Why are golf statistics so unhelpful?

I remember several years ago Tiger was no. 1 in scoring and no. 145 (or so) in putting. That struck me as not just odd, but virtually impossible. The baseball equivalent would be Chipper Jones leading the NL in hits while batting .225. Those sorts of statistical disparaties don't happen in baseball.

Where did the collection of golf stats go off the rail? Why are they so uninformative? I assume part of it is that the wrong information is being collected.

I bring this up on a gca site because measuring how pros perform is also a good way to measure how well a golf course "performs". Informative stats would shine a light on whether a course's strategic features held up, what shots it forces the pros to compromise, and so forth.

Golf needs a Bill James.

Bob
 

I don't know the details of how you purchase access to the ShotLink data but for the past couple season's PGA Tour events there is a wealth of useful information being collected.

My question is why the TV networks, the Tour or the magazines are still promulgating the same old garbage stats when real data is available. Occasionally on a TV broadcast you'll see a quick graphic of how often a player hits 150-yard approach shots within 20 feet of the hole or whatever but 90% of the time it's GIR, driving distance, putts per round and a bunch of jawboning.

When the Bill James of golf stats arrives he will be using the high-priced actual shot-outcome data not the crude counting stats that are widely available.

Bob,
Have you read Moneyball?  If not, I think you will love it.

I agree that it would be nice to see someone come up with a golf stat that really correlates to low score, but the game may be too subtle for that.  With every hole being different, every course being different, and every shot being different, I'm not convinced it can be done.

BTW, for those here not familiar with Bill James, the whole premise of James is that traditional baseball stats like batting average obscure truth rather than reveal it.  He has come up with a collection of new stats, such as runs created, to more accurately assess the impact of a particular player on winning and losing baseball games.  These stats are being used by a number of major league teams, most notably the Oakland A's who are the subject of the book Moneyball, to win consistently despite very, very low payrolls.  It is fascinating stuff, especially if you tend to be a gearhead.

Why are golf statistics so unhelpful?

I remember several years ago Tiger was no. 1 in scoring and no. 145 (or so) in putting. That struck me as not just odd, but virtually impossible. The baseball equivalent would be Chipper Jones leading the NL in hits while batting .225. Those sorts of statistical disparaties don't happen in baseball.
 

Bob,
I think I remember a thread here about this very statistic concerning Tiger.  My recollection is that the consensus was that Tiger had hit lots more par 5's in two shots, but was farther from the hole that another player who laid up and hit a wedge in.  Statistically, Tiger would then be much more likely to 2 putt, though finishing with a birdie most of the time, that the shorter player that made some amount of birdies by one-putting, but usually two-putted.

In any event, you are right; the putting statistic (even limited to GIR) MUST be obscuring something important that is really happening, since the 145th best putter couldn't possibly be the best player.

Do you have any ideas for how to adjust golf stats?

The trouble with simply measuring the length of each putt a player makes or misses is that we all know not all four-footers are created equal. There's a four-footer at Indian Wells during the Hope, and the four-footer Mickelson missed on #17 during his Sunday round at Shinnecock. To truly capture the skill level needed to make a given putt, you almost need a degree of difficulty rating, such as they have in diving or gymnastics.

Such a rating could be applied to all shots, if one were inclined to get that deeply into it; it's a whole hell of a lot harder to hit the 18th fairway at Oakland Hills than the first fairway at The Old Course, for instance, but each would count the same under the current system.

I think the reason golf stats are so much less usefull than baseball stats is that baseball is a team game played by individuals; it takes anywhere from one to four players to do something statistically measurable to get a run across the plate. That's why James' new stats, like Runs Created, are so helpful in determining a player's real worth. But since golf is about an individual's final score, his score itself tells us much more about his ability than the final score of a Mariners game tells us about, say, Ichiro's ability.

That said, baseball stat freaks would love to have a system like Shotlink, which provides a much more precise look at how players arrive at their final scores. You can't help but learn something usefull if you know how long each shot was, where it was hit from and where it ended up. Someday we might even have all the info about each shot, such as the lie, the wind, the degree of elevation, the break, etc.

