The Empty Cell in a Golf Data Sheet: The Discipline of Not Inventing Numbers

The third-round data sheet opened on screen at 6:40 in the morning. The Strokes Gained: Approach column was empty. The SG: Putting column was empty. Only three fields survived: total score, putts, greens in regulation. The man across the table pushed his coffee over and said the sentence I have heard often enough in eleven years of this work: "Just estimate it, who is going to check?"

I closed the laptop.

There is nothing heroic in that. No audience, no elegant chart, nobody clapping. There is only one rule: a cell without data stays blank. That sounds simple until you realise a twenty-page report can collapse into seventeen blank pages, and someone is waiting for it before lunch.

The empty cell in golf analytics does not mean laziness. It is the trace of a measurement system that was never built out properly. How a person handles an empty cell says more about whether they are a data professional than any job title.

I write the report, I close the file, and then the market opens again on its own.

Golf came to data twenty years later than football

Mark Broadie, a professor at Columbia Business School, systematised the Strokes Gained method and turned it into the common language of professional golf. The PGA Tour added the metric to its official statistics in 2026. His book "Every Shot Counts" arrived in 2026. Since then, every serious argument about golf has to pass through four columns: off the tee, approach, around the green, and putting.

But Strokes Gained does not generate itself. It lives on ShotLink, the PGA Tour's data-collection system, with laser measuring devices on the holes, a dedicated operating crew and hundreds of volunteers per round. A machine like that costs money and people, and only exists where there is money and people.

The PGA Tour covers almost its entire schedule. The LPGA arrived many years later and still unevenly. The DP World Tour has full data at its biggest events and at co-sanctioned tournaments with the PGA Tour; the rest is thin. The Asian Tour, Japan Tour, Korean Tour and nearly every national tour sit outside the coverage area.

Vietnam has never had ShotLink. The last time a tournament on a major international tour stopped here was the 2026 Ho Tram Open at The Bluffs Ho Tram Strip, on the Asian Tour, won by Sergio Garcia. Since then, most golf data in this country has been manual data: paper scorecards, stroke counts, and human eyes.

So what does golf analytics in Vietnam live on? On what can be measured, and on honesty about what cannot. Data is never in a hurry; it only waits for someone who knows how to read it.

The technical boundary: what is measurable and what must stay blank

At a round with ShotLink, I have four metrics to decompose every shot. At a round without it, I have a much shorter list: driving distance measured from video, fairways hit, greens in regulation, putts, scrambling, sand saves.

That list carries a trap. Putts per round is a bad metric, and it is bad in the most dangerous way: it returns a number that looks perfectly reasonable. A player who hits more greens will take more putts. A player who misses greens and chips close will take fewer. Using putts per round to compare two golfers is using a bent ruler to measure two differently shaped objects.

In my tracking files, when ShotLink is missing, I substitute a narrow but clean set: greens in regulation, scrambling rate, average score on the last three holes, and the rate of bogey immediately after a bogey. Those four fields can be captured from a scorecard, cross-checked, and need no inference at all. Every other column stays white. Not for appearance's sake. It stays white because that is the truth.

Player and form: a ranking metric with a structural hole

The OWGR is a weighted average of results over the past two years, divided by a minimum number of events. That structure makes it sensitive to three things: events played, strength of the fields played in, and the timing of injuries. A player who misses six months with a wrist injury drops not because he got worse, but because the denominator shrank.

When I assess a player, I read the major record first: top-10 finishes, cut-made rate, and conversion rate from contention to victory. Those three are far more stable than form over the last eight rounds. They also reveal something a ranking cannot: whether that player can survive the pressure of a Sunday.

The age curve in golf has two peaks out of phase. Tee-to-green skill peaks late, usually after 30. Putting skill peaks earlier and declines more slowly than people assume. Overall performance for most elite players peaks somewhere between 30 and 35. That is why a 26-year-old who wins two events in a row has not necessarily entered his best years, and why a 34-year-old the media calls finished may not be finished.

Tournament system: points are generated by field quality

A tournament does not award fixed OWGR points. Points depend on field strength, meaning how many ranked players show up and how highly they are ranked. An event with twenty players inside the world's top 50 awards far more points than one with three. This structure pushes smaller tours into a spiral: few points, so hard to attract strong fields; weak fields, so even fewer points.

At the top, the season splits into four majors, the PGA Tour's FedExCup, the DP World Tour's Race to Dubai and Ryder Cup qualifying. Lower down, the only thing that truly matters is the tour card. Losing the card means losing access to the course, and losing access to the course means vanishing from every data table.

This is where analytics creates its own blind spot. A player ranked 120th in the world appears in my data roughly twenty times less often than a player ranked 12th, not because he is twenty times worse, but because he is broadcast twenty times less. An empty stadium does not lack noise; it lacks a dimension of data.

Governance: two data tables speaking two languages

On 10 October 2026, the OWGR body rejected LIV Golf's application for ranking points, citing the 54-hole format, the absence of a cut and the team structure. In March 2026, LIV Golf withdrew the application.

