Wednesday, October 06, 2010

Yay for Random!

When the Padres lost on Sunday afternoon and the Braves were officially into the playoffs for the first time in five years, my first thought was "how in the world are we going to score any runs off of Matt Cain and Tim Lincecum with this patched-together lineup?" My second thought was "if we somehow get past the Giants by winning a bunch of games 2-1, how are we ever going to beat the Phillies, given that the Phils just lit our pitchers up and they didn't throw any of their big three?" Then I remembered that the baseball playoffs are a random number generator, as we learned the hard way over the course of a decade as superior Braves teams routinely lost in short series to opponents who had finished a mile back over the course of 162 games.

Mark Bradley thinks that the quality of pitching has something to do with the results:

Bobby Cox, who has taken the Braves to five World Series, says it himself: “We played better in three Series we lost than we did in the one we won.” He refers to 1991, when the Braves outscored Minnesota but lost Game 7 in 10 innings; to 1992, when the Braves outscored Toronto but lost Game 6 in 11 innings, and to 1996, when the Braves outscored the Yankees, but blew a six-run lead and lost Game 4 in 10 innings.

And now we recall the Series the Braves won: They faced Cleveland in 1995, and the fearsome Indians — with Kenny Lofton, Albert Belle, Jim Thome and Manny Ramirez — had hit .291 as a team. (The Braves had hit but .250.) The Braves won in six games. The Indians’ batting average in the Series was .179.

And there you have it, the reason the baseball playoffs seldom track a coherent course: Because pitching skews the results. Nobody gets to hit off a No. 5 starter in October because rotations become three- or four-man things. Almost nobody sees a truly lousy starting pitcher in postseason, and the bullpens tend to be good, too. That’s why October is, to borrow the baseball argot, such a crapshoot.

I think that the explanation is simpler: we are talking about a bunch of short series, so anything can happen. Seattle was the worst team in the AL this year and they had six-game winning streak. Pittsburgh was the worst team in the NL and they had a five-game winning streak. Unless you think that teams can control when they hit hot streaks, the baseball playoffs come down to an exercise in hoping that your team has the good fortune to hit a hot streak at the right time of the year. Better teams are more likely to hit hot streaks, so it's a little more likely that the Phillies will hit a hot streak, but we're talking about probability (and not strong odds, at that). Thus, you have Bradley's examples of the 2006 Cards and the 1988 Dodgers winning despite having very flawed teams. Those sorts of events don't happen all the time, so it's not like we can rationally expect the Braves to roll in October, but we have a puncher's chance.

Additionally, the Braves' poor September doesn't mean a thing if you believe in precedent($):

A look at "hot" teams heading into the postseason uncovers this truth: Momentum is a myth. Of the 20 playoff teams in the wild-card era that went 8-2 or better over the final 10 regular-season games, only two won the Series (the 2005 White Sox and 1998 Yankees). That's right, 18 of 20 sizzling teams flamed out. The flip side is just as revealing. Three of the seven playoff teams that stumbled to a 3-7 finish or worse during the same period won it all. Now check out a fascinating study by Dave Studeman of the sabermetrics website The Hardball Times. Ten years of data show that when teams face off in the playoffs, September records actually have a negative correlation with winning the series.

Fans, though, remain fooled by what's called "recency bias," the tendency to put greater emphasis on what has just been witnessed. "Confirmation bias" is a related psychological culprit, causing humans to favor info that supports their existing beliefs. For example, those who remember the Rockies' 2007 run often conveniently forget the 2000 Yankees, who lost 15 of their final 18 regular-season games, then won yet another Series.

I'm not sure that I totally buy the notion that the Braves' September slump is irrelevant. The slump was caused in no small part by Martin Prado's injuries, combined with the fact that Chipper was already out. Voila, the team is missing two of its best hitters and the offense goes into the tank. Pair that with an injury to Jair Jurrjens on top of Kris Medlen having to go through the Tommy John rite of passage and the rotation was also whittled down by the end of the year. Then you have bullpen fatigue as a result of Bobby's usage patterns. The bullpen issue remains a concern going into the Giants series (specifically Jonny Venters, although Craig Kimbrel's emergence is encouraging). The depth issue for the rotation isn't as big a deal in the playoffs, although the prospect of Brandon Beachy starting a do-or-die game four is a concern. (Would a rusty, gimpy Jair Jurrjens be a better prospect?) The lineup issue is a major concern. This team has holes aplenty. So, the question is this: do we buy the meta numbers that show that a team's form going into the playoffs doesn't matter or do we believe our lyin' eyes?


Jeff said...

Color me a cynic but it seems like many of these arguments have big holes in them. For instance, part of the reason that team's September record doesn't indicate playoff success is that teams that wrapped up a playoff berth a while ago rest their key players/pitchers, etc. like the Ryas and Yanks did this year. By contrast, teams that make the playoffs based on a hot streak often just aren't that good but got hot.

Also, looking solely at teams' records over 162 games is misleading because regular season records evidence a combination of talent, depth and luck that doesn't translate to playoff success. To take your point about a fifth starter a bit farther, a team with two flame-throwing aces and three crappy starters is much better off in the playoffs than a team with five average or even above-average starters.

It seems like there's a better test than the stats sportswriters discuss--Vegas odds. I'll bet that teams Vegas thinks are more likely to win in the playoffs do win in the playoffs much more often than the teams Vegas doesn't think will win. But I'm too lazy to check.

Anonymous said...

That random number generator has been generating repeat numbers a lot: Twins lose (10 straight playoff game losses). Even more to the point, Twins lose to Yankees in game that Twins lead early, with at least one atrocious missed call (4 straight).