Tuesday, June 30, 2009

Data-Mining 1.6 Million Putts for Irrationality

A recent study by Professors Devin Pope and Maurice Schweitzer of the University of Pennsylvania concluded that professional golfers are more likely to make the same putt if it is for par than for birdie. (Par is the expected score for a hole; anything worse is bad. Birdie is one stroke better than par, which is good.)

Duh. Par putts are usually shorter than birdie putts because the player has used an extra shot to get the ball closer to the hole.

The study only compared par and birdie putts of the same distance. It used a database of 1.6 million putts, accurate to the inch, from Professional Golf Association tournaments between 2004 and 2008.

What about position on the green? Greens have lots of different slopes, so the same-distance putt can be easier or harder depending on where it is. If you took an extra shot to reach your position on the green, as the par-putter did compared to the birdie-putter, you should have been able to pick a more favorable spot.

The authors accounted for that. Position on the green indeed led to variance of putting performance overall but did not have a significant effect on birdie versus par putting performance.

Well, par putts are often second putts, after a player has missed a birdie putt. That means the player has already seen the green’s speed and angle from the first putt. Even if the distance is the same between a birdie and par putt, this extra learning will help sink more par putts.

It does, but it only accounts for 20% to 30% of the difference. The authors controlled for that effect, as well as a player’s learning from watching other players’ putts on the same green.

Hmmmm, sounds like the authors were thorough.

We’ve only covered some of the effects the authors tested and controlled for.

But how widely does the finding hold?

The authors tracked 188 professional golfers. All of them were better at par putts compared to equivalent birdie putts.

How much better?

Other than for the shortest putts, which were rarely missed, the difference varied between two and four percentage points, depending on distance. The best players tended to have less difference, although it was still statistically significant.

And why does this matter?

First, it’s a great example of a rigorous data analysis, the kind that can handle challenges like those above with aplomb.

Second, the conclusions are not just about golf. They have wider applicability to the debate about whether people are predictably irrational. In the golfers’ case, they treated the exact same putt differently, even though it was worth one stroke either way. The authors hypothesized that the golfers were exhibiting a flavor of irrationality known as loss aversion: The players perceived a missed par putt as a loss because par is expected, whereas a missed birdie putt was more like an unrealized gain. Standard economics suggests that people do not act differently based on such perceptions; the study suggests they do, even if they are highly skilled professionals.

Where can I read more?

The academic paper is available, but it’s not exactly a beach read. For a more approachable summary, see this New York Times article.

No comments:

Post a Comment