The Blank Scouting Sheet in Mid-Season Badminton: Why N/A Is the Most Trustworthy Data

At three in the morning in Shanghai, I reopened a spreadsheet named BWF_MS_Season_Review. Forty-seven rows, each one a men's singles player I had tracked across three weeks of screens and four live broadcasts of the BWF World Tour. The name column was full. Every other column — rally-length distribution, unforced error rate across the final five points of a game, average movement amplitude per rally, net-pressure index, recovery time between games — sat untouched.

The emptiness was not the emptiness of unfinished work. It was the emptiness of data that does not exist in any public source. The Badminton World Federation publishes results, schedules, rankings and a handful of basic metrics at major events. Everything else — distance covered, landing distribution, the breathing rhythm of a game — sits inside national teams, inside coaching camera systems, and inside files that nobody hands to a journalist.

That night I did something the version of me from eleven years ago would have called failure. I left the empty cells empty, typed N/A into each one, and wrote a line at the top of the file: "Data never lies; it only stays silent in front of the wrong question."

A trade that survives by filling gaps too quickly

There is a tournament every week. Each tournament has thirty-two men, thirty-two women, and in doubles the number doubles again. Every round generates hundreds of headlines, thousands of summaries, tens of thousands of comments. Sports media is not paid to say "I don't know". It is paid to publish.

I understand that pressure better than most. I once filed four pieces in a single night after the quarter-finals of a Super 1000, simply because the desk needed copy before Chinese readers woke up. Of those four, one I am proud of, two are acceptable, and one I still remember with discomfort: I assigned a metric I had never measured, purely so the sentence would sound more certain.

The mechanism behind that mistake is very specific. When a data cell is empty, the writer's brain fills it with the most plausible option. A player who wins the last five points of a game is obviously "mentally stronger". A player who covers more ground in the third game obviously has "a better physical base". Those sentences are rhetorically true and evidentially empty. A badminton game lasts roughly eighteen to twenty-two minutes. In that window, the number of rallies usually falls between seventy and ninety. That is a tiny sample for any causal claim.

I have sat in enough arenas to know that a single shuttle landing on the line, a single line call, a single stroke half a millimetre off the frame can decide a twenty-one-point game. When the decisive variable sits inside the error term, every model built on the final score is built on sand.

Nine analytical dimensions, and nine times I had to write N/A

I run my evaluation across nine layers. Not because badminton needs nine, but because I learned the hard way in 2026. When the pandemic froze every competition, I built a prediction model on three thousand eight hundred European football matches across ten seasons. In the first month after football returned, it was right sixty-eight per cent of the time. In the second month, forty-seven per cent. When the model collapsed, I started listening to the noise. That lesson followed me into badminton and turned nine analytical layers into nine opportunities to admit ignorance.

The first layer is technique and tactics. To call a player "attacking", I need at least eight to ten matches across different venues against different types of opponent. Three matches at one indoor event in Asia is not enough to separate style from conditions. That column gets N/A.

The second layer is form and individual data. The BWF ranking system uses a fifty-two-week window and a player's best ten results. It is a lagging indicator by design. A player can be performing above their ranking for six weeks, or trailing it for six weeks, and the list will not tell anyone. To measure real form I need opponents, game scores, game durations and physical condition after each match. Public sources give me two of those four. The column gets N/A.

The third layer is tournament structure. Here I hold real data: Super 1000, Super 750, Super 500, Super 300, plus the World Tour Finals. A thirty-two draw, eight seeds, random allocation. I can calculate the probability that a seed meets a difficult opponent in round two. I cannot calculate the probability that a player cramps in the third game. That column gets half data and half N/A.

The fourth layer is the world landscape. Here I am allowed to say a great deal, because the picture is large enough to carry the sample. The leading men's singles group in this cycle still circles familiar names: Viktor Axelsen with Olympic gold at Tokyo 2026 and Paris 2026, Kunlavut Vitidsarn with the 2026 world title and silver at Paris 2026. In women's singles, An Se-young won the 2026 world championship and then Olympic gold at Paris 2026, while Tai Tzu-ying held the world number one position for a long stretch and took silver at Tokyo 2026. Those rows are filled, because they are history that happened and was recorded in multiple independent sources.

