Do Defensive Metrics (and WAR) work at Coors Field?

Ahhh, Coors Field. A beautiful place to watch a game, a wonderful place to hit, and a terrible place to pitch. If you’re a casual baseball fan, that’s probably the extent of your knowledge about the Rockies. Remember the Blake Street Bombers? What a fun group! Didn’t they go to the World Series once? They did! Is [Rockies player] any good? Nah, he’s just a Coors Field creation.

If you’re a more discerning baseball fan, you might know something else: Coors Field is an enormous disadvantage to the Colorado Rockies. Pitches move much differently at altitude than they do at sea level, and as a result, two things are true. First, it’s hard to pitch in Denver, as any glance at the team’s history shows: only eight Rockies’ pitchers have more than ten fWAR in their career, and John Thompson (?) ranks among the team’s top starters of all-time. Second—paradoxically—playing in Denver makes it extraordinarily difficult to hit on the road, as several baseball analysts have concluded in recent years. Since the Rockies’ inception, they’ve been the worst road hitting team in baseball on a park-adjusted basis, and it’s not remotely close. This is what people talk about when they say Reverse Coors: while Coors Field inflates offensive home statistics, it deflates offensive road performance.

But what about defense?

When Coors Field was built, the team knew about the effects of altitude, so it built the ballpark big. Really big. Coors Field has the most fair territory in baseball; compared to the average major league stadium, Coors has about 7,840 more square feet of fair territory. All of this additional space comes from the outfield—on average, the outfield in Colorado is about 10% bigger than the outfield in other stadiums.

The expansive outfield in Colorado has done what it was intended to do: Coors Field is not the easiest place to hit a home run. But it is the easiest place to get a hit, as there is simply more room for the ball to drop in than there is in other stadiums. On the defensive side, this means that there’s more ground to cover—those additional 7,840 square feet aren’t going to patrol themselves.

There are two primary sabermetric defensive statistics: Defensive Runs Saved (DRS), and Ultimate Zone Rating (UZR). Both statistics calculate how many runs better or worse than average a particular player is while playing defense, and for outfield defense, both statistics include a factor for range. To use UZR as an example, the formula estimates how likely each particular batted ball is to be caught based on the ball’s speed, location, and trajectory. Whether an outfielder receives or loses credit depends on whether the ball was “difficult” or “easy” to catch.

UZR includes a rudimentary park factor. But when we look at outfield UZR over the last fifteen years, something interesting emerges:

Your eyes do not receive you: the Rockies have had the worst outfield UZR in baseball over the last fifteen years. In fact, the Rockies outfield UZR is approximately 33% worse than the next closest team, and over twice as bad as everyone but the White Sox and Pirates. Sure, the Rockies may have had some bad outfields, but even on a per-season basis (UZR/150), the team has only had four seasons where they’ve been close to average. What’s more, as compared to the bell curve of every team’s performance, the Rockies have had no season where they’ve been anywhere close to good:

On a per season basis, the average major league outfielder should have a UZR of zero. (And if you average out every single outfielder’s season since 2004, this is actually true.) But the average Rockies outfielder has a UZR/150 of negative six—meaning that over the course of the season, UZR thinks the average Rockies outfielder gives up six more runs than a league-average defender. That the Rockies have been consistently rated poorly in outfield defense suggests that something systemic is going on. Put another way, it’s probably the statistic, not the team.

So the enormous Coors Field outfield is distorting a somewhat obscure defensive statistic. Why does this matter?

In one sense, it doesn’t. The fact that one statistic underrates Rockies outfielders doesn’t actually have an impact on the game—this doesn’t actually limit David Dahl’s range, or decrease the number of fly balls Charlie Blackmon can get to. But it does impact how Rockies players are valued, and importantly, it impacts fWAR. Along with base running runs, batting runs, and positional and league adjustments, defensive runs are one component of fWAR. In other words, if UZR underrates Rockies outfielders, fWAR does, too.

How much does fWAR underrate Rockies outfielders? Well, we can calculate it. In 2019, here’s how UZR rated the Rockies outfielders, and how they ranked based on fWAR (the “defense” rating isn’t exactly the same as UZR, but for our purposes is doesn’t matter):

As discussed, the average outfielder should have zero UZR/150 over the course of a season, but the average Rockies outfielder has negative six. We can counteract this bias in an easy and admittedly crude way: we can give each Rockies outfielder six more defensive runs—annualized depending on their outfield playing time.

