Fifteen Years of Strokes Gained: When Japanese Golf Still Lacks Shot-Level Data

Fifteen Years of Strokes Gained: When Japanese Golf Still Lacks Shot-Level Data

The seventh column in my JLPGA tracking file has been empty for four straight seasons. The header reads "SG: Approach." Below it sit more than four thousand rows of tournament names, dates and round scores — and not a single filled cell.

I reopen that file every April, out of a professional habit I cannot shake. In April 2026, days after Hideki Matsuyama wore green at Augusta National, I sat in my Nagoya office and asked myself a very specific question: if Japanese golf truly is "the nation of iron players," that fact must appear in approach metrics. To test it I needed shot-level data. I did not have any.

Fifteen years before that, a finance professor at Columbia Business School named Mark Broadie published a measurement method that changed how the sport reads a round. The PGA Tour made it an official statistic from the 2026 season. The metric is called Strokes Gained. As of today, the two largest professional tour systems in Japan still have no publicly available equivalent.

Gaps in a table speak, if we bother to listen. What this one says is not small: everything the Japanese public — and Vietnamese readers who follow Japanese coverage — knows about the shot quality of its own players rests on a set of metrics that Western analysts stopped treating as the primary yardstick roughly fifteen years ago.

ShotLink and the cost of measuring every shot

Strokes Gained runs on an idea so simple it is easy to dismiss: every ball position on a course carries an expected number of strokes to hole out. From 150 yards in the fairway, an average PGA Tour player needs about three strokes. From 150 yards in the rough, the number is higher. From eight feet on the green, far lower.

When a player hits a shot, we compare the expectation before and after. The difference, minus one stroke, is the value of that shot. Sum every shot in a round and you have that round's Strokes Gained. Split it by category and you get four buckets: Off-the-Tee, Approach the Green, Around the Green, Putting.

The hard part is infrastructure, not arithmetic.

To compute SG, every shot must be logged as coordinates: where the ball lies, how far from the hole, from which lie. The PGA Tour does this with ShotLink — laser devices and personnel stationed at every hole, for every group, throughout the event. It is a logistics machine, not a spreadsheet.

I once stood through an entire JLPGA round in Aichi to count by hand. I captured ball positions after every shot for three players in one group. By the 12th hole I had two errors, because one player changed clubs after the ball was airborne and a rules official had to intervene. One round, three players, one recorder. At that scale I cannot cover a forty-event season.

The JLPGA and JGTO publish round-level aggregates: scoring average, fairway-hit percentage, greens in regulation, putts per round, average driving distance. Those numbers are real and useful. They are counting statistics of outcomes, not valuation statistics of difficulty. An approach from 190 yards into a bunker-guarded green and an approach from 120 yards into a wide green both count as a green in regulation if the ball finds the putting surface. The metric cannot tell two actions of very different difficulty apart.

That is why my seventh column is empty. I have a method, time, and raw scoring data. What I lack is shot coordinates in public form.

Why the infrastructure never appeared

Cost is the first answer and the easiest one. But cost does not explain everything, since the JLPGA is among the richest women's tour systems in the world and the JGTO has the resources to stage large events.

The second answer is structural. Japanese professional golf survives mainly on partnerships with corporations and host clubs, not on broadcast rights. When revenue comes from sponsors rather than from television, the incentive to invest in granular data that enriches broadcast content is much weaker. ShotLink in the United States is fed by a television machine that needs a new story every week.

The third answer is rarely spoken aloud: shot-level data can unsettle stories that have already been handed down. If the approach numbers of Japanese players are not better than the rest of the world's, then the claim that Japanese golf is a golf of iron craftsmanship has to be rewritten. No institution moves quickly toward a table of numbers that may contradict its own identity.

I have no evidence that the third answer is the actual cause. I keep it on the hypothesis list because it is the only one that explains why people have tolerated the gap for so long.

Vietnam starts from a different position

Vietnam's professional and semi-professional tour system is far younger, with fewer events and narrower broadcast reach. But it holds a structural advantage that few people name correctly: no legacy infrastructure to defend.

The cost of shot-tracking sensors has fallen sharply in recent years. Devices that clip onto the grip, log GPS coordinates and recognise shots automatically now sit within reach of an ordinary amateur. A young tour system can begin with large-scale amateur data instead of waiting for American-style infrastructure.

I have to state the limit clearly, because this is where people fool themselves. GPS data from grip sensors carries positional error measured in metres, not yards. It is enough to compute trends and classify errors; it is not enough to calculate SG to two decimal places. When data hides its face, error becomes the guide — and a guide must be named in the report, not buried in an appendix.

Four categories, four kinds of questions

| Category | What it measures | The question it answers | Known limit | |---|---|---|---| | Off-the-Tee | Value of drives and par-3 tee shots against the tee baseline | How many strokes a player's driving gains against the field | Includes par 3s, so it is noisy for short hitters | | Approach the Green | Value of shots into the green from outside it | Distance control and accuracy as a skill | Highly sensitive to how the lie is recorded | | Around the Green | Chips, pitches, bunker shots near the green | Ability to save par after missing the green | Small samples, large week-to-week swings | | Putting | Value of strokes on the putting surface | Conversion of chances into score | The most widely misunderstood of the four |

This table is not for memorising. It is for a more practical purpose: when you read that player X has "an excellent short game," ask which bucket the claim rests on, and whether that bucket has enough sample to say anything at all.

