There's some silliness going around. No secret there--that's just about "nature's way" at this point in human history. But the field of athletics is a fertile breeding ground for some of the silliest of the silliness, and another nugget from that gold mine has caught my jaundiced eye. I'm talking about the dreaded "quality start."
In baseball circles, some folks have been throwing that term around for a few years now, trying to define a good starting pitching performance versus a bad starting performance. The idea is, if a starting pitcher goes six innings and gives up three runs or fewer, that's a "quality start," meaning he did his job and gave his team a chance to win the ballgame. This concept started with the stat-heads, but has crept further out into the baseball community, to the point where a "quality start" is now part of the sport's standard vernacular.
This sort of thing is typical of baseball, a sport that loves to quantify everything that can be quantified, for the sake of comparative benchmarks and (presumably) an ever deeper understanding of the game. In this case, however, there's one glaring problem: the definition is remarkably stupid.
I'll back up that statement in a minute, but first let me get a few things out of the way. The definition for a "quality start" should read: "a game in which the starting pitcher pitches a full 6 innings and gives up 3 or fewer earned runs." That's a bit more precise than what I wrote above, and it needs to be for this to make sense. Note especially the distinction about "earned" runs; that plays an important part later.
First, let's take the definition as stated above and extrapolate it out to its reasonable conclusion. Even those of you who know little about baseball can probably tell me how many innings are in an official regulation game: nine innings. The math majors out there can tell you that, last we checked, 6 is not 9. Those 3 earned runs over six innings would not necessarily be 3 earned runs over nine innings. A pitcher who went longer than six innings, even in a "quality start," might give up more runs.
Fortunately, baseball has a way of figuring such things out. It is, after all, the stats game. In this case, we have ERA--"earned run average," which means "the number of earned runs a pitcher gives up, on average, for every nine innings pitched." ERA is quite easy to figure out: just take the number of earned runs, divide by innings pitched, then multiply that figure by 9. For our hypothetical pitcher giving our team a "quality start" as defined above, that means:
3 runs ÷ 6 innings = 0.5 x 9 = 4.50
So our stalwart starter who pitched 6 innings and only gave up 3 earned runs has an ERA of 4.50.
Folks, a pitcher with a 4.50 ERA is nothing to write home about. Back in the day, a starting pitcher with a 4.50 ERA would be setting himself up for a quick exit from the rotation and into the bottom half of the bullpen. Bob Gibson, who posted a record ERA of 1.12 for the entire 1968 season, would probably want to fight you if you patted him on the back for having a 4.50 ERA.
One can argue that the game has changed in recent years: scoring is way up, expansion has diluted the talent pool, and these days a 4.50 ERA really isn't so bad. It's a seductive line of thought; unfortunately, it's just simply wrong. Even today, a team that runs out a five-man rotation of 4.50 arms is not going to win a lot of ballgames. And the numbers back that up.
Think about this: a 4.50 ERA really works out to 9 runs every two games. Since a team can't score half a run in a game, the stat works out to giving up 5 runs in one game and 4 runs in the next. How often do you see a baseball team win a game where the other team scored 5 runs?
But wait--that's not even the worst of it. Because those 5 runs and 4 runs are earned runs. Teams also give up unearned runs, when someone makes an error and such. So those numbers above should actually be higher than 5 and 4. And remember: you can't win the game unless you score more runs than the other team. If your pitchers have a 4.50 ERA, that means you've got to score, on average, at least 6 runs a game, just to win half your games.
Is scoring 6 runs a game really a tall order? Even in these days of big time offensive baseball and watered down pitching? Yes, it is. Case in point: the 2007 Detroit Tigers.
As of this writing, the Tigers are the highest scoring team in baseball. Through their first 83 games, they scored 494 runs, the most in the majors. (All team stats quoted here are taken from Yahoo! Sports as of July 5, 2007.) That works out to 5.95 runs scored per game. That beats the "quality start" standard of 4.50 ERA; it certainly tops their team ERA of 4.39 (better than the "quality start" standard, but fair to middling compared with the rest of MLB). The Tigers are scoring more runs per game (on average) than they allow (on average). Of course, as noted above, teams also give up unearned runs. Detroit has given up 398 total runs, of which 364 were earned runs. That difference of 34 runs works out to 0.4 unearned runs per game--almost half a run, which raises their total runs per game to 4.79. That's still lower than their average runs scored (5.95), and the Tigers have parlayed those numbers into a 49-34 record, good for second in their division, one game behind Cleveland. (Note: the Tigers subsequently won a couple more games and improved their record, taking over first place in the AL Central by the All-Star break.)
So given the Tigers' example, it all works out. You can have an ERA of 4.50 and still be one of the best teams in the league. Right?
