Why Vietnamese Badminton Is Harder to Analyse with Data Than Football

On the last Saturday of September, I sat in front of a screen in my flat in Nha Trang, rewatching a women's singles semi-final at an international badminton tournament that featured a Vietnamese player. The match ended close to eleven at night. I opened the tournament information page looking for something very basic: the average length of a rally. Nothing. The page returned the score, the match duration, and a single line reading "fastest smash". I closed the tab, opened the stopwatch on my phone, and counted by hand.

Forty minutes later I had 96 rallies. An average of 11.4 seconds per rally. But broken down by game, everything flipped: game one averaged 8.1 seconds, game three climbed to 15.7. The player who won game one ended rallies early; the player who won the match won by extending them. The tournament page could not tell those two players apart. It only said who won.

People remember the smash that won the point. I remember the nine shots before it.

Badminton has cameras, but no data warehouse

This is the paradox I have run into across seven years in this job. Badminton owns enviable technical infrastructure: Hawk-Eye at events on the BWF World Tour, shuttle-speed sensors, high-speed cameras used to determine whether a shuttle landed in or out. A single semi-final can generate tens of thousands of raw data points about shuttle position, foot position, movement direction.

Almost none of it is opened to the public.

Football is ahead here. A single English Premier League match yields more than 3,000 labelled events, from passes and duels to the coordinates of every shot. From that people built xG, PPDA, progressive passes — terms now used even by people who do not watch football. Badminton has none of it. The scorecard of a professional badminton match, even at world level, usually stops at three columns: score, duration, number of service faults.

With those three columns, I cannot answer the simplest thing any spectator wants to know after a match: what actually decided the result?

I entered this trade through an Excel spreadsheet, but I stayed for the stories inside it. And most of Vietnamese badminton's stories have never been recorded.

From missing data to building my own measures

In 2026, when tournaments resumed inside empty arenas, I started noting by hand. For every match I watched, I built a table with seven columns: rally length, serving side, final shot, point winner, number of direction changes, estimated distance covered, and the moment within the game. Seven columns, written by hand, while my eyes still had to follow the shuttle.

After roughly 40 matches — more than 3,800 rallies — I had a metric I called the long-rally win rate, calculated as points won in rallies lasting more than 15 seconds divided by the total number of long rallies that player contested.

The metric produced a result that made me recheck my notes three times. For some players whose attacking style the media described as "dominant", the long-rally win rate was markedly lower than their short-rally win rate. Conversely, some players labelled "defensive, lacking penetration" posted the highest long-rally win rate in the sample. The style label fans attach to them holds for the first 10 seconds of a rally, and fails after 15.

I also tested a second measure, the forced-error share. Today's scorecards lump every player mistake into a single "unforced error" column, regardless of whether the error came from a self-inflicted miss or from a rally where the opponent pushed them past the point of any return. Those are different in nature: one is a technical fault, the other a consequence of the opponent's tactics. Merging them erases the credit of whoever created the pressure.

Once separated, I found that in many matches the losing player recorded fewer self-inflicted errors than the winner. They lost because they were forced, not because they collapsed. An entire tactical story sits in that gap, and the traditional scorecard does not tell it.

Nguyen Tien Minh and the lesson of a decade nobody measured

Take the biggest case in Vietnamese badminton. According to ranking data from the Badminton World Federation (BWF), Nguyen Tien Minh reached the world top five in September 2026 — the highest mark any Vietnamese men's singles player has achieved to date. He won the Vietnam Open multiple times and was a familiar face at several Olympic Games.

But ask one simple thing: during his peak years of 2026 to 2026, what was Tien Minh's win rate in rallies lasting more than 20 seconds? Nobody can answer. Not because the question is hard, but because nobody recorded it.

For a whole decade we had a world-class player and not a single rally-level dataset to retell how he won. All that remains is memory, and memory is always worn down by time into legend.

The story does not stop with Tien Minh. In the current generation, Nguyen Thuy Linh is a regular presence in the deep rounds of World Tour events, while Le Duc Phat is the leading figure in men's singles. When I rewatch Thuy Linh's matches, I want to test a hypothesis: whether her results come from extending rallies or from sudden acceleration. No public data allows the test. I end up counting by hand again.

