ShotLink, Strokes Gained and the Data Gap in Professional Golf

Last week, an eight-section golf analysis landed on my desk. All eight sections carried the same line: insufficient information to assess. No tournament name. No player name. Not a single data point to hold on to. The report was not wrong in its conclusions. It was wrong at the intake stage.

I kept it instead of deleting it, because gaps in a data table can speak, if we are willing to listen. After re-checking the entire data pipeline — source retrieval, formatting, extraction — I realised the problem sat deeper than a single technical fault. Professional golf runs on several very different data tiers, and most of the golfing world does not sit on the richest one.

To understand how a data table can be entirely empty, you first need to know what professional golf measures.

In 2026, the PGA Tour began publishing Strokes Gained as part of its official statistics. The method was built by Mark Broadie, a professor at Columbia Business School, using ShotLink data, and presented in full in his 2026 book Every Shot Counts. The core idea is simple: instead of counting strokes, measure the expected value of each stroke against the tour average. A three-metre putt on a fast green carries a very different expected value from a three-metre putt on a slow one.

ShotLink is the data-capture system the PGA Tour deployed in 2026. Every shot in every round is logged with coordinates, distance remaining, club type and situation. ShotLink is the infrastructure. Strokes Gained is the product running on top of it. Without the first, the second does not exist.

Golf has three data tiers, and only one of them is universal.

The results tier holds scores, finishing positions and cut rates. It exists everywhere, from the PGA Tour to the Japan Golf Tour, from Vietnam's professional events to provincial amateur tournaments. It is cheap, easy to store and needs almost no infrastructure.

The basic-statistics tier holds fairways hit, greens in regulation, putts per round and scrambling rate. It appears on most major tours, but coverage and accuracy vary sharply. The Japan Golf Tour publishes basic statistics for its own events. Many Asian events publish partial data.

The shot-data tier holds per-shot coordinates, distance remaining, expected value and Strokes Gained split into four categories: off the tee, approach, putting and around the green. This tier exists almost exclusively on the PGA Tour.

The distance between those three tiers is the real frontier of modern golf analysis. An analyst in Nagoya watching the Japan Golf Tour has the results tier and part of the basic-statistics tier. An analyst in Hanoi watching domestic events has mainly the results tier. Neither can calculate Strokes Gained: Approach, because there are no shot coordinates to calculate from.

When the shot-data tier is missing, people reach for proxy metrics. And proxy metrics fail systematically.

The classic example is putts per round. At first glance it looks like a sensible measure of putting skill. In reality it is contaminated by approach quality. A player who hits many greens in regulation leaves longer first putts, and therefore records more putts per round. A player who hits few greens but scrambles well can post fewer putts per round despite being the weaker putter.

Strokes Gained: Putting was created precisely to fix this systemic error. By comparing each putt against an expected-value baseline by distance, it separates putting skill from approach quality. But it can only do that where shot data exists.

On tours without a shot-data tier, we are still reading golf through noisy metrics. That makes models less accurate. It also shapes how players are judged. Every number is an unwritten confession: if I rank golfers by putts per round, I have admitted I have nothing better to work with.

Infrastructure cost is the real reason behind the gap.

A ShotLink-style system requires laser measurement devices, processing software and an operating crew at every hole of every round. The PGA Tour has broadcast rights revenue large enough to sustain that system for more than twenty years. A regional tour does not have equivalent revenue.

I am not quoting a specific cost figure, because I cannot verify a source reliable enough to do so. But the structure is clear: shot data is an expensive commodity, and it only appears where someone can pay for it.

Hideki Matsuyama won the Masters on 11 April 2026, becoming the first Japanese man to win a major championship. That victory took place inside a data ecosystem entirely different from the one where he started. While he was playing the Japan Golf Tour, nobody measured his Strokes Gained: Approach. When he moved to the PGA Tour, every shot he hit began to be logged. I keep an open question: if that infrastructure had existed in Japan ten years earlier, how many other Matsuyamas would have been seen?

At another end of the same ecosystem, Rory McIlroy completed the Career Grand Slam at the Masters on 13 April 2026. Every shot of his final round carried complete Strokes Gained data, published within hours. A young golfer in Southeast Asia having the same moment at a national championship will leave no data trace at all.

The 2026 season taught golf a lesson few people want to revisit.

