Vietnamese Golf: When an Empty Cell Gets Read as Zero

11 p.m. in Binh Duong. I reopen the spreadsheet from an amateur event played a few days earlier: 132 rounds, seven metric columns, and 47 empty cells scattered across the fairway and greens-in-regulation columns. Nobody forgot to write. The recorder simply was not standing in the right place when the ball left the club.

What stopped me was not the 47 blanks. It was the reflex. My head already had an answer for every cell: this player hits it straight, that one leaks left. I could fill them all in four minutes. And if I did, the dataset would look immaculate, and every conclusion drawn from it would be neatly wrong.

Data does not lie. But reputation whispers into the ear of anyone who will not read the sheet.

Golf is the most recorded sport and the least analysed. Every shot costs a line, yet most tournaments on this planet keep only the total. The PGA Tour has run ShotLink since the early 2000s, logging every shot with distance, lie and club. In Vietnam, the VGA Tour system and nearly every amateur field keep only a scorecard: strokes, points, placing.

On Vietnamese amateur courses, most data is recorded on an honour basis: players count their own strokes and their group's. That convention preserves the courtesy of the game, but it produces a dataset that cannot be compared across groups, because each group applies its own level of strictness.

The gap between those two worlds is not a matter of hardware. It is a matter of method, and of a question few organisers ask: what do we need to know in order to understand a round?

One fact deserves to sit beside that question. On 6 December 2026, the USGA and the R&A announced a limit on golf ball flight distance. Elite competition adopts it from January 2028, recreational players from January 2030. A rule change of global scale drags a long chain of adjustments in clubs, tactics and how distance is calculated on every hole. Anyone holding shot-level data will measure the effect within weeks. Anyone holding only scorecards will need several seasons to notice that something has changed.

At the professional layer, I always separate two kinds of data. One is outcome: strokes, points, placing. The other is process: where the ball went, from what lie, by what means. Xander Schauffele won the PGA Championship in May 2026 and The Open that July. Looking at the final score column, those two weeks look alike. The routes to those two titles were entirely different in approach quality and putting quality.

From a bare scorecard you can compute six things: total strokes, strokes by round, the spread across par 3, par 4 and par 5 holes, birdie count, bogey count, and round-to-round variance. Seven things you cannot compute: tee-shot direction, approach quality, putts taken on the green, putting quality, scrambling after a missed green, the lie the ball finished in, and the weather at the moment of the shot. Those seven variables are where a round is decided.

The problem is not the missing seven. The problem is how people handle the missing part. In every model I have built, an empty cell faces two fates: flagged as insufficient information, or filled with a zero. The second fate is far more dangerous, because a zero is a claim: this player never missed a fairway. In reality, he simply was not recorded.

The silence of data and the absence of risk are two different things. But in a spreadsheet with no "unknown" column, they look identical.

This is a process illustration, not data from any specific event. Assume a sample of 96 rounds has a fairway column missing 18 per cent of its entries. One group fills every gap with zero and computes an average fairway rate of 61 per cent. Another group marks nulls and calculates only on the 82 per cent of rounds with real data, arriving at 74 per cent. Same source file, opposite conclusions. If a national squad selects its players on the 61 per cent figure, it is cutting people with a data-entry error.

For Vietnamese golf, the work required is far smaller than it feels. Seven columns. Two volunteers per group: one at the tee recording ball direction and club type, one at the green recording the ball's finishing position and putt count. No radar, no sensors, no equipment budget. It takes a printed sheet and one inflexible rule: any cell not observed is marked null, and no guessing is permitted.

I built this process for a team of assistants in Binh Duong. In week one, 30 per cent of cells were null. By week four, 9 per cent. Nobody got better over those four weeks. The recorders simply learned where to stand.

The counterintuitive angle sits elsewhere. People often say golf lacks data because golf is hard to measure. The reverse is true: golf has an abundance of data and a shortage of people recording the right thing. Once someone is recording, instinct reaches for the easiest columns to fill — total strokes, birdie count — and skips the columns that decide outcomes. A thick dataset is not the same as a useful one.

The bigger risk lies in ranking systems. Models that evaluate golfers, even good ones, reward players who compete often and compete early. A Vietnamese amateur who appears in no ranking is not a player without risk. He is a player who has not been measured. Those two states differ in kind, yet in a selection report they are usually written as the same line: no data.

I hate uncertainty. But 2026 taught me that one unforeseen variable can outweigh every algorithm. When stadiums closed, the home win rate fell from 49 per cent to 38 per cent. A single variable — the crowd — was enough to rewrite the entire formula. Vietnamese golf sits in a comparable state around one unmeasured variable: tournament density and travel distance for the leading amateur group.

Next round, try something small. Open the scorecard, find an empty cell, and read it as a question rather than an answer. The answer for Vietnamese golf does not live in measuring hardware. It lives in two volunteers standing in the right places, filling the right columns, and being willing to leave a cell blank when they did not see it.

I do not predict. I read the data and accept the consequences.