When Golf Data Goes Quiet: The Standard of Saying 'Insufficient Information' in Major Week

At 9:40 on a Sunday night I opened the file for the week ahead: four Strokes Gained columns, three distance columns, one greens-in-regulation column. The sheet ran to 4,312 rows. Every value was blank. No red flags, no error notices. The feed simply had not pushed yet.

The young analyst beside me said the line I have heard at least twenty times in eleven years in this trade: "Just use last season's numbers. Nobody checks."

I closed the laptop. By eleven that night I had rewritten the entire opening section, deleted every comparison table, and typed one line into the internal document that coaches never enjoy reading: "Insufficient information to conclude."

That was the moment I understood something golf scoreboards can never catch. A wrong number gets corrected by somebody. A correct number that means nothing survives forever, quoted from article to article, season to season, until it becomes part of the myth.

Data does not lie. Reputation whispers into the ear of anyone who does not read the table.

A data ecosystem that is thick at the top and empty at the bottom

Golf has an unusual information shape. The PGA Tour has run ShotLink since the early 2000s, capturing nearly every shot by every player, down to putt distances measured close to the millimetre. Mark Broadie of Columbia University built the Strokes Gained framework on top of that, and the PGA Tour added it to its official statistics in the early 2010s. Since then, the question "who is playing well" has had a quantitative answer.

Only at that level, though. The DP World Tour has far thinner coverage. The Asian Tour, developmental circuits and national amateur events have almost no shot-level data at all. In Vietnam, events on the VGA Tour publish gross scores and hole-by-hole results, occasionally driving distance, but no Strokes Gained, no ball speed, no green topography.

The practical consequence is simple. When a Vietnamese player shoots 68 on a 6,800-yard course, we know the result and not the process. We do not know whether he gained strokes on approach and lost them on the greens, or the reverse. We do not know where his fourteenth tee shot finished. We do not know what a shifting wind at the sixteenth cost him in par probability.

A writer then faces two choices. State plainly that the data is insufficient and analyse what genuinely exists, or import PGA Tour benchmarks onto a different course, climate and grass type and produce something that sounds authoritative.

I have seen the second option used often enough to know how dangerous it is. It is not arithmetically wrong. It is wrong because it maps one frame of reference onto another place's data.

Strokes Gained and the small-sample problem

Strokes Gained is the best tool the sport has. It answers what a scorecard cannot: how much advantage a player gained in each skill area against the field. Good tools still break, and this one breaks at sample size.

A PGA Tour round gives roughly 70 shots. Putting accounts for about 28 to 30 of them. Split that into putts inside three metres, three to six metres, and beyond six metres, and each cell holds only a handful of shots. Four rounds stacked together still leaves a small sample. That is why serious analysts use a rolling 20 to 30 round window for putting skill and a full season for approach play.

Putting is the highest-variance skill in golf and the one most carelessly cited in coverage. A player who putts well for a week has not proved he is a good putter. He has just had a week where probability leaned his way. Equally, three poor putting weeks do not prove decline, because across roughly 90 putts the standard error is large enough to swallow the gap.

If my method had to fit into one sentence, it would be this: a metric may only speak after its sample size, its frame of reference and its uncontrolled variables have been declared. Without all three, it is merely noise.

Player form: the age curve and the small-sample trap

Four data layers must be separated when assessing form: world ranking, tour tier, recent results and position on the age curve.

The Official World Golf Ranking is weighted by recency and by field strength, but it cannot distinguish a player protecting accumulated points from one climbing on a hot run. Two players can sit side by side in the ranking while moving in opposite directions.

Tour tier matters more than most coverage admits. A top five in a weak field is not equivalent to a top twenty at a major once ranking points are counted. This is introductory knowledge that still gets ignored.

Recent form is the most inflated layer. A five-event window can produce a beautiful trend line that vanishes in a fortnight.

The age curve is the most underrated layer. Golf peaks between roughly 27 and 33 for power players, and can extend further for those who live on accuracy and experience. A 22-year-old with three straight top tens has not proved anything about his ceiling. He has proved he outplayed the field across twelve recent rounds.

