David R. Henderson  

Overdetermination vs. Randomness

Henderson on Landsburg's "Armc... Bailouts are for Bondholders...

Here are the opening sentences from an article by San Jose Mercury News sports reporter Tim Kawakami:

All the melodramatic twists, tweaks and breaks should've made the Giants virtually unbeatable on Saturday.
They had their ace, Matt Cain, on the mound, and the Reds lost their ace, Johnny Cueto, after eight pitches.
The Giants were at AT&T Park, their beloved and spacious home, and the scene of so many 2010 playoff raves.
Game 1 of this NLDS was all set up for the Giants, yes it was.
Until Cain got hit, the Giants' bats and home crowd stayed mostly silent, they lost 5-2, and suddenly the plot got flipped on them.
Now the Giants are the team in a scramble for answers to questions that they thought they might never be asked in this best-of-five series.

Why do I quote them? They're similar to many lines that sports reporters write, right? Exactly.

And that's why I quote them. This is the kind of report you read all the time and it's presumably by someone who should be expert at following game he's writing about.

But he's not. Why? Because he doesn't get randomness. For Kawakami, a few advantages the San Francisco Giants had against Cincinnati should have been enough to make the Giants "virtually unbeatable." Now, if this had been a 10-year old writing, someone who had never really paid attention to baseball, I could understand it. But this is someone who has followed the game for years. Has he ever seen a playoff game won by the visiting team? Of course he has. Has he ever seen a high-quality pitcher come in and pitch less than a high-quality game? Sure.

But there's no learning.

For more on randomness in sports, see my NCAA Outcomes: Margins and Randomness and my Moneyball and Randomness.

Comments and Sharing

COMMENTS (6 to date)
BLM4L writes:

To be fair, given Laplace's demon, I think you have to acknowledge that "random" is just a term to describe something that we just don't understand fully. A pitcher may have a subpar game for a variety of reasons (e.g. weather, scouting, dyspepsia) and it is the sportswriter's job to find out (or at least speculate) about the cause of this specific subpar game. I don't think a sportswriter can just say "well, when you win 2/3rds of your starts, you lose 1/3rd, and variance means that today was in that losing 1/3rd. The end."

Ken B writes:

I had similar thoughts yesterday listening to the announcers on the Tigers' game. Stuff like Player X was batting 600 after he swung at the first pitch from left handers in the early innings. There is something both appalling and lovely about using naive and bogus statistical reasoning to gin up excitement in an audience. It's awful but it's also the closest math gets to cool and sexy.

Richard writes:

Hypothesis #1: We economics-types are enlightened while the sportswriter is an idiot.

Hypothesis #2: The sportswriter intelligently responds to the incentives in his profession, and "beaten by randomness" makes for terrible copy.

I'm going with #2

(To be fair, BLM4L suggests something similar at the end of his comment above, but the force is diluted by his linking it to an unrelated point about limited knowledge.)

David R. Henderson writes:

Good point. And I'm trying, on the margin, to change his incentives. Also, I'm using his piece to make a point about statistics that our readers can learn from. To the extent I get people thinking more clearly about randomness, they change his incentives too.

BZ writes:

This reminds me of the old joke "Last weekend I took my girlfriend to see E.T. During the scene where the kid is flying through the air on his bike, she said 'Yea, right.'. To which I replied 'Honey, it's not live footage.'"

My point being, sports writers are not like economics writers trying to disseminate knowledge, but entertainers trying to touch something in readers that will make them come back for more. The real question should be "What do sports writers know about human nature that we don't."

David R. Henderson writes:

I confess that I'm almost never of the opinion that there's one "real question." Rather than reply de novo to your point, though, I recommend that you see my reply to Richard, above, making a similar point.

Comments for this entry have been closed
Return to top