Well, yes, but--and it's a big 'but'--the Tigers are the highest scoring team in baseball. And they're still only beating that "quality start" ERA by one run a game. If their "unearned run average" of 0.4 unearned runs a game is typical--and the Tigers are a good defensive team; that number may be superior to the average team--then a 4.50 ERA translates in real terms to 4.90 runs a game. For all practical purposes, that's 5 runs given up every game. That is, you've got to score six runs a game, every game, in order to win. And only Detroit, the highest scoring team in the majors, is approaching that mark. The next closest team, Cleveland, scored 456 runs in their first 85 games; that's 5.36 runs scored per game, much too close to that 4.90 standard listed above to get by with a rotation of "quality starters." And it only gets worse for less accomplished baseball teams.
As a case in point, let's look at the team with an ERA closest to the "quality start" standard of 4.50 ERA. "Ladies and gentlemen, your 2007 Baltimore Orioles," clocking in with a 4.51 team ERA. How's that working out for the O's? Not so good. They've only scored 383 runs in their 84 games--that's 4.55 runs scored per game, barely better than their earned runs allowed. And they've been good in the field, allowing only 21 unearned runs in their first 90 games (not even .25 unearned runs per game). Baltimore parlayed this prowess into a 37-47 record, and saw their manager, Sam Perlazzo, get fired halfway into the season. That's hardly "quality."
The numbers don't lie. No, wait--actually, the numbers often lie. But in this case, the basic math is dead on: coughing up 3 runs every 6 innings you pitch is a recipe for mediocrity. If you need a "quality start" standard, at least make the pitcher go 7 innings. Three earned runs over 7 innings shows a 3.86 ERA--still a little high, but respectable in this day and age. And, most importantly, it's a number that may actually give your team a chance to win. That, and only that, is the true measure of quality.
Addendum: I couldn't let this go without doing a little more numbers crunching to back up my position. Hence, I studied the results from all the major league games played during one week, in this case July 14-20, a total of 100 games. I noted the results for each starting pitcher who exactly matched the "quality start" standard as defined above (6 innings pitched, 3 earned runs allowed), as well as the results for every starter who exceeded the standard, either by pitching more innings or giving up fewer earned runs. The breakdown produced both surprising and not so surprising results.
I began the week sweating profusely, because of the first seven pitchers to match the standard, all of them saw their team win their games.
(Note: in all cases listed here, the final result I cared about was whether or not the team won the game, not whether or not the starting pitcher got the decision. The whole idea behind the quality start is that the pitcher in question gives his team a chance to win. The basic tenet is not about individual statistics.)
I was genuinely shocked by these results, and was grateful when the next five exact matches all lost their games. The remaining exact matches rebounded with three more wins. Ultimately, fifteen starting pitchers exactly matched the standard, with their teams going 10-5 (a .667 winning percentage) after those starts.
That's a pretty good record. Win two thirds of your games each season and you'll probably win the World Series. Surely that result backs up the logic behind the "quality start" definition?
Well, maybe. It's suggestive, but those numbers also come from a very small statistical sample. Only fifteen starts can hardly be considered definitive. Without pursuing further research, it's hard to say what the long-range record would be for six innings with three earned runs a start.
However, what's also suggestive are the results for starters who exceeded that "quality start" standard. Starters who pitched better than six innings, three earned runs saw their teams go 56-27 (.675) in those games. That's already a better percentage than the exact matches, and the results come from a much larger sample (eighty-three games vs. just fifteen). Clearly, doing a little more helps.
But wait--it gets better. Because among that 56-27 record, there are twenty-one pitching duels--games where both starting pitchers exceeded the standard. That's twenty-one games where one pitcher was automatically going to see his team pinned with an "L" despite his efforts to exceed the "quality start" standard. Presumably, those twenty-seven losers would have had a better chance for a "W" if the other starter had not manned up and hurled a gem of his own.
Taking the duels out of the above record confirms this idea. In games where only one starter exceeded the standard, the teams' records clocked in at a dominating 35-6 (.854). The evidence is crystal clear: if your guy steps out to the mound and throws for more than six innings, and/or gives up fewer than 3 earned runs, you're going to win--unless the other team's guy runs out there and goes him one better.
Overall, the numbers from a week of baseball show that a good starting performance will usually result in a victory. No surprise there. But what the numbers also demonstrate backs up the bottom line of my argument above: in 100 games played, 62 of those games saw at least one starting pitcher exceed the paltry standard of 6 innings pitched with 3 earned runs. You can not characterize anticipating a better performance than the standard as an unreasonable expectation. A starter who can't go out and consistently pitch seven innings, with 3 runs or less allowed, does not belong in your rotation. Period.
Quality matters. But you're not going to get quality from your starting pitchers if you don't truly ask for it.
Posted 7/15/07; Revised 8/8/07