The Empty Strokes Gained Table and the Discipline of Reading Golf Data

The sheet of paper sat on the desk in the temporary office in Cam Ranh. Twelve pages. Every page had tables, ruled lines, proper column headers. And every data cell carried the same single entry: N/A.

It was a golf analysis. It had a Strokes Gained: Off the Tee section. It had Strokes Gained: Approach. It had course fit. It had OWGR. It even had a six-by-six risk matrix. Not one cell was formally empty. Not one cell contained information either.

The man who handed it to me asked: what do you make of this player's form?

I said I could not say anything.

Not out of modesty. Not for lack of opinion. I simply had no data point on him. No average driving distance. No greens-in-regulation rate. No scoring average over the last three rounds. No event name. No date.

The biggest temptation in this trade is not to be wrong. It is to say enough. Numbers do not lie. But reputation whispers into the ear of anyone who will not read the table. And that whisper is always ready to fill every N/A with a story that sounds perfectly reasonable.

I work as a data consultant. My daily job is to turn a match, a round, a swing into numbers that can be verified, and then turn those numbers into a decision. In golf, my most-used frame of reference is Strokes Gained.

Strokes Gained was popularised by Mark Broadie, a Columbia Business School professor, in his 2026 book Every Shot Counts. The core idea is simple. Every position on a course has an average number of strokes required to finish the hole, calculated from professional data. If you move the ball from position A to position B, the drop in expected strokes is the value of that shot. Subtract the tour average and you have Strokes Gained.

The metric splits into four buckets: off the tee, approach, around the green, and putting. Together they describe a player's scoring picture.

In the United States, where I grew up watching the PGA Tour, this data is captured by ShotLink. Every shot by every player on every hole is laser-measured, with thousands of volunteers positioned around the course. ShotLink is not an app. It is infrastructure.

Vietnam has no ShotLink. That is a foundational fact, and I state it first because ignoring it sends every local golf analysis drifting toward guesswork.

Golf data in Vietnam exists on three tiers. The lowest and most common is the scorecard: total strokes, hole-by-hole scores, birdies and bogeys. One level up is tournament data: finishing position, margin behind the leader, number of rounds. The top tier is shot-level data, and at a system level that tier is essentially empty.

Which means most golf analysis Vietnamese readers encounter, including mine, is built on the bottom two tiers. We know how many strokes a player took. We do not know where he earned them.

That is why I opened with an empty table instead of a name.

The four Strokes Gained buckets and the problem of translating them into Vietnamese conditions

Off the tee measures the value of the drive. It merges two variables that fight each other: distance and accuracy. A 300-metre drive into heavy rough can be worth less than a 265-metre drive down the middle. On paper both are one stroke. In data they are entirely different.

At The Bluffs Ho Tram Strip, the Greg Norman design that opened in 2026, the wind variable is large enough to invert the value of a drive within two hours. Same player, same club, same line. In the morning, with the sea breeze, a 280-metre drive is optimal. By afternoon, with the wind reversed, that same shot is a risk. A standard Strokes Gained table does not know this automatically. The person reading it must.

Approach is the heavyweight. According to Broadie's research, most of the scoring difference between tour professionals comes from the long game, with approach at its centre, while putting accounts for a far smaller share than television audiences assume. This is where Vietnamese golf media and American golf media commit the same error: devoting most of the coverage to the putt that decided it, when the deciding had already happened three shots earlier.

Around the green measures par-saving when the green has been missed. It depends heavily on grass type. Paspalum, common at many Vietnamese courses because it tolerates hot, humid conditions, responds differently from the bermuda and bentgrass of American courses. The same chip from the fringe rolls differently. A PGA Tour baseline was built on American turf. Applying it to a central Vietnam course in July creates systematic error, not random error.

Putting is the noisiest bucket. It is also the one fans trust most. The paradox lives there. A player holing a 12-metre putt three rounds in a row creates the feeling of an elite putter. But the sample size of one putt is one putt. Three rounds is roughly 90 putts. That is a small sample, and in a small sample, variance beats meaning.

The rule I keep: never let a single metric speak alone.

Every number must sit beside three things: the opponent, the course conditions, and the stage of the season. A 72 percent greens-in-regulation rate at a local amateur event is not measured in the same units as 72 percent on the Asian Tour. Same figure, different meaning. The writer is responsible for stating which frame of reference applies.

The mathematics of the world ranking and the cost of isolation

The Official World Golf Ranking runs on a rolling two-year window with a minimum divisor of 40 events. Points are averaged across events played, but if you have played fewer than 40 over two years, the divisor is still 40.

That rule has a direct consequence for Vietnamese golf. A player competing mainly domestically will carry a low field rating, because the quality and quantity of opponents cannot generate high points. He can win repeatedly and still sit outside the world top 500. The ranking does not punish him for being weak. It punishes him for being isolated.

That is the kind of conclusion reading a results list will never produce. You have to read the formula.