The persistence problem and the putting trap

Here is the point I want readers to carry away: a metric has predictive value only if it repeats across seasons. If a player's SG: Putting jumps from +0.4 to +0.9 and back to +0.2 across three seasons, the middle figure is not evidence of improved putting skill. It is noise wearing a headline.

Broadie showed this in his research: of the four categories, putting is the least persistent season to season, and approach is the most persistent. The reason is more physical than psychological. Whether a 12-foot putt drops depends on things outside control — a grain of sand, a pitch mark, green moisture shifting through the day. A 7-iron approach from 175 yards depends on technique, and technique repeats.

The consequence for media is clear. If you get one metric to tell a player's story, choose approach. If you choose putting, you are telling the story of this week, not of a career.

I do not believe in luck; I believe in cultivated probability.

Four names, and one test I cannot run

Hideki Matsuyama, April 2026, Augusta National. The familiar Japanese framing is that he won with nerve and iron play. That framing may be correct. To prove it, I would need his SG: Approach across four rounds against the field baseline. I have the scores. I do not have the rest.

Hinako Shibuno, August 2026, Walton Heath. Collective memory ties her to a smile and a run of putts down the stretch. I have no shot-level data for her that week. And here is the real problem: collective memory preserves what shocks, not what decides. A holed 15-footer is remembered forever; ten approaches that found the correct part of the green are not.

Ayaka Furue, July 2026, Evian Championship. The familiar framing is "small swing, large output." Again, I have the result and not the process.

Yuka Saso, US Women's Open champion in 2026 and 2026. Two events, two different course contexts, two results under one name. That is an ideal sample for a test: which metric repeats across two victories? Without data, I cannot answer.

Every number is a confession not yet written into prose.

Vietnam and Japan at the operational level

| Factor | Japan | Vietnam | |---|---|---| | Course density | Dense, many long-established private clubs | Sparse, concentrated in cities and resort areas | | Driving-range culture | Widespread, focused on technique and repetition | Expanding, tied to recreation and fundamentals | | Cost of a weekend round | High relative to median income | High relative to median income, wide variation by course | | Junior tour system | Long-standing, tied to schools and clubs | Forming, small but growing fast | | Public shot-level data | None | None |

The last row is identical in both columns, and it is the most important row. Two golf nations at different stages share the same data blind spot. For Japan it is the product of a settled ecosystem that resists change. For Vietnam it is the product of an ecosystem that has not yet had time to build.

Where I have to break my own argument

Throughout the section above I quietly assumed that having Strokes Gained would make the story clearer. That assumption deserves challenge.

Correlation is not causation. High SG: Approach and a victory occurring in the same week does not prove that approach play produced the win. A player can hold average approach numbers and still win because rivals faltered late. A metric describes a round that already happened; it does not describe the mechanism that produced the result.

Second, we only have data on winners. Full leaderboards are published, but public analysis clusters almost entirely around the top of them. Compute the field average of a stat and you see a distribution. Compute only the champion's and you are measuring a sample already selected by outcome, then using it to explain the outcome. That loop is closed and empty.

Third, and this is the point I most want to press: the data gap creates a systematic bias, and the bias tilts toward putting. Why? Because putting is the easiest category to see with the naked eye on television, the easiest to log in a single line of notes, and the easiest to narrate in one sentence. When infrastructure is missing, people tell stories with whatever is easy to tell. The result is a generation of fans taught that golf is decided on the putting surface.

What did NOT happen often tells the truth more plainly than what did. We have no approach data for Japanese players; we have an abundance of memories about their putts. That imbalance reflects the nature of record-keeping, not the nature of golf.

I once asked the wrong question in exactly this way. In 2026, working as a data analyst for a J.League 2 club, I built a prediction model from video and omitted the home-venue factor across a four-match losing run. The model missed six of the last ten rounds. Data is never wrong; I simply asked the wrong question. The lesson transfers directly to golf: if I ask whether a player putts well and all I have is putts per round, I am asking a question my data cannot answer.

Signals for the next cycle

Four things I will watch next season, written down so I can later check whether I bet correctly.

First, the adoption rate of shot-tracking sensors among Vietnamese amateurs. If the user base grows large enough, aggregate data will appear before any tour invests in infrastructure. That is the reverse of the Japanese and American paths.

Second, whether Japanese broadcasters push harder on data graphics during live coverage. Content demand usually precedes infrastructure rather than following it.

Third, the quality of lie recording in any new data system that appears in Asia. If the lie is logged incorrectly, every downstream SG figure drifts, and no algorithm repairs that error.

Fourth, a question I have no way to verify yet: whether the claim that Japanese golf is a golf of iron craftsmanship survives its first encounter with Strokes Gained. My suspicion is that it survives, but in a narrower, more specific and more conditional form. Every good conclusion should look like that.

The seventh column in my file is still empty. But I have renamed it, from "SG: Approach" to "question I cannot yet answer." An honest spreadsheet labels what it lacks.