At that point, golf's Bill James might well emerge.

I haven't read all the posts on this subject, but a more important stat is puts per green in regulation. If you miss a lot of greens, you stats will be better, because the pros tend to get close from off the green...

For nearly 20 years baseball old timers told Bill James that his stuff was bumf (whatever bumf is).

A couple of years ago the Oakland A's hired him as a consultant. Last year the Red Sox hired him full time. The rest is history. Nobody in baseball calls his stuff bumf any longer.

Bob

How about a Moe's scale, based on degree of accuracy as a function of distance. If you hole out a shot, it's Moe's 1.000. If you hit it to one yard from 100 yds, Moes is .990,
from 200 yds is .995, 60 ft from 60 yds is .333, etc. It shouldn't be too hard to figure out.

How about a Moe's scale, based on degree of accuracy as a function of distance. If you hole out a shot, it's Moe's 1.000. If you hit it to one yard from 100 yds, Moes is .990,
from 200 yds is .995, 60 ft from 60 yds is .333, etc. It shouldn't be too hard to figure out.

Except that often good players are not aiming at the hole.

I believe the quote is something more along the lines of "There's lies, damn lies, and then there's statistics." I'm sure Dan Kelly knows it or can google it in 2 seconds flat.

I've always been more partial to "It's not how, it's what." Of course, the corollary to that is that when it comes to enjoying the game, whether watching or playing, for me it's more about how than what.

Go figure (pun intended).

Golf is such a multi-faceted game with each hole and even each shot being so individual, mathematical models just can't be made.

Statistics can only set out to prove what they set out to prove.

GIR?    An isolated event?
Putts/GIR? many factors affect
Sand saves?  ditto
I'll stop now.

I have a couple of suggestions, one is a nitpick and the other a conceptual misunderstanding.

First, the nitpick. Statistics can neither prove nor disprove anything. A statistician never sets out to prove anything. Statistics is a way of summarizing and interpreting evidence to see if it supports a given thesis or not. Supporting is not the same as proving, not supporting is not the same as disproving.

Now the misunderstanding. When you pick a small set of useless pseudo-statistics like GIR, putts per GIR and sand save percentage of course you can't understand anything about the game no matter how you combine or formulate them. They are not informative.

It's a logical fallacy to say that since no combination of those statistics produce a useful model then no useful model of the game of golf can be made. Go collect me some real data and I'll build you a real model. You may ask what I consider real data...

Q: What is the fundamental event in the game of golf?
A: The stroke.

Q: What sort of statistics would be useful for understanding golf?
A: Outcomes of each stroke.

Q: What determines a successful outcome for a stroke?
A: How far the ball ends up from the hole (in the hole is best) and in what kind of lie.

This line of reasoning leads me to want data on where each stroke during a round ends up, distance and lie. Examples of such data records would be: 1) an approach shot ended up ten feet from the hole on the green, 2) a drive ended up 140 yards from the hole in the thick rough, 3) a putt ended up one foot from the hole or 4) a chip shot was holed out.

By stringing together these records you know where a stroke went from and where it went to. Maybe one stroke was a drive that started out on a tee and ended up 180 yards from the hole in the fairway and then the next shot started 180 yards from the hole in the fairway and ended up twelve feet from the hole and then the next shot started out twelve feet from the hole on the green and ended up holed out.

The final thing that would be nice to know is what club was used for each stroke. Given the distance and lie before the stroke, the distance and lie after the stroke and the club used for each stroke in each round for each player, you'd have a dataset that would be very informative about how the game was played.

The PGA Tour uses volunteers to collect this data at Tour events but I don't know anyone with access to it. You can subscribe to ShotLink but that just gives you predigested summaries in a flashy multimedia package. I don't think it is straightforward to get the data you'd need to actually do golf statistics. Playing around with GIR, putts, etc. is fun but it isn't going to tell you much that you don't already know.

BCrosby:

This thread is pretty old.  I still don't know where you come off saying they are "unhelpful".

