Outs Above Average: A Test Case in Statcast’s Newest Number

A fun thought thing to talk about, if you’re the kind of person who enjoys angry arguments about baseball statistics, is the 2016 stat lines for Jason Heyward and Matt Kemp:

Jason Heyward vs. Matt Kemp, 2016

If you look at the triple-crown stats, Kemp had a great year and Heyward was awful. Even by more advanced metrics like OPS and OPS+, Kemp was clearly a better offensive player in 2016 than Heyward was. Get as advanced as you want, and every stat will tell you that Kemp was a better hitter than Heyward last year. You like wOBA? Kemp’s .333 is significantly better than Heyward’s .282. Fan of wRC+? Kemp leads 109 to 71. Even if you just look at the offensive side of WAR, Baseball-Reference has Kemp at 1.8 and Heyward at -0.3.

So while Kemp’s 2016 was pretty lousy compared to most 35-homer seasons — he is one of only 37 players in MLB history to hit 35 homers and have a slugging percentage below .500 — he was indisputably a better hitter than Heyward last year.

So how, a WAR skeptic might ask, does Heyward have a significant edge in both versions of WAR? The bulk of that answer is one word: defense.

(For the record, let’s get this clear: Heyward was not good last year. Baseball-Reference’s little WAR cheat sheet lists a WAR between 0.0 and 2.0 as a “Sub,” meaning a bench player, a non-starter, a guy who doesn’t get 592 plate appearances on a team that wins the World Series. Heyward was bad. Just not as bad as Kemp.)

Yesterday, the fine folks at Statcast released a cool new stat: Outs Above Average. Here’s how they explain it:

Outs Above Average (OAA) is the cumulative effect of all individual Catch Probability plays a fielder has been credited or debited with, making it a range-based metric of fielding skill that accounts for the number of plays made and the difficulty of them. For example, a fielder who catches a 25% Catch Probability play gets +.75; one who fails to make the play gets -.25. Read more about how Outs Above Average works here.

Expected Catch Probability expresses, based on the difficulty of balls hit to the fielder, how many an average outfielder would have caught.
Actual Catch Percentage is the actual performance of the particular fielder on those plays.
Catch Percentage Added is the difference between the two, showing how much the fielder added (or didn’t) based on the opportunities he was presented with.

Among advanced metrics, this one is a bit easier to grasp than most. There are plenty of intricate details that go into determining “catch probability,” but this is basically just a way to define, statistically, the concept of a “nice play.”

Let’s look at Kemp and Heyward from last year:

Heyward vs. Kemp, 2016 Defense

Among the 208 outfielders with at least 25 chances, Heyward was seventh in Outs Above Average. Kemp was 208th, aka “last.”

We have always known that defense matters, but the question has been: “How much?” For outfielders, anyway, Outs Above Average is a pretty big step forward in quantifying that, and the Kemp vs. Heyward discussion is an excellent test case.

For the rest of this discussion, let’s stipulate one thing: an out is an out. Every team gets three outs per inning. A batter’s main job is to not make an out, and a fielder’s main job is to make an out. Making an out on defense is roughly as valuable as not making an out on offense.

So let’s do a quick, clunky merge of Kemp’s and Heyward’s defensive outs into their offensive stats. Heyward made 13 more defensive outs than average, and Kemp made 26 fewer. So what we will do is add 13 singles to Heyward’s offensive line and subtract 26 singles from Kemp’s. That gives us this adjusted slash line:

Heyward vs. Kemp, Clunky Adjusted Slash Line

Suddenly, our 172-point gap in OPS becomes a 45-point gap. The main criticism of OPS is that it lumps OBP and SLG together as if they are equivalent, when in fact OBP is quite a bit more important than SLG. So Heyward’s 63-point advantage in OBP probably more than makes up for his 108-point deficit in SLG.

And then there’s one other factor: We’re being pretty conservative on the value of defense here. Quite often, when a good outfielder makes a catch that a lesser outfielder would not have, he’s preventing an extra-base hit, not just an ordinary single. I don’t have the data to know how often, but surely our clunky adjustment of calling every saved play a single was conservative. If we count each one as 1.5 bases instead of a single base, that bumps Heyward’s SLG up to .364 and drops Kemp’s down to .437, and suddenly their OPSes are almost identical.

Heyward and Kemp both made about $21.7 million last year, and the only correct answer to the question “Which would you rather have?” is “Neither.” They were both bad. But Heyward’s defense more than made up for Kemp’s offense, and thanks to Statcast, we can start to quantify that.