When I was an international communication student in Binh Duong, I started a blog called Numbers Do Not Lie and spent three months building an expected-goals model in Excel to analyse 26 rounds of a domestic football season. The output showed the eventual champion had below-average possession but one of the league's best conversion rates. I was mocked for writing against the consensus. Three months later that team won the title.

I retell that not to boast. I retell it because it taught me that data does not speak on its own. It has to be contextualised, and the person contextualising it is accountable for how the number is framed.

November 2026 and the variable no algorithm anticipated

In 2026, the pandemic closed stadiums to spectators worldwide. I was working for a football club in Binh Duong at the time, and I found that the home win rate in the domestic league dropped sharply when matches were played without crowds. The coaching staff wanted to keep the same home-and-away approach. I objected and presented a comparison table across 42 matches. The team switched to proactive defending away from home and won four of the next five.

Golf had a parallel natural experiment, and I watched it through the same lens.

The 2026 Masters was played in November, the first time the tournament had moved off April in its history, and without patrons. Dustin Johnson won with 268, twenty under par, a 72-hole tournament record. That same year, the PGA Championship was played at TPC Harding Park in August without spectators, and Collin Morikawa won. The U.S. Open at Winged Foot in September went to Bryson DeChambeau at six under par, the only player under par for the week.

The question I asked was not who won. It was which variable actually changed.

There were three candidates. Crowds. Calendar. Course conditions. For the Masters, the third candidate was the strongest and the only measurable one: the overseed followed a November schedule, temperatures were lower, moisture differed, green firmness differed. Those leave traces in data. Applause does not.

I hate uncertainty. But 2026 taught me that one unanticipated variable can be stronger than every algorithm. What I took from it for golf was not a conclusion about crowds. It was a rule for classifying variables: what can be measured goes into the model, what cannot be measured gets a note and a downgrade in confidence.

Behind those numbers are people. Morikawa won his first major at 23 in a year when the schedule was completely scrambled, with no crowds and no normal seasonal rhythm. DeChambeau won at one of the hardest courses in America with a physical strategy most of his peers considered extreme. Data describes them. Data cannot explain why they withstood that pressure. I deliberately keep that paragraph in every report I write, placed directly after the table, so the reader remembers that behind the column of figures is a person standing on a tee with wet hands.

When two tours stop speaking the same language

Since 2026, professional golf has split into two parallel systems. LIV Golf launched with 54 holes, a shotgun start, no cut, and team competition. The PGA Tour kept 72 holes with a two-round cut.

In data terms this is a far more serious event than the online arguments suggest. Two formats produce two different scoring distributions. No cut means no third and fourth rounds under elimination pressure. A shotgun start means late-hole conditions vary by tee time. A Strokes Gained figure computed from 54 holes is not directly comparable to one computed from 72 holes with a cut.

In October 2026, the OWGR board rejected LIV Golf's application for ranking points. The consequence is that one group of world-class players competed in a system that generated no points while another group competed in one that did. Anyone claiming to know precisely who the world number one was during that period is selling you a conclusion the data cannot support.

For Asian golf and Vietnam specifically, the effect sits in money flows and schedules. The international events staged in Vietnam in recent years, mostly at coastal central courses, gave domestic players the chance to compete against opponents with high field ratings. That is the only route out of the isolation trap. There is no second route.

The risk surface: reading what has not happened yet

A proper analysis must include risk. I break it into six categories: competitive, psychological, injury, career and commercial, governance, and systemic.

For a young player, injury is the category I weight most heavily and the one most often underweighted. A seventeen-year-old golfer's musculoskeletal system is not finished. Pushing him into adult competitive volume, with the rounds, the practice hours and the repeated rotational load, creates a biological debt. That debt does not appear on a scorecard. It appears later, at 24, as a back or wrist injury.

This is where performance data and physiological data contradict each other. Performance data says: play him more, he is winning. Physiological data says: stop. In most cases I have witnessed, the first one wins.

For a player at his peak, the biggest risk is a swing rebuild. An overhaul takes 12 to 18 months to stabilise, and during that window every Strokes Gained comparison loses its validity. Anyone writing the analysis must state which month of the process the player is in. Skipping that detail turns a technical report into a match recap.

For a rising player, the risk is sample size. Three good events in a row produce a trend line on a chart. That trend line has three data points. Three data points do not make a trend. They make three data points.

A chain of evidence instead of one talking number

My method for building a player profile for a coaching staff has five layers.

The first is sample. Rounds, events, time span. Under 20 rounds, I flag low confidence at the top of the report, not the bottom.

The second is score decomposition. With shot data, I split into four Strokes Gained buckets. Without it, I substitute raw data: fairway rate, greens-in-regulation rate, scoring average on par 3s, par 4s and par 5s. This is the fallback when premium data does not exist, and in Vietnam the fallback is the default rather than the contingency.

The third is course context. Grass type, green speed, prevailing seasonal wind, altitude. A course in Da Lat at over 1,400 metres of elevation makes the ball fly considerably further than a coastal course at the same temperature. Same player, two courses, two different numbers.

The fourth is tournament context. Field strength, cut or no cut, number of rounds, weather during the week.