Statistics are meaningless to tell you what will happen, but very helpful when trying to tell what has happened.  (Example:  Stroke average of 70.5 in college vs. 71.5 on the minitours vs. 73.5 on Tour... who is favored in a match.)  

Golf has a TON of statistics that are compiled.  Dave Pelz can stat you to death on how players score.  The Tour publishes some okay summaries, but the best stuff is unpublished.

You cite James.  He illustrates that statistics are meaningless without someone to interpret them.  I actually met with a very well placed Golfweek official about four years ago to develop what I called Statman - the person (in this case me) who delved into an issue of the (at the time) Sal Johnson stats to break it down to meaningful material for a reader.  Things kind of broke down when I told him I was suggesting THEY could do and he was hoping I could do it for them.  Not to mention that Seanor was skeptical, probably because he didn't know how to get it ready in time for their publication each week.  (It would take some time.)

I will share with you some of my work.

Statman – August 4, 2000

Interesting items, notable achievements, and other oddities from the world of championship golf


·   Players in the Deere Classic that received sponsors exemptions had mixed results.  Charles Howell finished 3rd.  Dan Halldorson, Charlie Rymer, and Chris Wilkins were three of the last six.
·   It’s always something.  Kazuhiko Hosokawa was among the leaders in driving and putting at the Kemper.  His 62 or 72 GIR was tops at the Deere.
·   Sunday Bloody Sunday.  Cam Beckman’s week in Quad Cities:  69-69-69-75.
·   Too Late.  76-65-MC for Barry Cheesman and 79-67-MC for Steve Scott.
·   Too Often.  John Daly’s 80, following an opening 68, was only good enough to beat two players in Friday’s second round.  In a similar fashion, Tom Anderson shot 69-83 in the Idaho Club Pro.
·   You want to have the LEAST putts.  Emilie Klein’s 46 of 54 GIR were the most.  So were her 99 putts.
·   Isn’t that all of them?  Melissa McNamara hit 42 fairways in 3 rounds.
·   Fortunately, there’s no room for pictures on the scorecard.  25/42 fairways (only one player hit fewer), an average of just 8 GIR per round (no one hit less), and even par!  58 putts!  An average of 1.074 per hole.  How many times did you chip in, Stefania Croce?
·   39/42 fairways (most), 48/54 GIR (only one player hit more), and 3rd fewest putts.  Any wonder Bruce Fleisher won wire-to-wire?
·   Age has its effects on Cal Peete.  His 216 yard driving average was 17 yards shorter than Gene Littler, this week’s next shortest driver.
·   Look what happens when you take a week off.  The fabled ironman finished an un-Bruce Summerhays like 57th a week after caddying for his daughter in the U.S. Open.
·   Never give up.  David Good and Jerry Bruner made the cut at Royal County Down by firing 80-68.
·   I thought he won.  Scan closely and you’ll see Christy O’Connor shot 80-78.  Easily half of the low father-son team in the Sr. British Open.
·   Winner Spike McRoy’s 63 GIR were 5 more than anyone else on the Buy.com Tour.
·   Steady as she goes.  In Europe, Bernhard Langer, Angel Cabrera, and Mathias Gronberg all finished T3 or better by shooting only 68s and 69s.
·   Leif Olson’s matches in the Colorado Match Play went just 12, 17, and 13 holes before losing on the 28th! hole in the Quarterfinals.
·   Couple more rounds and he might have won.  Jack Melnicoff posted 81-75-68 on the Collegiate Players Tour.
·   Runaway.  Abigail Spector won the Maine Am by 12 after leading by a Tiger-like 8 after 18 and 9 after 36.  She had the low round all 3 days.
·   Clay Homan, 3 up on the field after an opening 66, turned it around with 79-77 to finish 13 back on the Tight Lies Tour.
·   Crowded at the Top.  9 players within 2 shots of the winner on the Euro Challenge Tour.  32 players within 5.
·   Spotted.  Jim Benepe, winner of the Western after his college career at Northwestern, found on the Canadian Tour.  Henry Liaw, who fashioned a 58 in a junior event a few years ago in California, posted a nice finish on the AJGA Tour.

- Statman brought to you by John Conley