Baseball Needs a Leveraged Version of Outs Above Average

Last week, Colton Cowser put forward a strong contender for catch of the season. With two outs in the bottom of the ninth, a runner on second, and the Orioles up by one, Rockies batter Jake McCarthy hit an Andrew Kittredge sinker to straightaway center. Cowser leapt with his back to the wall and his glove raised high in the air. He caught the ball. Baltimore won.

A week later, USA TODAY MLB columnist Bob Nightengale had this to say about Cowser’s catch:

If WAR was really this all-encompassing stat, then why wasn’t Baltimore Orioles center fielder Colton Cowser credited a full 1.0 when he robbed Jake McCarthy of a game-winning two-run homer, preserving the victory for the Orioles? Isn’t that a full win above replacement?

It’s more than fair to ask this question. I do see where Nightengale is coming from. At the same time, there is so much wrong with the point he’s trying to make that the fact that he originally wrote Brandon McCarthy (who has been retired for six years) instead of Jake McCarthy (the actual batter in question) wasn’t even one of the top three most flawed things about it.

First of all, the question starts with a straw man. Who says WAR is an “all-encompassing” statistic? FanGraphs makes it clear that their version of WAR for position players is a context-neutral metric. So does Baseball Reference.

WAR doesn’t care how many outs there are when a player hits a go-ahead home run or what the score is when the defense turns a double play. That’s not a defect, either. It’s by design.

You don’t have to like it. You don’t have to use it. But criticizing WAR for not acknowledging game context is like giving your Uber driver a one-star review because they didn’t offer to cut your hair.

Second, this argument seems to misunderstand the fundamental idea behind wins above replacement – specifically, the “replacement” part. You see, if Cowser wasn’t playing center field, someone else would have been. And someone else could have made that catch.

I’m not saying they would have. It was a difficult play, and Cowser is pretty darn good at his job. Regardless, unless you think there is literally no chance that anyone else the Orioles could have put in center field would have reeled that ball in, it wouldn’t make sense to credit Cowser with a full win above replacement. Even then…

No team has ever won a ballgame on a great catch alone. If you’re familiar with the basic rules of baseball, you’ll know it takes two things for a team to claim a victory:

• Score more runs than your opponent
• Make at least 27 outs

As important as Cowser’s catch may have been, it was still only worth one out. It saved two runs, but it didn’t bring any across the plate.

If Orioles pitchers had given up a couple more runs at any point earlier in the game, Cowser never would have had the chance to make that catch. If Coby Mayo and Dylan Beavers hadn’t scored that day, nothing any Orioles did on defense would have mattered.

Cowser wasn’t credited with a full win for the same reason that McCarthy didn’t take all the blame for his team’s loss. If McCarthy had hit the ball a few feet farther, the Rockies would have won. His actions, just like Cowser’s, directly made the difference between a win and a loss. But you could say the same thing for every Colorado batter who came before him.

So, long story short, the answer is no. Cowser wasn’t credited with a full win above replacement, nor should he have been.

Even so, Nightengale’s question got me thinking about something else. He’s right to suggest that a major category is missing from the current sabermetric landscape: context-dependent defensive metrics.

Baseball Is Missing Context-Dependent Defensive Metrics

There is no well-known, publicly available metric that properly measures the value of Cowser’s catch.

Let me be clear: WAR doesn’t need to be that metric. But I think that metric should exist.

Statistics like wRC+, OAA, and WAR help us evaluate players in a context-neutral capacity. A home run in the first inning boosts a player’s wRC+ just as much as a home run in extras. A diving catch with the bases empty is equal to a diving catch with the bases loaded as far as OAA is concerned. You get the point.

When it comes to hitting, however, we can also look at statistics that don’t erase context. Win Probability Added is a big one. More recently, Baseball Savant added leveraged batter run value to their stable of stats.

Unfortunately, we don’t have access to numbers that let us evaluate defense the same way.

Imagine, if you will, a BBWAA voter who thinks context matters when it comes to deciding the MVP. Perfectly fair. This voter might give less weight to WAR and more weight to WPA and leveraged run value. The problem? That also means this hypothetical voter is giving less weight to defense.

You want to know how much more to value a go-ahead home run than an insurance homer? Great, I can show you the stats that measure exactly that.

But if you want to know how much more to value a home run robbery in a tie game than a home run robbery in a blowout? Sorry, you’re out of luck.

I’d like to see that change.

A direct defensive equivalent to WPA or batter run value wouldn’t work. Win Probability Added measures the change in win expectancy from before and after a batter steps to the plate. He is credited (or debited) with the difference. Similarly, batter run value is measured by assigning a run value to the outcome of every pitch and adding it all together.

This works for batters because a batter can take full offensive responsibility for the outcome of every pitch and plate appearance. On the other side of the ball, a group of fielders (plus the pitcher) work together to record outs. You can’t assign full credit or blame to any one player.

What’s more, a defensive equivalent to WPA or batter run value wouldn’t properly measure the difficulty of different plays. It’s harder, and more valuable, to hit a home run than a single. Stats like WPA and batter run value acknowledge as much. In contrast, a high-probability catch and a low-probability catch have the same outcome – an out – so WPA or batter run value wouldn’t treat them any differently.

That’s why I called for a leveraged version of Outs Above Average in the title of this article. (Truthfully, I’d prefer a leveraged version of Fielding Run Value, but Outs Above Average sounds better, and it’s the more well-known metric.)

The brilliant minds at Statcast have already figured out a great system for measuring the difficulty of different defensive plays and properly assigning value to the fielders who make those plays. Baseball Savant also has stats to measure leverage and win probability.

All I’m asking is for those brilliant minds to smush that all together.

Like peanut butter meeting chocolate, I think it would work. And I think it would be amazing.

Now, a metric like this would still undervalue Cowser’s home run robbery. That’s because OAA and leveraged run value don’t account for the theoretical outcome of a missed catch.

Someone much smarter than I am could take this idea a step further and add expected outcomes into the equation. But that would be a much more complicated statistic to create, and I’m not trying to be greedy. I’m already asking someone else to do all the work here. The least I can do is ask nicely.

So, all I want is some way to consider defensive prowess in context, the way I already can for offensive excellence. I don’t think that’s too much to ask. After all, we already have the constituent parts!

Colton Cowser doesn’t deserve a full win above replacement for his home run robbery. However, he does deserve more credit than any stats on Baseball Savant or Fangraphs or Baseball Reference are giving him.

Cowser didn’t just make a great catch. He made a great catch when his team needed it most. Wouldn’t it be nice if we had a statistic to reflect that?