On 12 March 2026, the PGA Tour halted its entire schedule because of the pandemic. It was 11 June 2026 before the Charles Schwab Challenge at Colonial marked the return. Three months without an official round. For a model built on form sequences, three months is a gap wide enough to render every weight meaningless.

I lived through a smaller version of this problem while working with training data for a football club in Japan in 2026. No matches, no performance metrics, only GPS data from training sessions. The solution was to shift focus from outcome data to process data: running volume, intensity, movement symmetry. It could not predict who would win. It could only estimate who was ready to compete.

Golf did the same thing, later. The 2026 Masters moved to November, the first time in the tournament's history. Season-long form sequences were scrambled so badly that many ranking models had to reduce the weight of recent rounds. When data hides its face, error becomes the guide.

The Official World Golf Ranking is a system designed to coexist with missing data.

OWGR does not use ShotLink. It uses a player's finishing positions, the strength of the field and a time-decay factor. It does not need to know which shots were good. It only needs to know who beat whom, at which event, and how long ago.

That approach has a clear advantage: it works on every tour, including those without shot data. The disadvantage is equally clear: it is a system-level proxy, and every proxy has limits.

In October 2026, OWGR rejected LIV Golf's application for ranking points. The reasons given centred on event structure: 54-hole rounds, no cut, and a team format. The consequence was that many LIV players fell out of the world ranking, which in turn affected their access to major championships.

Seen through a data lens, this is a valuation problem. When there is not enough data to assess quality, a system tends to refuse to price rather than price wrongly. That is understandable as a design choice. It also shows the true cost of a data gap: it does not stop at inconvenience for analysts. It reaches into golfers' careers.

Japan and Vietnam face two versions of the same gap.

Japan has a long-established professional tour, a statistics system and a training culture I have followed for more than a decade. The Japan Golf Tour publishes the results tier and most of the basic-statistics tier. What Japan lacks is the shot-data tier: infrastructure to log every shot across an entire schedule.

Vietnam sits elsewhere. The number of golf courses has grown fast over fifteen years, mostly serving tourism and recreational play. Data infrastructure starts close to zero. Results from domestic professional events are stored mainly as scores and finishing positions.

For an international recruiter, the consequence is concrete. A nineteen-year-old Vietnamese golfer may post strong results, but the file sent abroad contains only scores. No Strokes Gained, no distance distribution, no putting data. The decision-maker has to choose based on something he cannot read.

Based on my experience following matches and tournaments in both places, I would argue the gap sits in record-keeping infrastructure, not in talent. A golf nation can produce good players without ShotLink. But that nation will keep being undervalued, because the market only pays for what it can read.

My assumption at the start of this piece was that data gaps are always harmful. I ran the numbers against that assumption myself, and the result forced me to change the question.

There is an argument that missing data sometimes protects players. When there are no detailed metrics on an eighteen-year-old, nobody labels him early. No ShotLink means no poor Strokes Gained figure, no mispriced valuation. In other sports I have watched advanced metrics box in young players before they finished developing. Golf could repeat that script if shot data spreads globally.

But when I tested the argument against historical data, it collapsed faster than I expected. Data gaps are not randomly distributed. They cluster on small tours, in low-money regions, and around young players without strong representation. The gap protects those who already hold an advantage, and shackles those who hold nothing.

Data is never wrong; I am simply asking the wrong question. The question is not whether we should collect more data. The question is who controls the quality of that data, and who has the right to correct it when it is wrong.

I also have to admit something about myself. There was a period when I used the missing-data argument as a shield for my own faulty models. When a forecast missed, I cited incomplete inputs instead of admitting my assumptions were wrong. The line between honest acknowledgement and sophistry is thin. The only way to stay on the right side is to label everything clearly: this part is data, this part is inference.

The eight-section report at the start of this piece did exactly that. It did not guess. It stated plainly that it did not know.

What is worth watching in the next round sits in the infrastructure, not in the leaderboard.

If the DP World Tour extends ShotLink coverage to more events, if the Japan Golf Tour signs with a shot-measurement provider, if a professional event in Vietnam begins logging shot coordinates — those are the signals that change the sport, before they ever become headlines.

If none of those signals appear, we can still keep analysing golf. We just have to be honest about how much of our analysis is data, and how much is inference.