Reading Golf Data Again: When OWGR and Strokes Gained Tell Two Different Stories

In October 2026, the Official World Golf Ranking board announced its decision to deny ranking points for LIV Golf events. That decision did not come from an emotional vote. It came from three technical criteria that any ranking system must follow: an event must be played over at least 72 holes, must have a 36-hole cut, and must be open to all eligible golfers through a published pathway. LIV met two of the three and missed exactly the remaining one. A 54-hole, no-cut, team-based format. Technically, this was a clean decision. Numerically, it created a paradox that took me a long time to accept: the golfers playing best in one system receive no ranking credit, and the world ranking — long treated as the most objective measure in the sport — begins to measure something different from what fans see on screen.

I make my living reading data. I have no right to complain about a decision that is methodologically correct. But I have an obligation to point out that a measure that is methodologically correct can still be wrong in meaning. That is the subject of this article.

Context: two systems, two philosophies

To understand why the OWGR and LIV story matters so much, we must separate two concepts that golf fans often merge into one: ranking and skill.

The Official World Golf Ranking began in 2026. Essentially, it is a weighted points system over a rolling window of roughly 104 weeks, or two years. Each event carries a certain value depending on field strength, and the points a golfer earns depend on his finishing position. The system divides by a minimum divisor to limit the advantage of a golfer who plays few events but only big ones. This approach has clear logic: it rewards consistency and competing against strong fields, not a single explosive week.

Strokes Gained is a different philosophy entirely. The concept was popularized through the work of Professor Mark Broadie at Columbia University, especially through his 2026 book. The core idea is simple but revolutionary: instead of counting strokes, measure the value of each shot against the tour-wide average from the same position, same distance, same lie. A three-metre putt has a different expected value from a three-and-a-half-metre putt. An approach from 150 metres in the rough differs from one from 150 metres in the fairway. Strokes Gained splits into four categories: off the tee, approach, around the green, and putting. Add the four together and you get a picture of how many strokes a golfer gains or loses against the baseline.

These two systems answer two different questions. OWGR answers: how well has he competed, in what context, over the past two years. Strokes Gained answers: how well is he striking the ball right now, which skills carry him, which skills leak. One is outcome, one is process. And when the two diverge, that is where the work of a data reader becomes interesting.

In Vietnam, an extra layer of context makes this story relatable. Vietnamese golf is in a growth phase in courses and players, with domestic events run by the Vietnam Golf Association, alongside amateur and professional tournaments far smaller than the PGA Tour. When I sit down to analyse data for a domestic event, I always remind myself: do not apply US tour data standards directly to a field of a few dozen golfers with a much thinner competitive calendar. A metric that means something with a sample of thousands of shots can be meaningless with a sample of a few hundred. This is the first principle, and the most violated one.

Core: when the number is right but the story is wrong

The minimum divisor and the illusion of consistency

The OWGR system uses a minimum divisor, generally understood as 40 events over the two-year window, to prevent a golfer who plays only a few big events from holding a high position. If he plays fewer than 40 events, the average is still divided by 40. This mechanism has a good purpose: it encourages a full schedule. But it creates a side effect few notice.

Imagine two golfers. The first plays 40 events over two years, with a few explosive weeks and a few slumps, averaging out to some number. The second plays 20 events, all majors, and plays extremely consistently. Because the divisor is pushed to 40, the second golfer's score is diluted by 20 events he never played. Mathematically, the system punishes him for not showing up, not for playing poorly. This is a textbook example of a rule that is right by design producing a signal that is wrong in reality.

When LIV appeared and drew a group of golfers to compete there, those golfers fell straight into the divisor trap. They still played, still won, but the events they played earned no points. As a result their ranking slid, not because form declined, but because the input source was cut. This is the point I want to stress: a ranking is only as objective as the data source that feeds it. Cut the source and the ranking still runs, still produces numbers, still looks scientific. It is simply measuring something else.

Strokes Gained: four doors into one room

Switching to Strokes Gained, the picture becomes far subtler. I usually treat these four categories as four doors, each telling part of the truth about a golfer.

Strokes Gained: Off the Tee measures the value of the drive. But understand it correctly: it does not measure distance alone. It measures the combination of distance and accuracy, converted into expected strokes. A golfer who hits it far but often finds the rough can have a lower figure here than one who hits it shorter but always finds the fairway. This explains why the distance race on tour does not automatically translate into a scoring advantage.

Strokes Gained: Approach is the category I value most when judging long-term potential. It is the most stable skill over time, less affected by short-term luck than putting. A golfer with a consistently positive Approach figure across multiple seasons is usually a durable title contender, even before he wins anything.

Strokes Gained: Around the Green measures short-game handling. It is easily undervalued because it produces no flashy numbers.

Strokes Gained: Putting is the most dangerous door. It has the highest variance of the four. One hot putting week can push total Strokes Gained very high, and the next week it can collapse back to average. This is why I never conclude that a golfer has improved across the board from a single good putting week.

Course fit: a metric means nothing detached from the course

A common mistake in reading golf data is comparing figures across events while ignoring course characteristics. The same Strokes Gained: Approach means something entirely different on a windy coastal links than on a calm parkland course. The same putting figure means something different when major greens are far faster and more sloped than those of a regular annual event.

I call this the contextualisation problem. A number detached from the course, the weather, and the stage of the season is just a number, not a conclusion. When analysing for a team or a golfer, I always require data to be split by course type and condition, not pooled. Pooling produces pretty averages that are useless.

