May 6, 2004
Not Earning Its Keep, Part II
More on the Unearned Run
One big issue I didn't address when I wrote about the wrong-headedness of the earned run rule last month is the idea that, while the rule may have outlived its usefulness today, it was necessary and meaningful in the error-filled early days of baseball. An old friend, Steve Thornton, put the argument well in a recent letter:
Your article on UERA, and the follow-up piece in Mailbag, are interesting. While I agree with you that the current system hasn't made a lot of sense for the past 50 years or so, I think you're missing, or glossing over, the history of the earned run.
The reason: errors. Players hardly make them anymore. But in baseball's formative years, when players didn't wear gloves, errors were EXTREMELY COMMON, almost as common as hits (Cincinnati in 1876 had 555 hits, 576 errors); and when they started wearing tiny blob-like gloves with no webbing, errors continued to be very high, and in fact right up to around the WWII era, if my memory of error rates isn't failing me completely, they were much more common than they are now. The history of error rates is a nice downward-sloping curve, though of course it's very nearly flat now.
But when ERA was invented, those errors MATTERED. And when you look at old-time pitchers, you really do have to adjust for defense somehow. A lot of errors today are pretty tight judgement calls that frequently miss the point, but you simply don't have fielders routinely dropping or muffing balls hit right at them like you used to. I don't mind saying that Roger Clemens should be on the hook for his UERAs, but for Walter Johnson or Cy Young, it's just not fair.
This argument makes a lot of sense on the surface. Steve is absolutely right, of course, about the game's changing error rates. There were four times as many errors per game in 1900 as there were in 2003, and the rate of unearned runs was almost five times as high. And errors and unearned runs were even more prevalent in 19th century baseball. In an era where fielders are dropping balls left and right, didn't the earned run rule make a lot of sense?
No, it didn't. It's important to distinguish two arguments here. One is that errors were once a reasonable measure in the webless-glove, rocks-in-the-infield days of yesteryear. The second is essentially the same argument, except about unearned runs. These two arguments are often treated as interchangeable, but they're not. There's plenty of room for debate on the first argument, but we're not going to have that debate here. This is about unearned runs, not errors. And the key thing to know about unearned runs is this: The earned run rule does not logically follow from the error rule. The earned run rule is a monster unto itself. It's partly built on errors, sure, but it's also built on this philosophy that you can remove the effect of an error by hypothetically reconstructing the inning pretending as if the error never happened. It's that philosophy that's the main problem with unearned runs. And that philosophy was just as wrong a century ago as it is today.
In fact, it was arguably more wrong a century ago, at least in the impact it had in pitching evaluation. The role of pitching in determining unearned runs may well have been even stronger in the old days than it is today. If we compare earned run rates to unearned run rates for pitchers with 2,000+ innings pitched, and break it down by era according to the year the pitcher's career started, here's how well they correlate:
Years ERA-UERA Correlation ----------------------------------- 1880-1899 0.49 1900-1919 0.52 1920-1939 0.23 1940-1959 0.35 1960-1979 0.22 1980-2003 0.43
The relationship between ERA and UERA was actually at its strongest in the error-happy old days. This might seem surprising at first glance, but it makes sense when you think about it a little. As we argued in the last article, there's very little reason to think that unearned runs are about fielding. The earned run rule--especially the part about ignoring all runs that score after the third out "should have" been made--really only serves to wipe out a nearly arbitrary subset of a pitcher's runs when errors (inevitably) happen. More errors mean a bigger subset--i.e., a larger sample size of unearned runs. So the relationship between ERA and UERA was strongest when pitchers were giving up a larger number of unearned runs, smoothing out some of the randomness that exists in the smaller samples of unearned runs in the post-war era.
Contrary to the notion that the earned run rule is necessary to be fair to Walter Johnson and Cy Young, those guys and the other old-timers make one of the strongest arguments in favor of dumping the rule. While unearned run prevention is a relatively small part of today's game, it played a huge role for pitchers of Johnson's and Young's vintage. Measures of run prevention based on earned runs alone, like Pete Palmer's Pitching Runs, are ignoring a significant portion of the old-timers' value by ignoring the unearned runs they prevented.
If you measure unearned runs prevented compared to an average pitcher, all of history's best totals come from baseball's early days (and nearly all of them come from the era's top earned run guys):
Pitcher Years Unearned Runs Prev. ------------------------------------------------- Cy Young 1890-1911 328 Charley Radbourn 1881-1891 275 Kid Nichols 1890-1906 273 Tim Keefe 1880-1893 249 Jim McCormick 1878-1887 249 Tommy Bond 1874-1884 230 Bob Caruthers 1884-1892 225 Tony Mullane 1881-1894 198 Charlie Buffinton 1882-1892 180 Pud Galvin 1875-1892 178
A better way of showing the disproportionate impact on the old-timers is to compare their run prevention numbers to those of modern pitchers. Below, the first column shows run prevention (compared to league average) based on earned runs alone, the second column shows run prevention based on all runs allowed, and the third shows the difference between the two. Here are the numbers for five of baseball's early greats, and five more recent stars:
Runs Prevented, Based On: Pitcher Years ERA RA Diff ------------------------------------------------------------- Cy Young 1890-1911 799 1127 328 Kid Nichols 1890-1906 618 891 273 Walter Johnson 1907-1927 669 818 149 Pete Alexander 1911-1930 508 651 143 Christy Mathewson 1900-1916 398 523 125 Roger Clemens 1984-2003 593 648 55 Tom Seaver 1967-1986 414 513 98 Pedro Martinez 1992-2003 436 468 31 Jim Palmer 1965-1984 326 387 61 Bert Blyleven 1970-1992 312 375 63
Looking at earned run prevention alone underestimates the value of modern pitchers by a relatively small amount, but it really robs the old-timers. Tom Seaver is by far the top unearned run preventer since WWII, but even he doesn't approach the older stars' unearned run numbers. Cy Young prevented more unearned runs than most Hall Of Fame pitchers prevented runs. Ignore those runs, and you ignore a huge part of his value. The arguments against the earned run rule apply regardless of how many errors are being made. It was a bad idea then, and it's a bad idea now.
I'll be hosting a chat tomorrow at 3pm EST (not 9pm as we had it for a while). Stop by to talk about unearned runs, spider webs on the bases, Bud Selig's hair, and other communist plots.