Once more with gusto: Forget about September!
With the Royals heading to the playoffs again, people are going to bring up their 11-17 record in September.
If the Astros do squeak into the playoffs, people will mention their 11-16 record in September.
Conversely, it’s already fashionable to cite the Rangers’ tremendous finishing kick; the same will go for the Angels if they ace out the Astros and the Twins for that second wild card.
Well, pardon me if I’m not worried about the Royals and Astros, or particularly impressed by the Rangers and Angels.
In 1971, the Orioles finished the regular season with 11 straight wins, then swept the A’s in the ALCS … before losing the World Series to the Pirates in seven games.
In 1977, the Royals were in third place on the 20th of August … and from there, they went 35-9 — including an incredible 24-1 stretch — to win the American League West, going away. Of course, then they lost (again) to the Yankees in a heartbreaking (for me) ALCS.
The Mariners in 1995, the Twins in 2003, the Twins again in 2006, the A’s in 2012 … all these teams played tremendous baseball down the stretch, only to lose their first postseason series. Well, except for the Mariners, who squeaked by the Yankees in their Division Series before losing to the Indians in the ALCS.
Okay, so all these are just data points, and not many data points at that. Fortunately, there are a lot more data points. From Ben Lindbergh’s Grantland piece last year:
In 2009, BP’s Jay Jaffe analyzed every 1995-2008 playoff team’s performance over the final week, two weeks, three weeks, and month of the season, then tried and failed to find evidence that a team’s record over any of those periods predicted division series play.
Earlier this month, Dave Cameron tackled the same question for Fox Sports, examining the correlation between second-half winning percentage and postseason winning percentage for 1995-2013 playoff teams. He found none, even when he looked at extreme examples. “The five teams that stumbled into the playoffs with the most meager second-half records combined to go 25-21 in the postseason; the five teams that played like the ’27 Yankees for the final three months of the year went 20-19,” Cameron wrote.
Yes, of course it’s counterintuitive. If nothing else, wouldn’t we expect the personnel in September to more accurately reflect a team’s true quality than its personnel from April through August?
Well, maybe April. A little. But the personnel differences between September and the rest of the season are small enough that they’re overwhelmed by the natural variance and random fluctuations experienced by a team over the course of 25 or 30 games.
So just in terms of wins and losses, the only predictor worth anything at all is a team’s record over the whole season. Same as it ever was.