely because he played 1B. A SS gains +7.5 runs solely because he played SS. That math for the 1B/SS comparison somewhat checks out. The Runs Per Win total (as seen in the denominator of the WAR calculation) was just north of 10 R/W in 2017, meaning that a shortstop who contributed 0 runs in batting, baserunning, and fielding after league adjustment, is worth 0.7-0.8 WAR for the season. A 1B who provides 20 runs in batting, baserunning, and fielding is also worth about 0.7-0.8 WAR for the season. In this specific example of 1B and SS, it makes sense, both mathematically and logically; a hitter who can play a league average shortstop is more value in nature than the same hitter who is unable to play that league average shortstop and must instead play LF or 1B. In fact, it may not even be harsh enough. If you throw a league average 1B at SS, the chances he costs the team more than 20 runs defensively over 162 games are actually pretty decent, I would think.
With other examples, such as comparing infielders to outfielders or comparing catchers to literally any other position, it’s a bit less intuitive. However, my real issue with the positional adjustment, aside from the fact that the methodology in which these adjustments have been formulated, is based primarily on the players that have played 2+ positions, is that it seems to punish elite defensive players in a way that isn’t truly representative of their worth. It attempts to throw a blanket over the entirety of the league that isn’t always fair to better defenders at their position. Take Brett Gardner for example. In 2017, with a minimum of 650 defensive innings (roughly half a year), Gardner was tied for 13th in baseball with a 7.1 Ultimate Zone Rating (UZR), which Fangraphs uses at it’s defensive metric in WAR, 2nd among LF only behind Alex Gordon. By Defensive Runs Saved (DRS), which BRef uses for it’s WAR, Gardner was the #1 defensive LF in baseball. Ultimately, no matter which way you try to flip it, he’s an elite defensive “left fielder”.
So why is Gardner in LF and not CF? Range isn’t the issue. According to Baseball Savant (Statcast), Gardner’s top sprint speed of 28.7 mph was tied for 37th best in baseball in 2017 out of over 450 players tracked. Arm strength isn’t the issue. Over the last 10 years, out of players that have logged similar innings in CF, Gardner’s arm grades very well there. He’s only relegated to LF due to better center fielder(s) that are already on the Yankees’ roster. Aaron Hicks logged a relatively huge 12 DRS and 4.8 UZR in a mere 440 innings before getting hurt. After the injury, Jacoby Ellsbury took over, and although he isn’t as electric as he once was, he makes for an acceptable CF. Gardner probably should have played over Ellsbury, but that’s neither here nor there. Simply by being relegated to LF instead of CF (a position which he can and probably would play on another team) Gardner was docked net 10 runs in his WAR calculation, or almost 1 WAR. Gardner would have to be one of the worst CF in baseball for that trade off to make sense, something I don’t think would be the case given his average speed (for a CF-type) and above average arm. Is Gardner truly a 0.5-1.0 WAR less valuable asset solely because of the position he played in 2017? Of course not. He is still the same player, just in a different position due to roster depth. That’s where this positional adjustment can become an issue.
Where it really began to make less sense is the infield. With the outfield, sure, there is less grey area in the adjustments. The center fielders are typically the best and fastest, the right fielders typically have the greatest arm strength, and the left fielders are typically the remains, or potentially a place where you hide a great bat like Rhys Hoskins when you sign a new 1B like Carlos Santana. It’s easier to adjust these positions. The infield is a little tougher, and my particular qualm is with third base.
Let’s take Andrelton Simmons and compare him to 2 of the best defensive 3B in baseball, Nolan Arenado and Matt Chapman. Arenado has been the best for some time, logging 104 DRS since his rookie year, the highest DRS total for a 3B since the statistic was invented in 2003, behind Adrian Beltre and Scott Rolen. Chapman, the newcomer, had 19 DRS in just 727 innings* last season, making it the 2nd highest DRS/inning total EVER at the position, only behind 2013 Arenado. Let’s call them 20 DRS 3B, as that is Arenado’s career average. The positional adjustment difference between SS and 3B is +5 runs. Do we think that if Andrelton Simmons moved to 3B, he could consistently be a 25 DRS 3B? Even with how good Simmons is, I find it hard to believe that would be the case, as there have only been 7 25+ DRS single seasons ever.