The fifth layer is regulation and institutions. This is the layer that irritates me most and that I believe matters most in a standard season. Badminton has published rules on withdrawals, mid-match retirements, participation obligations for leading players, and injury handling. The actual physical condition of a player is not published. I once sat in a press conference where a player said he "felt fine" after leaving court with a taped ankle, then withdrew from two consecutive tournaments three weeks later. Nobody lied. Medical information is managed as an asset, and it is released only when release benefits someone.

The sixth layer is coaching and support systems. This is where cross-border work shows its value, and where the difference between Indonesia and China is clearest. One tradition builds on shuttle feel, wrist rhythm and long sparring sessions. The other builds on programmed training volume, weekly video analysis and load management by numbers. Identical figures in those two places can tell completely different stories, and the world ranking will never distinguish between them.

The seventh layer is the risk surface. I build a matrix of injury, schedule density, ranking pressure, personnel structure, regulation and public opinion. Of those six, I can measure exactly one precisely: the schedule. Rest days between events, matches within days — that is arithmetic. For injury I have only disclosure, and disclosure always lags the injury itself.

The eighth layer is public narrative. Here the only data I genuinely possess is data about myself: I know which story is currently pulling me along. When an entire community celebrates a player after three rounds, I know I have to wait. After 2026, I no longer trust winning streaks, I trust cycles.

The ninth layer is industry transmission. A title win lifts racket sales, broadcast contracts and junior academy enrolment. But the lag differs at every link, and I have never seen a study that measures the transmission coefficient from a Super 1000 title to the number of children enrolling in a Vietnamese provincial academy. That column gets N/A, with a note: this is a question worth funding.

What a bronze medal from 2026 taught me

There is one way to fight the impulse to fill gaps, and I found it in the file of Nguyen Tien Minh.

In 2026, at the World Championships in Guangzhou, he won bronze in men's singles. It was Vietnam's first World Championship medal in badminton. In the same year he entered the world's top five. A player born in 2026, from a country with no deep badminton development tradition at that level, competed at four consecutive Olympic Games and was still on court past the age of forty.

If I look only at the ranking, the story is a curve rising and then falling. If I look only at win totals, the story is a data series about ageing. But when I sat through scattered recordings spanning more than a decade, what I saw was something else: the ability to restructure movement patterns to compensate for lost speed. No ranking records that process, because rankings record outcomes.

From the 2026 SEA Games, I learned that data needs time to whisper. I do not write about Tien Minh by adding and subtracting wins. I write about him by contrasting what the metrics cannot say: how he chose placement against younger opponents, how he controlled the tempo of the opening game, and how he accepted losing one game to buy another.

Nguyen Thuy Linh and the problem of samples that are too small

In women's singles, I have followed Nguyen Thuy Linh across several seasons. Born in 2026, she competed at the Tokyo 2026 Olympics and has been inside the world's top twenty-five. Every time she beats a top-twenty opponent, Vietnamese forums erupt into a new round of analysis: "she belongs at this level", "she lacks nerve", "she needs a new game plan".

I tried to test those conclusions with data. The problem appeared immediately. In one season, a women's player ranked in the world's top thirty typically meets a top-ten opponent three to five times. Each meeting is a twenty-one-point game, times two or three. The number of rallies between the two across an entire season may not be enough to compute a reliable coefficient.

If someone tells me a player "wins seventy per cent of points after taking the lead", I ask over what horizon: twenty minutes, three months, or one season. The answer changes everything. Across twenty minutes, every ratio is essentially random. Across a season, a seventy per cent figure can start to carry meaning. Between those two points lies the zone where I have learned I am not permitted to conclude.