Every season is a lifetime of practice; every margin of error is a session of meditation.

The heatmap has become the new fortune-telling

This may sound like I am asking for too much. But football, where data is already abundant, has bred a different disease: fortune-telling with heatmaps.

A heatmap shows a player covering a lot of ground on the left flank. It does not say whether he ran there because he was instructed to, because he lost his position, or because a teammate abandoned theirs. The map says where, not why. Then someone attaches a story to it: "he dominates the left side". Three months later the coach changes the shape, the player still runs there, and the old story collapses.

I guard against exactly that trap in badminton. Because badminton lacks data, analysts in Vietnam fall easily into two extremes: either narrating events, or inventing a metric and then worshipping it like a talisman.

I nearly fell into the second. When I first built the "long-rally win rate", I was eager enough to nearly draw conclusions about an entire generation of players from 40 matches. Luckily I stopped to ask myself: of those 40 matches, how many were against opponents of the same level? How many took place in arenas with drift? How many involved a player carrying an injury?

Data does not lie; it stays silent until you learn how to listen. But it is also silent about the things you forgot to ask.

What fills the void

When data is absent, what fills the gap is not neutrality. It is sentiment, rumour, and sometimes betting lines built on a handful of recent matches.

In the days before a major event, I receive messages asking about a player's "form". The word sounds like data, but on follow-up it usually means: people remember the last two matches. The three before those, nobody remembers. The niggling injury from three weeks ago, nobody remembers. The fact that the player just changed their serve, nobody remembers.

I once got a message from a young U.19 coach asking how to measure the distance between lines during a defensive rally. He had a camera, software, and players. What he lacked was a recording system long enough to show which change actually worked. A coach cannot evaluate a new drill when all he holds is a feeling.

Across the region, a few federations with long badminton traditions are said to keep internal data for their national teams, though most do not publish it. The gap between them and us is not in the players. It is in the decision to record.

I have nothing against the emotions of fans. Emotion is also a form of data, just uncoded data. But when emotion is used to replace evidence, it becomes something else: belief that cannot be verified.

Reading a match through what disappears

There is a direction of reading I find more useful than chasing numbers that do not yet exist: reading a match through what disappears.

When a player suddenly stops smashing in game three, that is data. When a coach stops asking a pupil to come to the net, that is data. When applause in the arena fades after each long rally, that too is data — about an entire stadium tiring on the player's behalf.

In the summer of 2026, with tournaments held in empty arenas, I measured something that normally cannot be measured: the gap between rallies. Normally the crowd fills the interval between points, so we never sense how long a player takes to recover. In an empty hall that interval is exposed in full. Winning players tended to take intervals roughly 1.5 seconds shorter than losing players late in game three. A difference so small it is meaningless on a scorecard, yet a clear signal about conditioning.

Numbers never tell the whole story, but they know where the story begins.

Correlation is not causation — and that is good news

I have to say plainly something sports-data people are often reluctant to say: correlation is not causation, and in badminton the distance between the two is wider than in football.

A player with a high long-rally win rate is not necessarily good at long rallies. They may simply choose to extend rallies they already control, and to end unfavourable ones early. The metric reflects tactical choice, not physical capacity. Separating the two requires data on the score state of every rally — which my handwritten table does not have.

That is why I never publish a metric without its calculation method and sample size. A measure without a stated sample is just a story dressed up in numbers.

And that is also why I see good news in badminton's data shortage: we have not yet built bad metrics only to worship them for twenty years.

Football has walked that road already. A whole generation of fans believed that the team with more possession was the better team, until the 2026 World Cup taught us that possession is an illusion dressed up. Badminton has a chance to learn that lesson before paying for it.

The signal for the next cycle

The thing I am watching is not a particular player, but an infrastructure change: whether domestic badminton tournaments start recording rally-level data.

The first sign will appear at small events, where a group of volunteers sits noting by hand the way I once did. If within the next two seasons a national championship produces a rally-level dataset clean enough to publish, we will have something Vietnamese badminton has never had: a way to verify memory.

When memory can be verified, legends will have to stand on their own merits, rather than on the silence of data.