Vietnamese Golf and the Empty Data Table: What Replaces Strokes Gained?

Vietnamese Golf and the Empty Data Table: What Replaces Strokes Gained?

A query that returned zero

On January 14, 2026, I sat in front of a screen in Nagoya and ran the opening query of the annual season: every shot-level record for Vietnamese golfers who competed on the VGA Tour in the 2026 season. The spreadsheet was ready with sixty columns, from first-putt distance to greens-in-regulation from rough. The system returned zero rows.

I assumed I had mistyped the syntax. I retried using the tournament name, the event code, the abbreviation. Still nothing. I moved to launch monitor data, something nearly every golf academy in Japan archives session by session. Empty. I tried training GPS logs, then hole-by-hole scorecards, then fairway and green statistics. All empty.

What exists is three columns: total score, finishing position, prize money. Three columns for a full season. For comparison, a four-day JGTO event in Japan generates tens of thousands of coordinate data points in a single tournament week.

I sat with that screen for a long while. Data is never wrong; I simply asked the wrong question. My original question — which part of the game do Vietnamese golfers lose strokes in — sits far beyond the information I actually hold.

Courses are growing faster than data

Vietnam currently has roughly 80 to 100 golf courses in operation, according to industry figures widely circulated in recent years. The number of regular players is rising, the VGA Tour calendar is filling out, and a younger generation has begun producing names tracked on the World Amateur Golf Ranking. Nguyen Anh Minh is the clearest example among amateurs, alongside figures such as Truong Chi Quan on the professional side.

Alongside course infrastructure, another layer has to grow: measurement infrastructure. That is the slowest part.

In Japan, where I work, golf data is built in tiers. The bottom tier is total score. The middle tier is hole-by-hole score, fairways hit, greens in regulation, putts per round. The top tier is shot-level data recording ball coordinates before and after every strike. That top tier is what allows Strokes Gained to exist, and it is also what determines the distance between two golfing nations.

Strokes Gained was not invented by a technology vendor. Mark Broadie, a professor at Columbia Business School, published the method in 2026; the PGA Tour formally adopted it from the 2026 season. Before that, analysts judged golfers by fairways hit, greens in regulation and putts per round. Broadie showed those metrics are systematically biased: they measure conditions, not skill. A golfer who reaches a 480-yard par-4 green from the rough has done something far harder than one who reaches a 350-yard green from the fairway, yet the old statistics record both identically.

That is why Strokes Gained became the standard. It is also why its absence cannot be covered by intuition.

Hideki Matsuyama won the Masters on April 11, 2026 at Augusta National, becoming the first Japanese man to win a major championship. The story told most often involves a putt, an approach, a final-hole nerve. The story told least often is that two decades earlier, Japanese golf had already completed the data tier that anyone can now look up.

I have reasons to be sensitive about this. In 2026, at 24, I did data analysis for Nagoya Grampus in their first season in J.League 2 after relegation. I built a manual xG model from video and omitted the home-venue variable. The model got six of the last ten matchdays wrong. I had to re-watch the footage, check every phase of play, and conclude that raw data is not enough — without tactical context, numbers are just noise.

In 2026 I collected PPDA figures for the Japan versus Belgium round-of-16 match at the World Cup on July 2, 2026 in Rostov-on-Don. The metric showed Japan pressing well. I ignored Belgium's running distance after the 70th minute. Belgium won 3-2, with Nacer Chadli's winner in the 90th minute plus four. I publicly criticised my own work online, and since then every pressing conclusion I publish carries a running-intensity chart broken into fifteen-minute blocks.

In 2026, stadiums stood empty because of the pandemic, Nagoya Grampus went two months without a match, and I had to rebuild the form model from youth-team GPS training data plus precedents from historically disrupted seasons. The coaching staff objected. I persisted using J.League 2026 data from after the earthquake disaster. The team lost only two of ten matches after the restart.

Those three episodes taught me one thing: when data hides its face, error margins become the guide.

