Run differential a flawed statistic
Orioles' winning ways this year, D-backs' run to NLCS in 2007 stand as proof
Those D-Backs had committed an unpardonable sin. They had been a winning team, a division-winning team, a team, in fact, that eventually advanced to the National League Championship Series. But they had been outscored during the regular season, 732-712.
For devotees of run differential as a principle governing life on planet Earth, this was an event that went beyond the realm of the improbable into the impossible. How could this have happened? Maybe it didn't. Perhaps it was a mass hallucination.
But those D-backs could pitch, at least relative to the competition. They were fourth in the NL in team ERA. True, they had problems with the other half of the game, finishing 14th in runs scored. But their pitching kept them in games, and it was widely agreed that Bob Melvin, who managed the D-Backs that season, did a splendid job of maximizing his team's resources. Melvin, now managing the A's, was justifiably named NL Manager of the Year.
You might have thought that an episode of this sort could have stopped the run-differential disciples in their tracks, but it didn't. Its adherents merely dug in for the long haul, produced examples favorable to their argument and refused to budge.
This summer, though, there is another high-profile exception to a supposedly inviolate rule. The Baltimore Orioles, through 116 games, have been outscored, 531-488, yet have not only a winning record (63-53) while occupying second place in baseball's toughest division, the AL East, but are in qualifying position for a Wild Card berth.
Again, the run-differential mathematics scream that this simply cannot be. But it is. Maybe this is merely a transitory condition. Some of us have doubts about the Orioles' staying power, but the doubts are not grounded in run differential, but in questions, for instance, about Baltimore's starting rotation. The fact that the Orioles lead the Major Leagues in errors also does not automatically seem to be a ticket to the postseason promised land. But this sport, honestly, is not quite an exact science. Therein lies part of its beauty.
The Orioles, to their credit, have thus far defied the doubters and played a successful brand of baseball, even if the whole experience has been statistically unlikely. But the fact that somebody could go 63-53 while being outscored by 43 runs tells you something about the advisability of worshiping at the run-differential altar.
This discussion reminds us of the notable quotation from the 19th-century British statesman, Benjamin Disraeli: "There are three kinds of lies: lies, damned lies, and statistics."
Much less well-known, by the way, is another Disraeli quote: "As a general rule, the most successful man in life is the man who has the best information."
So, this fellow was not advancing the cause of superstition and ignorance. I think that if Disraeli were alive today, even he would find room in his "best information" category for on-base percentage. But not run differential.
Run differential is biased in favor of high-scoring teams in hitter-friendly parks, bashing homers and running up scores. It is a statistical outgrowth. Cause should not be confused with effect. Or, as one National League general manager put it:
"It starts at zero. The pitchers can't get better than that. It's tilted toward teams that score a ton of runs."
And it is all aggregate information, which won't help you win tonight's game. "You can score 11 runs one night and one run the next," an NL manager said. "You'll be averaging six runs a game, but you still won't be going anywhere."
I believe that any statistical measurement that predicted disaster for a team that won 90 games and a division title comes in somewhere south of perfection. Run differential is not an immutable scientific force.
But rest assured that if the Orioles don't hold up, there will be people saying that run differential saw it coming. There is a chance, though, that if the Orioles don't hold up, the real issue could be starting pitching.