I still remember an evening I spent four hours logging every rally from three matches of one women's player. The result: her win rate was markedly higher in rallies lasting more than fifteen shots. I nearly wrote a piece concluding she had "superior physical endurance". Then I checked again: two of the three matches took place in a poorly air-conditioned arena at high temperature, and in both her opponent had come through a three-game match the previous day. The correlation I found was not causation. It was playing conditions wearing the clothes of data.

Injury: where public data is deliberately blindfolded

Of my nine layers, this is the one that produces the most N/A entries.

Badminton demands load tolerance from shoulders, ankles, knees and lower backs. Its common injuries are not collision injuries but accumulation injuries. Those do not appear in any statement until they have already become a problem.

I witnessed one clear case: Kento Momota, after winning the world title in 2026 and 2026, was in a road accident on the way to the airport in Malaysia in January 2026. In data terms, that is a break point. In human terms, it is a long sequence of recovery months that media could only approach through a few short announcements. What the public received was "back in training", "will compete at the next event". What actually happened sat behind a closed door.

Medical confidentiality has a direct consequence for my trade. When I do not know which stage of recovery a player is in, I cannot separate a dip in form from a body refusing orders. I would rather write N/A than write "declining form" when the only thing I know is a six-word statement.

I proposed a different format to two outlets: an injury piece that reaches no conclusion, listing only what was published, when, by whom, and explicitly listing what was not published. Both declined. Readers do not read to learn that some things are unknown. I understand. But that is precisely why that column in my sheet will always be full of N/A.

When I stopped predicting and started explaining

There is a line I use to explain my method to colleagues: a transfer is not the purchase of a player, it is the purchase of a gap in a system. Badminton rarely has football-style club transfers, but the logic is identical. When a national team slots a new player into a men's doubles position, the question is not how good that player is. The question is which gap in the system needs filling, and whether that player can fill it in six months or eighteen.

That is why I moved from prediction to explanation. A correct prediction helps little if I do not understand why it was correct. An incorrect prediction teaches nothing if I do not know which link in the reasoning chain broke.

In daily work I apply this with one mandatory question before any conclusion: what could make this model wrong? If I cannot answer that, I do not have a conclusion. I have a sentence that sounds good.

The numbers I refuse to place side by side

Over eleven years of watching the industry, I have seen one harmful habit spread: placing side by side numbers generated in different frames of reference.

A player wins eighty per cent of matches at Super 300 level and thirty-five per cent at Super 1000 level. Those two figures do not measure the same capacity. They measure two levels of competition, two schedule densities, and often two different physical phases. Combining them into a single "composite form index" is the act of manufacturing information that does not exist.

I have argued about this many times, and I have been rude in some of those arguments. That is my flaw. When someone rebuts data with feeling, I react as if data were a faith to defend. But xG is not a verdict, it is a lens. And a lens shatters when people use it to photograph something outside its field of view.

Expected-value metrics have been badly overused. They cannot explain a player's decision at a decisive point. They cannot explain why someone chooses a deep high serve instead of a short serve at nineteen-all. They do not reach psychology, and they are entirely blind to how a particular referee calls a particular day. Anyone treating an expected-value metric as the final answer is reading a lens and declaring it the whole panorama.

I still use it. I use it as a third cross-check, after I have watched the match and after I have read my handwritten log. If three sources agree, I write. If only two agree, I write with a warning. If only one, I do not write.

A season is a system of equations

There is a difference between football and badminton that took me years to fully grasp. In football, a season has thirty-eight rounds and eleven players per side. Large sample. In badminton, a season may have twenty tournaments, but an individual player competes in only fifteen to twenty matches, each of two or three games, each of about twenty points. The sample is small enough that every point carries weight.

This means the season's system of equations has more unknowns than equations. Injury is an unknown. Motivation is an unknown. Coaching change is an unknown. We observe only the visible portion of that system, and the visible portion is always smaller than the submerged one.

