Golf World UK - August 2017

MY RANKINGS EXPLAINED

Howascore-basedskillestimationrankingworks.

14 Golf WorldAugust 2017

“Last month, PGA Tour professional

KellyKrafttookissuewiththe

Official World Golf Rankings. ‘It is

amazing to me how fast some of

the Asian Tour/Euro Tour guys

rise in the world rankings,’ he

tweeted. His view caused a storm,

but in actual fact it was consistent

with a comprehensive academic

study completed by Professor

Richard Rendleman and myself.

Our aim was to show whether

or not there was any bias in the

rankings, and to do this we

compared the OWGR system

against two unbiased methods

for estimating a golfer’s skill and

performance. The first of these

methods was a score-based skill

estimation method, which used

scoring data to estimate a golfer’s

skill (see box, below). The second

was the Sagarin method, which

uses a win-lose-tie system and

scoring differentials to rank

golfers playing in the same events.

Neither method uses any tour

information, so neither can

favour one tour over another.

In contrast, in the OWGR

system, events with weaker

fields often receive as many

ranking points as those with

stronger fields because each of

the world’s tours has a minimum

points level that guarantees a

certain number of world ranking

points are available at each event.

Nine years of number crunching

After analysing data from nine

yearsof play on the PGA Tour,

European Tour, Asian Tour,

Australasia Tour, Sunshine Tour,

Japan Tour and Web.com Tour

and seven years on the European

Challenge Tour, we compared

the rankings produced by our two

unbiased methods with the

rankings from the OWGR and

found that there was a significant

bias against players who played

in more PGA Tour events.

The bias did not really affect

the top 10 or 20 players in the

world, as they tended to play

against each other a lot, but it

was very evident from 40-120,

which is a vital grouping as entry

intothemajorsandWorldGolf

Championship events are

determined, in part, by whether

a player is ranked in the top 50,

top 60 or top 100.

A good recent example of this

bias involves Martin Kaymer and

Jeunghun Wang. In April of this

year, both players had very

similar OWGR positions –

Kaymer ranked 44th, Wang 47th

• but very different records in
  events they both played in. To
  summarise, there have been 18
  events over the last two years
  where their paths have crossed.
  Kaymer has finished ahead of
  Wang in 13 of these tournaments,
  including every major and WGC
  event. Wang has finished ahead
  of Kaymer three times. And they
  have matched each other’s finish
  on two occasions. To me, this
  suggests that Kaymer has had
  far better results, so does it
  make sense that just three places
  separate them in the OWGR?
  Put it this way, I would pick
  Kaymer to do better in any
  event where they are both
  playing. You have to suspect
  that Wang’s OGWR is too
  high, particularly when you
  note he’s above the likes of
  Zach Johnson, Chris Wood,
  Charley Hoffman and Andy
  Sullivan, all of whom rank higher
  in my strokes gained category.

Now I’m not telling you this

because I have an alternative

formula to promote. I’m simply

telling you this, because Kelly

Kraft took a lot of criticism for

his Tweet when it was actually

consistent with our study. If three

players of identical skill levels

divide up so one plays the Asian

Tour, one plays the European

Tour and one plays the PGA

Tour, the two who play the Asian

and European Tours have a

greater chance of a faster rise up

the Official World Golf Rankings.”

Q Mark Broadie is a professor at

Columbia Business School, the

man behind the PGA Tour’s strokes

gained statistics and author of

the best-selling book Every Shot

Counts (everyshotcounts.com).

PINION

adinggolfstatisticianProfessorMarkBroadieontheTweetthatsparkedanonlinefurore.

“Kelly Kraft was right: the Official

World Golf Rankings are biased”

“When players are

competing on

different tours, we

don’tknowiflower

scores are due to

superior skill, easier

courses or more

benign

conditions. But if

we use connections

between the tours

(e.g. players who

play in events on

multiple tours), we

can estimate player

skill while separating

out differing course

and condition

difficulty effects.

A simple example

involves Tom, Dick

and Harry. Tom

plays the Red

course and always

shoots 71. Dick

plays the Blue and

always shoots 69.

Harry plays both

courses and always

shoots 72 on Red

and68onBlue.

Harry’s scores show

the Red is four shots

harder than the

Blue. Which means

Tom would shoot

67 on the Blue, so

isabettergolfer

than Dick – in spite

of Tom always

shooting higher

scores than Dick.”

It’s unfair to call the OWGR biased.

They don’t deliberately favour certain

players, and when you look at world rankings

in other sports, golf’s is remarkably accurate at

ordering the best players based on recent

results. Ultimately, the PGA Tour has more

money and ranking points available, so it’s

tougher but more lucrative. But if Kraft

wants a quicker rise up the rankings,

he’s free to switch tours and

play in Europe.

GW SAYS