There, in between all stats and some slight fawning over Matt Chapman, lies the problem. This WAR positional adjustment assumes that the average SS could slide right over to 3B and be +5 runs, or someone like Simmons could slide over and be better than the best. I’m sure this is true for a lot of players, but the assumption is too broad to be totally comfortable with. 20 years ago, the shortstop was the most important defensive position in baseball. This is still most likely true, but with the fly ball revolution spreading throughout the game and with shortstops being a part of almost 20% fewer plays than they did 10+ years ago, third basemen now have to make nearly 7 plays for every 8 that shortstops do. In the words of Tom Tango, “the share of responsibility for 3B, relative to the others, is going up.”
This assumption on the part of the adjustment that playing the shortstop position is automatically +5 runs worth of value over all third baseman is wrong. Can Andrelton Simmons make these plays?
Matt Chapman is yummy.
I’m going to say Andrelton Simmons could potentially make those plays, but only because he’s a generational defensive shortstop. He’s may be the only SS that could, and there is still a decent chance he couldn’t make those plays. Simmons will most likely retire with us, the fan, not knowing the answer to that question. The reason behind the hesitancy to say that he could is because ingrained in this play is twitchiness and reaction times that shortstops simply aren’t accustomed while playing defense. A 100 mph scorcher down the line will reach the third baseman in 0.6 seconds, giving the him somewhere in the range of 0.45 seconds to react. Just based solely on defensive positioning, that same 100 mph ball would take 0.8 seconds or so to reach the shortstop, giving him 0.65 seconds to react. That 200 milliseconds difference doesn’t sound like a lot, but it is, and it is one of a few things that aren’t accounted for by position adjustment, along with other things like hand-eye coordination and additional arm strength, both things I would argue are more important at third base. The inverse is true also; we have no idea how elite 3B would handle having to range extremely far to his left and his footwork/transitions around the bag when turning a double play. Kiley McDaniel, prospect writer at Fangraphs, recently had this to say about attempting to try out weak offensive prospects with a strong arm at catcher, since the offensive bar for catchers is much lower than other positions:
“Catching is really hard and it also demands a specific arm action/exchange, so some guys arms play down in that setting. Like some guys throw 95 on the mound and have an average arm in the OF, then plus at SS…there’s some mechanical things about the positions that do weird things to arm strength, mainly body position and kinetic chain and all that. Very specific to the player.”
The question/answer was about the catcher position specifically, but at the same time, it also serves as another example of how when comparing how players would fare at other positions, sometimes it’s apples to oranges. There are too many variables.
As I’ve said before, the idea behind some sort of adjustment for the ability to play certain positions in player evaluation is a genuine one, but in it’s current state, it’s incomplete and doesn’t do a fantastic job encompassing all types of players. Any shmuck can play left field; a true center fielder definitely can, so in that regard, positional adjustment does a fine job. However, I don’t think there are any players in baseball that can do what Nolan Arenado and Matt Chapman types can do at third base, and that includes 99% of shortstops. So why do shortstops get a +5 run boost in positional adjustment when they can’t play 3B to the level of those 2 (and perhaps others) and those 2 see nearly the same amount of balls in play? This sentiment is true to a degree with all positions, and on the infield especially. Hopefully one day, Statcast and any other advanced camera tracking systems will be able to create a defensive metric that can allow baseball statisticians to remove positional adjustments all together, but for now, this article simply serves as a reminder that player evaluation can’t be done in one measurement such as WAR. It’s as good as you’re going to find, and extremely useful in quick citing or glancing at a player year/career, but it will forever be somewhat inadequate. Given the choice, I may prefer a Arenado/Chapman type glove over any non-Andrelton infielder in baseball and happily play them at 3B, despite what current player evaluation would lead you to believe. WAR may not give them nearly the credit they deserve.
*727 innings is a rather small sample size for defensive metrics. It’s similar to roughly 1 month of hitting. It’s very impressive, and clear he’s talented and perhaps elite defensively, but this small sample most likely does not mean he is a 35+ DRS third baseman.
Photo Credit: AP Photo/Ross D. Franklin