When we do that, the Rockies fWAR numbers look a little different:

The average “bias” of fWAR against Rockies outfielders becomes clear: an outfielder who plays an entire season in Colorado will likely be underrated by about half a win.

Of course, this crude math simply assumes that each Rockies outfielder is six UZR/150 better than UZR/150 states. That is not necessarily true. In any single season, UZR is quite volatile—it’s only when evaluated over time that the Coors Field bias emerges. But luckily for us, another defensive metric has emerged in the past several years that is both more accurate and more reliable on a one-season basis: Outs Above Average (OOA), developed by Statcast.

Like UZR and DRS, OOA calculates how many outs a player has saved compared to the league average. But Statcast uses more precise technology that tracks exactly where each outfielder stands prior to the play. This leads to more accurate results than either UZR and DRS, and importantly for us, OOA doesn’t appear to have a Coors Field bias: over the last four years, Rockies outfielders have averaged slightly under zero OOA—well within the range of what you might expect of a neutral statistic.

Rather than simply re-calculating fWAR using the crude method, then, we can re-calculate it using OOA instead. (As its name suggests, OOA uses “outs” instead of “runs”, but we’re simply using it on a weighted comparative basis, so that doesn’t matter here). That calculation gives us this:

What we see here is rather interesting. OOA agrees that Charlie Blackmon and Ian Desmond are subpar defenders, and as a result, their fWAR barely changes with a more accurate statistic. But OOA strongly disagrees with UZR on Raimel Tapia (5 OOA) and David Dahl (2 OOA), as it thinks both were above average defenders. That translates into a considerable fWAR adjustment for both players: David Dahl’s 1.4 fWAR year becomes a 1.8 fWAR year, and Raimel Tapia sees a 1.1 fWAR swing into the positive. (I promise that I did not engineer this result because of my love of Raimel Tapia).

These differences may not seem like a lot, but they’re important. Remember 2017, when Charlie Blackmon was 5th in the National League MVP race? That year, he posted 6.6 fWAR, while Giancarlo Stanton—who was crowned NL MVP—had 7.3 fWAR. But the crude model states that Charlie actually had an fWAR somewhere in the range of 7.1 fWAR—still not as much as Stanton, but close. And as an aside, under Charlie’s current contract, that difference might have cost him a million bucks: Charlie gets $2 million for a 1st, 2nd, or 3rd place finish in the MVP race, but only $1 million if he finishes 4th or 5th.

The defensive bias of UZR and fWAR has even more of a pronounced effect when evaluated over a period of years. Despite winning five of his seven Gold Gloves with the Rockies, Larry Walker’s average defensive rating while in Denver was negative .44. That drags down his fWAR, which should probably be higher given the effects we’re talking about. A small adjustment of a half-win per Coors Field season puts Walker at 73.7 fWAR, leapfrogging him past Gary Carter, Jim Thome, Frank Thomas, and Derek Jeter. Thankfully, under-counting Walker’s defense didn’t get in the way of his Hall of Fame election, but the Coors Field bias can make a real difference when the margin of error is close.

More central to the team’s operations, consistently undervaluing Rockies outfielders suppresses their value—both for the trade market and for free agency. Despite winning the Gold Glove in 2010, 2012, and 2013, UZR never believed that Carlos Gonzalez was an above-average defender—part of the reason that other teams reportedly thought he was overrated. Accounting for the outfield in Colorado might have changed that, which might have brought the offers somewhat closer to the Rockies’ asking price. Raimel Tapia might be a modern example of this as well. For those who would (mistakenly) like to trade him away, Tapia has less value than he would if he were widely seen as a solid defender.

If the Rockies were shrewd, the Coors Field defensive bias could actually work in their favor. Unlike pitching and hitting in Colorado, there doesn’t seem to be any impact on actual defensive performance. And this makes sense: the negative effects of pitching and hitting in Colorado have to do with the difficulty of the home/road adjustment, a consideration that is absent when chasing fly balls. Building a home-grown, stellar-but-underrated defensive outfield would give the team an advantage at home. What’s more—provided the team isn’t looking to trade anyone—the fact that the Rockies outfielders are consistently underrated for their defense could help them do it for less.