What Strokes Gained requires, and what Vietnam has

For a Strokes Gained figure to mean anything, four things are needed. Ball coordinates before each shot. The lie type the ball sits in. Ball coordinates after the shot. And enough sample — typically forty competitive rounds or more in a season, so the confidence interval is narrower than the effect being measured.

Vietnam has the last item on that list and none of the first three. In other words, if I published a Strokes Gained number for any Vietnamese golfer today, its error margin would be wider than the very gap I am trying to explain.

That is when the method of exclusion becomes the only usable tool. Exclusion is the key.

The principle is simple: I do not ask what can be measured, I ask what can be removed without distorting the conclusion. From the three existing columns — total score, finishing position, prize money — I can extract four reasonably solid things.

Scoring average per round, counted on actual rounds rather than events. Cut conversion rate, which measures the ability to hold a minimum standard across consecutive days. The scoring gap between the first and final rounds at four-round events, a crude indicator of mental and physical endurance. And consistency across consecutive events, measured as the standard deviation of scoring average per tournament.

Those four answer a narrow question: who is improving. They cannot answer the wider question: improving in which part of the game. A golfer may lower their scoring average because of better putting, better approach play, or simply because they faced weaker fields. Those three causes lead to three entirely different training plans, and a composite metric cannot distinguish them.

Without hole-by-hole scores, I do not even know whether a golfer loses strokes in the first half or the second half of a round. In golf, the pattern of scoring loss by block of holes is the cheapest observable thing there is, provided someone keeps records. A golfer who averages 0.4 strokes lost over the first three holes but plays at tour level across the remaining fifteen is a golfer with a warm-up problem. A golfer who holds position until the 15th and then loses 1.2 strokes over the final three is a golfer with a decision-making problem under pressure. Those two diagnoses require two different training programmes, and neither needs expensive technology — only an honest scorer and one standard form.

This is where I test a lens I still use in football. In football, PPDA measures how quickly a team wins the ball back after losing it. In golf, the equivalent is bogey-recovery rate: after a bogey hole, how many holes does that golfer take to return to par or better.

I must state the conditions clearly, because this kind of comparison is easily abused as decoration. I only accept this lens after testing it on the JGTO dataset I collected and cross-checked from 2026 to 2026, and I keep only results where the deviation is large enough to be statistically meaningful. For Vietnam, I have no basis to do the same — and I will not publish a metric simply because it sounds plausible. Gegenpressing does not break data; it breaks my assumptions. But it cannot replace data I do not have.

One more variable is routinely ignored in international comparisons: field strength coefficient. A 72 scoring average in a field of thirty amateurs is not worth the same as a 72 in a field with twenty professionals and five major qualifying places. The World Amateur Golf Ranking handles this through field-strength weighting, but that weighting only works when organisers publish complete and consistent entry lists. When entry lists change late, the coefficient loses value, and every cross-event comparison collapses into a comparison between two courses with different conditions.

Below is the table I use to decide which items may enter the model.

| Item | Minimum input data | Current status in Vietnam | Current status in Japan (JGTO reference) | |---|---|---|---| | Strokes Gained: Approach | Ball coordinates before and after shot, lie type | Absent | Present | | Strokes Gained: Putting | First-putt distance, putt count | Absent | Present | | Bogey recovery | Hole-by-hole scores | Incomplete | Present | | Scoring distribution by hole block | Hole-by-hole scores | Incomplete | Present | | Season scoring average | Total score, actual round count | Present | Present | | Field strength coefficient | Entry list, world ranking points | Absent | Present |

This table explains why I am not writing a ranking of Vietnamese golfers by advanced metrics. I do not have the raw material. And a ranking built from three columns will not be wrong in its arithmetic — it will be wrong in its assumptions, which is the hardest kind of error to detect.