For a data person, the consequence is not about who was right. The consequence is that from that moment, there exists a group of players with complete competitive data, measurable round by round, but with no ranking yardstick compatible with the rest of the world. Jon Rahm, Phil Mickelson and Brooks Koepka play, win, and those results never flow into the same data pipeline as Scottie Scheffler or Rory McIlroy.

When two tables speak different languages, every direct comparison is a false comparison. Analysts have two choices: build a separate frame of reference for each table, or leave the comparison cell blank. The second is more honest and less often chosen.

Rules and equipment: a break point in the time series

On 6 December 2026, the USGA and the R&A announced changes to the test conditions for ball distance, applying to elite competitions from January 2028 and to recreational play from January 2030. Earlier, the anchoring ban, Rule 14-1b, took effect on 1 January 2026.

For data work, an equipment rule change is a break point. Average driving distance on major tours has risen almost monotonically for decades. Any long-horizon forecasting model built on that series has to add a dummy variable for 2028, otherwise it will extrapolate a trend that the rules cut in half.

At grassroots level, the 2030 milestone creates another problem: amateur and professional data will be measured against two different ball standards for several years. Anyone comparing metrics across those two groups during the transition is comparing quantities with different units.

Risk surface: six categories and one unnamed risk

In a player assessment file, I track six risk categories: competitive, psychological, injury, career and commercial, governance, and systemic.

One concrete example: a player ranked 90th in the world, aged 34, with a history of wrist injury, in the final year of his card. Injury risk in that age bracket is higher than at 25. Psychological risk rises when every round counts as a qualifier. Commercial risk rises when sponsors start asking about results. Those three compound, and they compound far faster than the decline of any single skill.

The biggest risk in my job is not on that list. It is pipeline risk: bad input data, output that still looks complete. A report with full headings, full colour, full charts, and an entire body built from estimates. That kind of report is more dangerous than a blank page, because it does not indict itself.

Public narrative: heat cycles and expectation gaps

A player wins two events in a row. Within forty-eight hours, the media has built a dynasty. In my data, that is eight rounds. Eight rounds are not enough to separate a technical improvement from a lucky putting streak.

The heat cycle of a golf story is usually shorter than the data cycle. A story lives on emotion and dies when new results arrive. Data lives on samples and only dies when the sample is refuted. The gap between those two rhythms creates what I call the expectation gap: the market prices a player on eight rounds, while a two-year file says something else.

In Vietnam, the strongest current story is about the next generation. Nguyen Anh Minh and Truong Chi Quan are the two names that appear most often in domestic coverage. To confirm a genuine step forward, I need three fields: international rounds played per year, the strength of those fields, and cut-made rate. Missing any one of the three, the conclusion stays blank.

Industry transmission: the downstream data layer is still empty

Golf runs on three layers. Upstream is courses, equipment and coaching. Midstream is tours and event organisers. Downstream is broadcasting, sponsorship, data and derivative data products.

Vietnam has a fast-growing upstream: more courses, more players, more academies. The midstream is thin: a small domestic professional circuit with few rounds. The downstream is nearly empty: no professional golf data provider, no standardised electronic scoring system across events, no data market.

That produces an economic paradox: data quality does not get paid for, so data quality does not improve. The loop only breaks when someone pays for accurate recording instead of fast recording.

The counter-intuitive angle: a blank cell is more trustworthy than a filled one

Correlation is not causation, and in golf that trap is everywhere.

Data people repeat three mistakes. The first is treating a blank cell as a signal. Having no data on a player and that player having no ability are two different sentences written in two different languages. The second is trusting a proxy without checking how bent it is. Putts per round is the most heavily distorted proxy in this sport, and it is still used daily. The third is defending an old prediction with a new argument.

I hold myself to one rule: a hidden variable only enters a conclusion when it appears in at least three independent seasons. Fewer than three, and it stays blank. That rule has cost me many good articles. It also means I have never had to retract one.

There is a symmetrical blind spot the data crowd rarely admits. Someone who says "I have twenty years in this game" is actually holding a large uncoded dataset. Those twenty years are data; they simply are not in a spreadsheet. My job is not to dismiss them but to ask clearly: how many cases do those twenty years contain, what were the outcomes, and how often were the conclusions wrong. Being pushed out of the room is the fastest way to see the whole board.

Signals for the next cycle

Three milestones to track. First, how far ShotLink expands beyond the PGA Tour, because every event covered is a data column saved from being blank. Second, the January 2028 ball rollback milestone, the first break point in the driving-distance series in nearly half a century. Third, in Vietnam: a domestic event with standardised electronic scoring in every round, and a Vietnamese player inside the world's top 200.

I do not need recognition in a newsroom; the numbers know their own way into a story. But when a golf tournament in Vietnam pays for accurate recording instead of fast recording, the question will flip: how much is an empty cell in Vietnamese golf data actually worth?