Brandon Crawford, Jason Kipnis and the Flip Side of the Coin

Like any baseball stat, Wins Above Replacement provides the answer to a question. The question, in this case? Something like this: accounting for all the main ways (hitting, running, defense, etc.) in which a player can produce value for his team, how many wins has this particular player been worth?

There are, of course, criticisms of WAR. Some valid, others less so. One prominent criticism is how defensive value is handled in WAR. Some don’t understand how it’s calculated. Others understand but also question how well it represents a player’s defensive contributions. These criticisms shouldn’t be dismissed. As with all baseball statistics, though, it’s necessary to consider WAR in the context in which it’s presented — that is, to remember the question a metric is intended to answer and the method by which it attempts to answer that question.

On Monday, I completed one such reminder in a discussion of players whose WAR totals this year are probably low based on what we know about their defense. Today, I’ll make another attempt — this time, by examining players whose WAR totals are probably inflated by defensive numbers unlikely to be sustained over the course of a season.

In the comments of Monday’s post, one reader, Ernie Camacho, noted:

[T]here is a weird tension in this article between quantifying and estimating what has already happened, on the one hand, and evaluating player talent, on the other. I’m not sure we should be blending the two.

This is a good point. That tension most definitely exists, and it’s possible that some of that tension is what causes people to discount defensive metrics — and WAR as a whole. I agree that, in terms of calculating WAR, we should not be blending what has already happened with what we think will probably happen. Over time, in an ideal world, WAR captures both. In smaller samples, however, this is more difficult to do. In fact, there’s actually something that does capture the blending when we have smaller samples: projections. If we want to capture a player’s talent level at any given moment, projections do that very well.

However, when we want to take a player’s recent results and try to determine value, projections don’t completely capture what actually happened, which is often what we want to see. It’s a big part of why we look at FanGraphs’ leaderboards. We want to see who’s having the best season so far, and in many ways WAR does capture that effect. It can be a bit misleading due to who appears on leaderboards. The standard for position player is generally qualified batters. If a hitter has 3.1 plate appearances per team game, then he shows up on the leaderboards.

There isn’t a similar qualification for fielders, however — and even if there were, it would still be difficult to use. As I noted Monday, every plate appearance has pretty close to the same difficulty level. There’s a difference obviously between facing Clayton Kershaw and… anyone who’s not Clayton Kershaw, but the range of possible outcomes is still pretty narrow. There are no “routine” pitches that batters convert to hits 99% of the time.

That’s not so on defense. In the field, players differentiate themselves both by not making mistakes on routine plays and also making difficult plays. Whether a fielder is actually receiving opportunities will have an effect on his defensive metrics in a way that is dissimilar to the hitting side of things. The middle 80% of hitters are currently separated by 21 runs; the middle 80% of fielders, only 13 runs. Defensively, players are clumped closer together when compared to average, so a greater number opportunities can have a big effect. The differences even out some over the course of a season — and even out even more over the course of several seasons — but they can provide strange results for some players in just one-third of a season.

By noting the challenges produced by defensive metrics, it might seem like I’m merely providing more fodder for the anti-defensive metrics/anti-WAR crowd. That isn’t actually the case. I wrote Monday, but it’s worth repeating here, that a vast majority of players in 2016 have actually produced fielding-runs marks fairly close to their three-year averages, meaning a glance at the WAR Leaderboards will provide a good indication of value thus far without any mental adjustments. There are also a few outliers, though, and those players are worth identifying.

As I did Monday, I looked at qualified hitters with at least 3,000 defensive innings from 2013 to 2015, came up with a one-third season average mark, and compared that number to their mark at the end of last week. There were only four players at least five runs off from their three-year averages.

Brandon Crawford is a very good defensive player, one of the best shortstops in the game, but his UZR/150 this year — which normalizes chances over 150 games — is three times his career-average mark. Even if you believe his DRS better captures his defensive talent, Crawford would still be about six runs above what we would expect. He’s an above-average offensive player at shortstop, where he provides very good defense, but right now his 2.6 WAR is ranked 20th among position players. If his offensive numbers moved up a little bit closer to where they were last year (117 wRC+) finishing 20th is well within his reach, but his defensive numbers will not keep up with this pace as the season continues.

It is possible that Jason Kipnis‘ three-year averages underrate him a bit on defense. Kipnis wasn’t fully healthy in his 2014 season, and that might have affected his defensive numbers negatively. That said, the probability that, in his fifth full major-league season, he’s all of a sudden the best-fielding second baseman in the majors by a fair margin seems slim.

Dexter Fowler’s main contribution has been with the bat this year; he’s produced a 141 wRC+ thus far. His defensive numbers have also been out of the ordinary. Fowler actually appeared on this same list last season before posting negative defensive numbers the rest of the way. It’s possible that Fowler’s one season with the Astros — roaming that club’s difficult and spacious center field — skewed his average. However, Fowler has never posted a positive UZR or DRS, and it seems unlikely that his skills would change at age 30, even if his positioning has improved by playing deeper this season.

Surprise! Putting Adrian Beltre in the same sentence with the phrase “defensively overrated” is kind of a daring move. Adrian Beltre is still very good defensively at age 37, which is a pretty amazing feat. However, it’s unlikely he’s headed to his best defensive season by UZR since his final year in Seattle. Adrian Beltre is great, and he’s well on his way to another very good defensive season, but he’s not quite this good.

*****

This post isn’t intended to cast doubt on UZR. To the contrary, I’m a big believer in UZR and think it succeeds in capturing defensive prowess in players. That doesn’t mean it’s perfect, however — and, like most stats, it’s always better in large samples. Context should be helpful not for dismissing stats, but to help better understand them so they can be better utilized when they are discussed. Defensive metrics are complicated and can be difficult to understand and apply, but they have great worth in helping us reach player value.