A season is a system of equations, and I only look for its approximate solution. I have no ambition to solve it exactly. I have one requirement: my approximation must come with an error band, and that band must be published alongside the conclusion rather than buried at the end of the piece.

This is the greatest advantage of a sport with a long competitive calendar. Audiences have time to be misled repeatedly and to draw their own conclusions, provided they are given enough raw numbers. The writer's responsibility is not to take that opportunity away from them.

The contrarian angle: the blank page is the product

I want to say plainly what few people in this trade say.

Over the past three years, twice I received a request to produce a pre-tournament preview for a content platform. Both times, I opened my data file and found it empty in the columns that mattered most. Both times, I submitted a version with a list of everything I could not measure, and asked the platform to run that section at the top rather than the bottom.

The first time, they refused. The second time, they ran it, and the piece drew about thirty per cent fewer reads than average while earning more saves and shares. One reader commented that it was the first preview he had read in which nobody pretended to know everything. That comment was worth more to me than any engagement metric.

There is a temptation built into this profession, and I call it the temptation of the empty cell. When a spreadsheet has a hole, the hand supplies a plausible number. That number gets copied into another article, then another, and within two seasons it becomes something people cite as fact. Nobody in that chain deliberately lies. Nobody simply took the trouble to write N/A.

When the model collapsed, I started listening to the noise. And I discovered that the noise tends to live inside the very cells that were left blank. A player underperforming at one tournament does not mean decline. It may mean he is reducing load to peak for a bigger event without telling anyone. Decisions like that only surface months later, when you look at the whole sequence rather than a single match.

At the other end, Western analysis is correcting itself

I read a fair amount of foreign sports analysis, not to imitate its voice but to learn how it argues against itself.

The notable trend of recent years is not the addition of new metrics. The trend is the reduction of published metrics and the expansion of explanation about uncertainty. Large data platforms have begun printing confidence intervals instead of a single figure. They have begun stating openly that certain metric families are unreliable at small scale.

I take this as a sign of an industry maturing, and it is arriving late in our region. Vietnamese and Chinese sports media have enormous resources: large audiences, fast distribution, and extraordinarily active forums. But those resources tend to flow into producing conclusions rather than testing them.

I do not think the problem is with the fans. The problem is with the writers, myself included.

What I will track for the rest of this cycle

In my still-empty spreadsheet, four signals sit at the top of my watch list. I put them here as a public commitment, so that later I cannot claim I did not know.

The first is withdrawal density. If a top-twenty player withdraws from two consecutive events in the same geographic swing, that is usually a signal about the body rather than about flight schedules. Method: count withdrawals, note dates, note tournaments, cross-reference against the following three months.

The second is the average age of the four semi-finalists at Super 750 level and above. If that figure falls steadily across three consecutive events, the next generation is taking territory, and every assessment of veteran players needs rewriting.

The third is the number of three-game matches a player plays within seven days. This is a figure I can measure precisely, and it is the best indicator of accumulated injury risk I have ever found. Method: accumulate actual time on court, not match count.

The fourth is how often a player faces opponents from their own ranking group. If a player repeatedly meets weaker opponents in early rounds, their results are being inflated by the draw and their ranking is overstating their level.

None of these four signals requires coaching cameras. They require only a public schedule and someone willing to sit and count.

What I left at the bottom of the file

The next morning I reopened BWF_MS_Season_Review and scrolled to the bottom. Forty-seven rows, mostly empty cells. I changed nothing. I only added one new column on day twenty-one, and that column was called "date to re-measure".

This approach does not give me a fast article. It gives me something else: after every round, I hold one more documented gap, and at some point in the season that gap answers itself.

I have done this work for eleven years, and the only thing I have learned with certainty is that data needs time. People see the result; I see the state before the result takes shape. The distance between those two things is my entire job — and most of that distance is silence.

If next season you read a badminton analysis with no N/A anywhere in it, ask yourself: does the writer hold more data than I do, or is the writer simply more confident than I am?