A 38-year-old in decline is also not finished. Injury, scheduling, equipment changes and variables absent from any table, such as a newborn at home or a sponsorship collapsing, can explain most of a gap a model cannot see.

I have to be honest about one limit of my own profession. Locker-room chemistry, caddie relationships, family pressure, comfort with a course: none of it is quantifiable. The analytics market overprices youth potential because it is easy to measure and underprices intangibles because they are not.

Tournament strength: field quality and ranking points

A major and a regular tour stop share a name type but not a frame of reference. Field strength, the number of top-50 players present and the ranking points distributed determine what a result actually means.

This is where comparisons drift. A player who wins twice in weak fields can end a season with fewer ranking points than one who posted three top tens in strong fields. Counting trophies alone ranks them wrongly.

Pressure is real and measurable in effect. The same three-metre putt has a different success probability in round one of a regular event than at the 18th on a Sunday at a major. I once built a putting-performance table by round for a group of professionals and found the widest gap in the final four holes of the last round. The data answers, provided there are enough rounds to speak.

Team events add another layer. Selection logic, order of play and four-ball or foursomes chemistry create new variables. Judging a player by team results without adjusting for format is methodologically wrong.

Governance: the PGA Tour, LIV Golf and the grey zone

This decade produced a split that data cannot heal. LIV Golf arrived with funding from Saudi Arabia's Public Investment Fund. Three dates matter: LIV's first season began in June 2026; on 6 June 2026 the PGA Tour announced a framework agreement with PIF; and in October 2026 the OWGR declined LIV's application for ranking points, citing format and lack of a qualifying mechanism.

What deserves attention is how data got dragged into the fight. A player who moves to LIV leaves the ShotLink system. He no longer has official Strokes Gained. Analysis of him must then rely on third-party data, self-collected data or estimates from scorecards, sources with very different reliability. Merging them into a single comparison table is a methodological error.

A player disappearing from official data does not become weaker on the course. He becomes weaker on the spreadsheet, and the spreadsheet is what most of the public reads.

Rules and equipment: the ball rollback and the limits of forecasting

On 6 December 2026, the USGA and the R&A announced revised golf ball testing conditions designed to limit distance. Elite competition adopts them from January 2026, recreational play from January 2028. Earlier, the anchored putting stroke was banned under Rule 14-1b from 1 January 2026.

Both cases show where impact modelling usually misses. We can forecast direction, since a ball flies shorter at high swing speed, but not magnitude. Players adapt. They change clubs, attack angles, training loads and course strategy, and each adaptation creates a new baseline.

I usually present three scenarios internally. The bad case: distance loss breaks the advantage of long hitters and forces majors to lengthen courses further. The neutral case: distance drops but accuracy compensates, leaving the distribution of results unchanged. The good case: slower ball speeds preserve the value of older courses, benefiting traditional events in Asia and Europe.

All three are assumptions. I label them as such. That is the line between forecasting and fabrication.

The risk surface: six layers

When assessing a player or a tournament plan, I separate risk into six layers: competitive, psychological, injury, career and commercial, governance, and systemic.

Competitive risk sits in field quality. Psychological risk sits in the ability to hold decision quality over the last 18 holes. Injury risk tracks rounds played, age, history and training load we never see. Commercial risk attaches to contracts, sponsors and broadcast rights. Governance risk attaches to whether a player belongs to a recognised system. Systemic risk covers what nobody controls: weather, compressed schedules, an outbreak.

2026 taught me this the expensive way. With stadiums closed to fans, home advantage in the domestic league all but vanished: home win rate fell from 49 percent in the 2026 season to 38 percent. The coaching staff wanted to keep the same home and away approach. I pushed back, presented a comparison across 42 matches played without crowds, and proposed active defending away from home. The team won four of the next five.

Empty stadiums in 2026 made me ask whether home advantage comes from the ground or from the crowd. Data has an answer. An unforeseen variable can outweigh any algorithm, and the only way to live with it is to treat it as its own layer rather than hiding it inside the error term.