There is one precedent I have used before and still consider right for this situation. In 2026, with no matches to measure, I did not wait for match data. I used youth-team GPS training data as a substitute and validated it against the J.League 2026 season — a campaign disrupted by the earthquake disaster, when clubs also had to play without a normal data stream. The result was imperfect, but better than having no model at all.

Applying the same logic to Vietnamese golf, the first data layer worth building is not a tracking system for professional events. Three cheaper and more valuable tasks should be done within twelve months. Standardise hole-by-hole scoring across all VGA Tour events and publish it in machine-readable form — this requires one rule and one form, not equipment. Establish a minimum protocol for launch monitor sessions, with the same file format and the same metric list, so one academy's data can be compared with another's. And open a public junior golf database, updated monthly, recording the actual number of competitive rounds and not just scores.

That last point matters more than it appears. For players under 18, rounds played per year is a biomedical variable, not a line of statistics. A 14-year-old playing twenty tournaments in twelve months, on top of daily practice, is accumulating load on a body whose skeleton and tendons are not yet mature. With no database recording that, nobody detects the pattern until injury appears and becomes the latest possible data point.

Vietnam's golf problem is not a missing Strokes Gained metric. It is the absence of the bottom data tier — hole-by-hole scoring — and every advanced metric built on an empty foundation is a form of decoration with numbers attached.

I want to add a note on the cost of this gap, because a data gap is not neutral. It distributes advantage to those who already hold information.

When a young Vietnamese golfer seeks sponsorship, their file is a sheet of paper with a ranking and prize money. When a Japanese golfer of the same age seeks sponsorship, their file also contains scoring distribution by hole block, competitive rounds per year and quarterly improvement. Sponsors are not necessarily smarter; they simply have more data points to feel confident with. That difference is not about talent, it is about record-keeping infrastructure — and it compounds year by year in ways that are hard to reverse.

Every number is a confession not yet written into prose. The worrying part is not that those confessions are hidden, but that they were never recorded at all.

The contrarian angle: copying Strokes Gained is the wrong move

There is a natural reflex in the industry: see that Japan has Strokes Gained, see that Korea has Strokes Gained, and go buy a tracking system. I do not think that is the right move now, and the reason is mathematics, not budget.

Strokes Gained was designed for major tour conditions, where a golfer plays twenty-five to thirty competitive rounds a year on courses with standardised difficulty. The forty-round threshold I mentioned exists so the confidence interval is narrower than the effect being measured. A Vietnamese golfer playing eight to twelve competitive rounds a year, mostly on courses that differ in green speed and grass length, produces a sample whose standard error exceeds the gap between first and tenth place.

In other words, an expensive system will produce numbers that look precise but are not strong enough to drive decisions. And a number that looks precise is more dangerous than an acknowledged gap, because it makes people stop asking questions.

There is another correlation I must state plainly to avoid being misread. The number of golf courses and the number of elite golfers do not move in a straight line together. Japan has many times Vietnam's course count yet still needed several decades to produce its first male major champion. If course infrastructure were the direct cause, that interval would have been far shorter. The variable I consider more important lies in junior competition structure and record-keeping culture — but I must admit this is a hypothesis under verification, not a conclusion.

And one self-criticism with corrective data attached. In 2026 I published a junior golfer ranking based on total scores from four events, and that ranking was methodologically wrong. When I recalculated the standard error for a four-event sample, the confidence interval was wider than the gap between first place and twelfth place. I withdrew the ranking and replaced it with a note on data limitations.

Signals to watch over the next six months

What I am waiting for is not the result of any tournament. I will be watching whether anyone publishes hole-by-hole scores from a VGA Tour event in machine-readable form, together with a complete entry list. If that happens, my first model can run, even with a single variable.

What does NOT happen often tells the truth better than what does. The fact that a hole-by-hole scorecard does not yet exist is not bad news about Vietnamese golf. It is only bad news about how we currently keep records.

And if that data table is still empty three years from now, my question will have to change again — this time not because data contradicted it, but because there was no data to contradict it.