The fifth is the gaps. I list what I do not know. That list is usually longer than the previous four combined.

The table reader and the storyteller

Golf has the widest gap in sport between what viewers see and what actually determines the result. Television shows the putt dropping on 18. It does not show the 180-metre approach on 14 that created the chance, nor the drive on 9 that forced that approach out of the rough.

Television's narrative structure produces a reputation ranking that does not match the data ranking. Players who appear often on screen are rated above what the data permits. Players who are consistent but create few moments are rated below it. The distortion is self-sustaining, because viewers remember emotion and forget scorecards.

Based on my experience attending rounds in person at various courses, I have noticed a pattern: players whose reputation ranking exceeds their data ranking tend to be those whose putting looks good to the eye, while players whose data ranking exceeds their reputation tend to be those with strong approach numbers but few highlights. The trophies go to the second group. The fame goes to the first.

Numbers do not lie — but they only speak within the sample and context in which they were collected.

This is where I have to say something plainly that will make some people in the industry uncomfortable. Most golf analysis produced for Vietnamese readers, including in places with serious investment, borrows the PGA Tour frame of reference and applies it directly to an entirely different context. Different grass. Different wind. Different event density. Different support depth. Different measurement budget.

The result is conclusions that sound highly professional and cannot be verified locally. A claim that a Vietnamese player needs to improve his putting to compete continentally may be true, may be false, and the only way to know is to measure. We have not measured.

Correlation is not causation, and that is where most analysis breaks

There is a pattern I see repeated in every market I have worked in, from football to golf. A writer finds an attractive correlation, then tells it as a causal relationship.

The typical example: the winning player posted an unusually high putting figure that week. The immediate conclusion is that putting decided the victory. But if that same player's approach figure was also unusually high, the easy putts were a consequence, not a cause. He putted well that week partly because he had brought the ball closer to the hole than everyone else.

In the other direction, a player who misses a putt on the final hole is routinely declared mentally fragile. Yet his data from the previous three rounds may show he putted above his own average all week. One miss on 18 does not erase the previous 71 holes. It only gets broadcast.

Golf is a sport with a very large denominator and a very small numerator. Four rounds is roughly 280 shots. Of those, perhaps ten genuinely decide the outcome. Ten shots are not enough to conclude anything about a skill. They are only enough to conclude something about ten shots.

Where data actually creates an edge

If golf analysis exists only to produce articles, it is a hobby. I do this job because it changes decisions.

For a national team preparing for a major event, data is valuable in three places. Selection by course profile: on a course with small greens and heavy wind, prioritise the player with a stable approach figure over the long driver. Practice allocation: if the data shows the biggest loss is from 100 to 150 metres, that is where the work goes, not the putting green. Contingency planning: if wind exceeds a defined threshold, switch the whole strategy to a low ball flight and accept fairways instead of greens.

None of those three requires a premium metric. All three can be done with raw data, provided the analyst takes the time to place each number in its proper frame.

The contingency plan is always the hardest part and the most frequently skipped. A strategy without a predefined breaking point is a prayer, not a plan. In every report I write, I ask: if metric X collapses in round one, what do we do in round two? If the answer is that we will figure it out later, the report is not finished.

What I learned from an empty table

Back to the twelve-page document in Cam Ranh.

There is another version of this story, and it is far more emotionally appealing. In that version, I look at the name, remember how he played at an event two years ago, blend in a few impressions from a round I once watched, and write a paragraph of analysis that sounds highly competent. The reader is satisfied. The person asking is satisfied. I am satisfied.

But the Strokes Gained table in my head would be empty, and I would be the only one who knew.

I do not predict. I read the data and accept the consequences. When the data is empty, the consequence I accept is an answer that sounds far less appealing: not yet determinable.

In the analysis trade, that is the most expensive answer and the one most often paid for. It is expensive because it requires the writer to give up the right to appear knowledgeable. It gets paid for because the market always rewards certainty, including certainty with no basis.

Three signals I will track in the next round

First, data structure. Whichever event begins publishing shot-level numbers, even roughly, will be the first to enable Strokes Gained analysis in Vietnam. I am waiting for that signal, and I rate the probability of it appearing within two seasons as moderate.

Second, the age curve of the young cohort. If a player under 20 competes in more than 25 rounds in a single year, I will flag him as high risk regardless of results.

Third, the flow of Vietnamese players into regional events. This is the only measurable indicator of Vietnamese golf's integration with the international ranking system, and it matters more than any domestic results summary.

Closing

An empty table is not a failure. It is a finding.

In a market where shot-level information does not yet exist, the greatest value an analyst can create is not in producing conclusions. It is in drawing the boundary between the known and the unknown precisely, and holding that boundary no matter how loudly the pressure demands an answer.

Eleven years ago I started a blog from a lecture hall and believed data would speak for itself. Today I understand that data only speaks when someone dares to act as its translator, and dares to stay silent in the places where there is nothing to translate.

That is the entire content of those twelve pages. Not a poor analysis. An honest one.