Public narrative and the expectation cycle

Every player gets assigned a story: coronation, redemption, defection, the career Grand Slam chase. These stories have their own heat cycles, and the cycle usually runs ahead of the data.

The clearest recent example is the career Grand Slam. Rory McIlroy completed the set by winning the Masters on 13 April 2026, after years of questions about that specific tournament. Before he won, market expectation had run far ahead of actual ability, then fallen far below it. Both phases produced mispricing.

My test for any narrative: does it rest on a data foundation, and over how many rounds. A rising-star story usually rests on 12 to 20 rounds. That is enough to describe the present and not enough to forecast the future.

A parallel story concerns generational turnover. Players born in the early 1990s still hold ground through major experience, while players born after 2026 have entered the world's top ten at unprecedented speed. Ludvig Åberg is the clearest case, moving from amateur ranks to a Ryder Cup team in under a year. But that speed also means the comparison baseline does not yet exist.

The sport's transmission chain

Golf runs on three tiers. Upstream: courses, equipment and talent development. Midstream: tours and event operators. Downstream: broadcast, sponsorship, data and derivative products.

A small change in equipment rules flows downstream across two to four years. Manufacturers rework product lines. Event operators reassess course design. Broadcasters adjust how distance is presented. Data platforms rebuild models. Each of those steps introduces new variables for anyone forecasting outcomes.

In Vietnam the chain is shorter but no less interesting. Courses multiply, domestic events multiply, young amateurs multiply, while the data layer stands nearly still. That is the biggest bottleneck for Vietnamese golf over the next decade: not a shortage of talent, but a shortage of the ability to measure talent.

The counterintuitive angle: a blank cell is itself data

A common belief holds that empty data means nothing to say. That is false.

When Strokes Gained does not exist for a group of players, that tells us who gets measured and who is left behind. When an event publishes no shot data, that tells us how professionalised it is. When a player declines to share equipment data, that is a signal too.

The most dangerous habit in this profession is the impulse to fill gaps. People complete an incomplete pattern rather than leave it incomplete. In analysis this produces four error types.

First, turning probability into certainty: a model showing 62 percent becomes "almost certain to win". Second, mistaking correlation for causation: a player changes clubs, wins the next week, and the club is declared the cause, even though he changed clubs four times before without winning. Third, imposing one system's data standard on another: ShotLink benchmarks do not transfer to an amateur event on a tropical course. Fourth, being so cold that you lose the human voice, so the warning never reaches the 19-year-old player or the 60-year-old manager who needs to act on it.

I dislike uncertainty. But 2026 taught me that an unforeseen variable can be stronger than any algorithm. Since then I write the "insufficient information" section first and the conclusion afterwards. If the first cannot carry the weight, the second does not deserve to be read.

What I carry forward

Eleven years ago I started a blog from a lecture hall, believing data would speak for itself. I built a model on a spreadsheet to analyse 26 rounds of a domestic season. It showed the eventual champion averaged only 48 percent possession, lowest among the leading group, with one of the league's best conversion rates. I wrote that champions do not need the ball and was mocked for it. Three months later that team lifted the trophy.

The lesson was not that I predicted correctly. It was that my model held no data on chance quality, only shot counts. I was right for a different reason than I believed. Had I been wrong, I would not have known where.

Data does not lie. It only speaks when placed in the right position.

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

What to watch in the next round

This week my sheet will open with blank columns again. The question is not how to fill them but which blanks are warning of real risk and which are merely weak infrastructure.

I will track three signals: the shot-data coverage rate of the event, the distribution of ranking points across the field, and the gap between media expectation and measured ability. When those three drift far apart, that is where the opportunity sits.

If you are a reader, once this week check whether the metric in your favourite article comes with a sample size. If it does not, you are reading a story, not an analysis. And if you are a writer, the hardest question is never what data you have. It is what data you are missing, and whether you are willing to say so.

Each season convinces me further that the best analyst is not the one with the most data. It is the one who knows exactly where their data ends, and says so out loud.