April 8, 2004

Not Earning Its Keep

Why The "Earned" Run Needs To Go

by Michael Wolverton

We at Baseball Prospectus occasionally hear the complaint that we make the game too complicated, with all the numbers and bizarre acronyms we throw around. So today I'm going to do my part to simplify the game. I'm here to suggest that baseball and its fans would be better off without one of its most fundamental, and most complicated, scoring rules. It's time to ditch the "earned" run.

The earned-run rule is widely accepted, or at least tolerated, throughout the baseball world, even in sabermetric circles. There are several reasons for that. For one thing, there's 116 years' worth of tradition behind the rule. I learned the rule because my dad learned the rule because his dad learned the rule, etc. ERA is on the back of every pitcher's baseball card, and it pops up in nearly every baseball-related article or news report you'll see. For another thing, believing in "earned" and "unearned" runs isn't nearly as harmful as, say, believing that RBI are meaningful for evaluating hitting. You have to pick your battles, and in the big scheme of things, this one may not be a battle worth fighting.

Perhaps most importantly, the earned-run rule might have gotten a pass because it's designed to achieve what everyone agrees is a noble goal: separating pitching from fielding. But good intentions aren't enough. The earned-run rule is a lame and counterproductive attempt at solving the pitching/fielding conundrum, one that deserves to be put out of its (and our) misery.

There are many angles from which to attack the idea of "unearned" runs. Here's the one that works best for me:

Pretend for a second that the earned-run rule hadn't been invented over a century ago. In this alternate reality, baseball is played the same as it is now, except there's one fewer column in the box score for pitchers. Pitchers are charged with all the runs they allow, and RA rather than ERA is the de facto standard for evaluating them. Everyone recognizes that good or bad fielding can affect a pitcher's RA, but no one has come up with a good solution to the problem.

Then one day, an enthusiastic columnist writes:

I've done it! I've solved the problem of removing the corrupting influence of fielding on pitchers' RA. We simply pay a sportswriter to sit in the press box, munch Cheetos, and decide which safeties came on plays that should have been made with normal fielding effort. Whenever one of these 'errors' occurs, we reconstruct the inning--not the game, mind you, just the inning--pretending as if the error never happened. Count up the runs that would have scored in this hypothetical reconstructed inning, and you have a revised run total for the pitcher. Things get a lot more complicated for relievers and team totals, and we'll broaden the 'plays that should have been made' definition a little bit, but you get the idea."

For me, shifting the rule's invention into the present day helps illustrate just how absurd it is. If it were proposed today, this idea would be dismissed out of hand, no matter who advanced it. If it had come from the blogosphere, people would use it as an example of the inferiority of the Internet. If it had come from a respected columnist like Peter Gammons or Thomas Boswell, people would wonder if senility was setting in. Absolutely no one would take this idea seriously. So why should it be any different just because the idea originated in 1888 instead of 2004?

It could be construed from our fictitious columnist's quote above that the main problem with the earned-run rule is its reliance on errors. And it's true that errors are one of the problems with the rule. The well-discussed shortcomings of errors as a defensive measurement--their miniscule coverage of defensive plays, their subjective nature, etc.--suggest that ERA isn't doing much to eliminate fielding from pitching numbers.

But focusing on errors is really missing a bigger problem. Even if errors were somehow the perfect measure of fielding performance, the earned-run rule would still be wrong-headed. The main problem with unearned runs isn't errors, it's the notion that the pitcher's job ends whenever an error is made.

The most glaring example of this problem is that the earned-run rule allows the pitcher to give up runs with impunity after the third out "should have" been recorded. We've all seen these kinds of situations. An error allows Michael Tucker to reach with two outs, and Barry Bonds comes up next to deposit one in McCovey Cove. The result: two runs, none earned. That's nuts. Never mind Tucker's run; you're telling me that Bonds wouldn't have hit his homer leading off the next inning if Tucker had been put out? In this situation, the earned-run rule essentially allows the pitcher to skip Bonds in the lineup, as far as his ERA is concerned.

OK, you might say, why not hypothetically reconstruct the entire game, rather than just the inning? That way, Bonds' homer would count as earned. But of course, that would be even more ridiculous than the current rule. Besides the fact that it's way too complicated to reconstruct an entire game this way, you'd frequently end up with absurd situations where the pitcher is charged with more earned runs than actual runs. It's possible, in hindsight, for errors to save you runs, if you buy into this hypothetical reconstruction philosophy.

No, the right solution to this problem is to recognize this simple baseball truth:

Errors will happen. Good pitchers will minimize the damage caused by them. That is, a good pitcher will allow fewer runners on base before the errors happen (so there aren't runners to score on the errors), and will allow fewer hits and walks after errors happen (so the runners who reached on errors won't score).

This isn't a hypothesis, it's a fact. Preventing unearned runs is a skill that pitchers have, and it usually comes hand-in-hand with the ability to prevent earned runs. I took all the pitchers since 1900 who pitched more than 2,000 innings, and compared their earned run averages (ERA) to their unearned run averages (UERA). (In both cases the values were normalized to the league averages for the years in which they played). A couple of results:

• The correlation between ERA and UERA was 0.36 -- not overwhelming, but pretty strong. That means that pitchers who are good at preventing earned runs are also generally good at preventing unearned runs.

• Of the top 50 pitchers in career (normalized) ERA, 46 of them were better than average at UERA.

What this means is that when you throw out unearned runs, you're throwing out part of the pitcher's performance. In other words, ERA is understating the run prevention abilities of the best pitchers in the league, and overstating it for the worst. Measures based on ERA, such as Pete Palmer's Pitching Runs from the Total Baseball books, are going to underestimate the seasonal and career values of baseball's all-time greats.

We noted above that the correlation between ERA and UERA wasn't perfect, so let's look at some of the exceptions. Here are a few of the pitchers who had average-to-excellent ERAs combined with terrible UERAs, along with their rankings (out of 317 total pitchers) in both categories. You may notice from the list of names that they also have something else in common.

                     Rank out of 317
Pitcher                ERA    UERA
------------------------------------
Hoyt Wilhelm             4     295
Dutch Leonard           36     307
Tom Candiotti          115     310
Phil Niekro            125     297
Wilbur Wood            126     302
Charlie Hough          174     301

Coincidence? Not bloody likely. Knuckleballers are a huge exception to the "good pitchers prevent unearned runs" rule. Despite being good pitchers overall, these knuckleballers were all among the worst all-time at surrendering unearned runs. Joe Niekro was the only long-time knuckleballer I could think of who didn't fit the pattern, finishing about average in both categories. (And Joe threw the knuckleball far less often than any of these other guys.) The reason is obvious: Knuckleballers produce a lot of passed balls, passed balls count the same as errors in the unearned-run rule, hence a bunch of unearned runs.

So now the situation is even worse than what we saw above. Not only does the unearned-run rule generally undervalue good pitchers, it's also systematically biased in favor of a particular type of pitcher. Phil Niekro, for example, looks quite different when you evaluate his career based on RA rather than ERA. He's a Hall Of Famer either way, but he's not the Inner Circle guy that Total Baseball makes him out to be if you take his unearned runs into account.

The bottom line is that there's little evidence that unearned runs have much to do with fielding, and there's plenty of evidence that they have a lot to do with pitching. ERA has a long tradition as a pitching metric, but if we replaced it with simple RA, we'd be getting a more accurate measure of a pitcher's contribution. And if the average baseball fan would free up the brain cells he's using to store the complicated earned-run rule, and use those cells to learn about other elements of the game, baseball would be the better for it.