What can stat geeks tell us about what it takes to win on the links?

The stadium scene.
June 18 2009 1:23 PM


What can stat geeks tell us about what it takes to win on the links?

Illustration by Robert Neubecker. Click image to expand.

Tiger Woods is standing on the 18th green at Bay Hill, a golf resort in Orlando, Fla., owned by Arnold Palmer. He's wearing a form-fitting red Nike shirt and a black baseball cap. Widening his eyes, he gauges a 24-foot putt that he needs to sink to win the tournament. The announcer whispers—why do golf announcers whisper?—the following statistic: "It probably means nothing right now, but he is 0 for 21 this week on putts over 20 feet." Tiger strokes the putt. He sinks it! Hello, Ben Hogan! Just further evidence that Tiger is a great clutch putter, and clutch putting wins tournaments.

Or not. As golf researcher Mark Broadie has explained, professional golfers make only 15 percent of putts from more than 20 feet. So Tiger was actually below average in long putting at the Bay Hill tournament. While making the putt took great skill, it was some combination of other skills—most likely Tiger's extraordinary driving yardage, iron play, and good putting from other distances—that placed him in a position to win.


Scholars like Broadie have been looking at golf and trying to answer a simple question: Why do golfers win? They write papers with titles such as "Gender, Skill, and Earnings in Professional Golf" and "Modeling the Determinants of a Professional Golfer's Tournament Earnings: A Multiequation Approach." This week, the New York Times featured the work of two Wharton professors who analyzed 1.6 million putts and found that pros missed more when going for birdie than for par on putts of identical length. They chalked this up to "risk intolerance" (i.e., fear of making a bogey) and calculated that it cost the top 20 golfers significant prize money each year.

As that 1.6 million figure suggests, golf has rich data. Since 2002, the PGA Tour has deployed a system called ShotLink, in which a platoon of volunteers uses laser-measuring devices to record each and every shot. (The tour claims that its greenside laser is accurate to within 1 centimeter.) Before ShotLink, pro golfers had what you might call a King Lear problem: They but slenderly knew themselves. Given that one's score might vary by only a few strokes per round, it's hard for a golfer to detect where he is losing ground with his peers. With ShotLink data, they can now discern clear trends. Phil Mickelson, for example, realized his ability to get up and down from the sand was subpar. In two years, he improved dramatically, moving from a ranking of 180 to 3 in sand saves.

So golf stats certainly help the pros fine-tune their game—but what about the fan at home who wants to put a little money down? Is there a killer stat to look for? Probably not.

In 2005, economics professor Stephen Shmanske published "Odds-setting Efficiency in Gambling Markets: Evidence From the PGA Tour." His determination: "Unfortunately, readers of this paper will not be getting rich anytime soon by handicapping professional golf." The odds-setting strategies that casinos have been using are sound. When Tiger was in the field, he was the favorite. When Tiger wasn't playing, the field was the favorite. Shmanske found that the odds captured the four skills that consistently win tournaments: driving accuracy, driving distance, greens-in-regulation, and putts per green. (Hitting a green-in-regulation means that you reached the green of a par 3 in one stroke, a par 4 in two strokes, and so on. It's a measure of accurate iron play.) He concludes by offering a slim hope that "other variables" such as the "hot-hand tendency" (golfers on a win streak) or "expected weather patterns" (some guys play better in the wind) may offer odds-beating predictive power.

What Shmanske's paper does offer is predictive power about golf studies in general: Tiger Woods warps everything, and, despite the explosion of golf data, a mere four skill stats have the greatest correlation with professional success.

My favorite example of the Woods effect came in the paper "Match Play: Using Statistical Methods To Categorize PGA Tour Players' Careers," which divided players into groupings of Elite, Distinguished, Established, Journeymen, and Grinders. (It's hard out there for a Grinder. The authors conclude: "We found that players on average spend significantly less time improving on their rank and more time staying the same or getting worse.") Woods' career stats are so good that he's statistically very close to forming a group of one. (Godlike?) He is literally a league of his own.

Though the four skills that correlate with winning have been established in the literature, there is debate about how influential each of them is. Specifically, has increased driving distance altered the traditional formula for victory? The short game—any shot from within 125 yards—has traditionally been more prized as a skill set than crushing the ball off the tee. "Drive for show, putt for dough," is the timeworn adage (usually said after your partner hits a great drive). But what if the ability to drive the ball 300-plus yards gives a player a significant advantage? An article in Chance, themagazine of the American Statistical Association, finds that greens-in-regulation and putts-per-round are still most crucial. But there is some evidence that, as players have gotten longer off the tee, driving accuracy is now less important.