This is also why I am cautious about judging Vietnamese golf by international standards. Wind, humidity and grass types at Vietnamese courses differ substantially from the big tours. A metric built on PGA Tour data cannot be applied directly to a coastal central Vietnam course without adjustment.

Age curves and the small-sample trap

Another dimension golf data handles poorly is the age curve. A golfer's form does not change linearly with age. There are surges, plateaus, and technical rebuilds. When a young golfer wins a big event, public opinion immediately draws a straight upward line. Data rarely follows a straight line.

The problem is sample size. A 22-year-old golfer may have only a few dozen recorded rounds at the highest level. At that sample size, the variance of every metric is large. Drawing a long-term trend from a small sample is the most basic analytical error, and the most commonly committed one in sports commentary.

I once wrote about this in a football context, and it applies to golf in the same way. When a young player is pushed to compete at adult pace too early, an immature body bears a load it is not ready for. In golf, this shows up in rounds played and practice volume. A teenage golfer pushed into a packed schedule may shine early, but later injury data often tells a different story. This is an area where public data is thin, and I always say so rather than pretending to have an answer.

Tournament systems: field strength is not about prize money

The strength of a golf event is not measured by prize money. It is measured by field quality. An event with a huge purse but missing the top golfers earns fewer ranking points than a more modest event that gathers the elite. This is a point mainstream media often overlooks.

With LIV, the issue becomes complex. The field quality is high, the money is large, but the format does not match the OWGR technical criteria. The result is a parallel system: on one side money and attention, on the other ranking points and the pathway to majors. The two sides no longer align, and that is the source of most of the controversy in golf in recent years.

Contrarian angle: correlation is not causation

This is the part I want to spend the most time on, because it is where golf data is most misunderstood.

When we look at a golfer who wins a major, we easily see a string of impressive metrics and conclude that those metrics produced the victory. But a major lasts 72 holes, roughly 280 to 300 shots. The gap between winner and runner-up is often just a few strokes. At a margin that thin, a lucky putt on the 17th, a ball hitting the pin, or a wind shift at the right moment can all be decisive. Data describes what happened; it does not prove causation.

More specifically, when I see a golfer with a sharply elevated Strokes Gained: Putting for a week, I do not conclude he is an elite putter. I check sample size, check putt-distance distribution, check green quality. If the sample is small and the distribution skewed, I treat it as noise, not signal. Conversely, if a golfer has a consistently positive Approach figure across multiple seasons, I treat that as a far more durable signal, even if he has never won.

Another correlation-versus-causation example: distance. Fans see long hitters win and conclude distance is the key. But distance only has value when it converts into better approach positions. A golfer who hits it far into the rough can be worse off than one who hits it short into the fairway. The distance metric, detached from accuracy and approach metrics, is a number that leads the reader in the wrong direction.

And here I want to restate my professional principle: Numbers do not lie. But reputation whispers into the ear of the person who does not read the sheet. A golfer famous for a magical putter may have lost that ability long ago, but reputation follows him through the commentary. Data does not care about reputation.

Rules and equipment: when the game is edited from outside

There is another dimension a golf data reader must track: changes to rules and equipment.

The most recent example is the plan by the two governing bodies, the USGA and the R&A, to limit ball distance, intended for elite competition from 2028 and for recreational players from 2030. This is a direct intervention into a variable every golf data model uses. When ball distance changes, all distance metrics, course-fit analyses, and even the value of driving skill must be recalibrated.

Earlier, the ban on anchoring the putter to the body, effective from 2026, is another example of how a rule change can affect the careers of a specific group of golfers. Those who relied on anchoring had to rebuild their entire putting motion, and the performance data of this group during the transition is a lesson in how long a model needs to adapt to a structural shock.

As a data reader, I place these changes in the same frame: they are exogenous shocks that alter the distribution of metrics. Ignoring them is deceiving yourself.

Industry transmission: from the course to the data

The golf data story does not stop at the ranking. It spreads through the entire value chain of the industry.

Upstream is the course system and talent development. How many courses a country has, their quality, how many junior programmes exist — all determine the data supply of the future. In Vietnam, the growth of courses and golf tourism in recent years creates a potential data source, but the quality of data recording remains an open question. Without standard data, there is no standard analysis.

Midstream are the tours and event operators. How they design events, how they record shot-level data, determines the quality of all later analysis. This is why PGA Tour ShotLink is so valuable: it is not just a product, it is data infrastructure.

Downstream are media, sponsorship, and data products. When data infrastructure is good, media can tell more accurate stories. When it is poor, media is forced to rely on sentiment and reputation.

Takeaway: a signal for the next cycle

What I want to leave is not a conclusion, but a signal to track.

I started my blog from a lecture hall, believing data would speak for itself. Eleven years later, I teach it to speak in words. But I also learned that data can only say what its source allows. When a ranking system has its source cut, when an event format is unrecognised, when a rule change has not yet been reflected in a model, the numbers still run but the meaning is gone.

I do not predict. I read data and accept the consequences. For golf, I will track three signals in the next cycle. First, whether new event formats find a mechanism to integrate into the ranking system, or whether a parallel system persists long-term. Second, whether equipment changes are reflected in skill-evaluation models quickly enough. Third, whether Vietnamese golf can build a data-recording infrastructure standard enough for domestic analysis without depending on imported metrics.

One question remains open: if a ranking that is right by method measures wrong in meaning, then what we need to fix is the method, or our definition of measurement? Data has the answer, but only when we are willing to read it the right way.