The Blank Between Two Points: Professional Badminton Has No Data to Tell Its Own Match

It is 11-9 in the deciding game. The umpire signals the sixty-second interval. The player sets the racket down on the Istora floor, drapes a towel over his head, and the coach leans in with a hand covering his mouth. The broadcast camera holds tight on the face and captures not a single syllable. The stands are still roaring, but on court there is a silence that was designed in advance.

In Surabaya it is three in the morning. I am sitting in front of a tracking sheet with forty-one columns. Thirty-nine of them are empty. There is no count of steps taken in the last sixty seconds, no shuttle speed on the final three rallies, no breathing rate, no recovery time, not one line about how much pace the player has lost since the cross-court smash a minute ago. The official scoreboard is still glowing: 11-9. That is everything the rest of the world knows about this match.

Those sixty seconds are unclaimed territory. When every tournament stops, I finally hear my own heartbeat.

A report that came back as N/A

Earlier this month I was asked to prepare a pre-match analytical report for a men's singles quarter-final at Super 1000 level. I opened the official data platform, downloaded the match statistics, and received exactly what I have received for nine years in this trade: two names, a head-to-head record, game scores, match duration, nationalities, seeding. That is all. I typed N/A into the notes field and sat still for a while. That word says nothing about the players. It says something about the system that produced it.

In football, a second-tier English league match can generate more than three thousand data events: every pass, every touch, coordinates, direction, pressure, expected goals attached to each shot. The sport I cover in depth, followed by hundreds of millions of people from Jakarta to Copenhagen to Kuala Lumpur, publishes a few dozen fields per match. Most of them can be read off the screen by anyone with eyes.

The World Badminton Federation's official data platform gives users the basic fields: results, scores, duration, round, head-to-head history, ranking and ranking points. At some major events the statistics page adds a handful of extra boxes such as points won by smash, net winners, and longest rally. That is the ceiling of public data.

The BWF World Tour is tiered into Super 1000, Super 750, Super 500, Super 300 and Super 100, plus the World Tour Finals. The top tier in recent cycles has included the All England, Malaysia Open, Indonesia Open and China Open. The Indonesia Open, played at Istora Senayan, usually carries total prize money around 1.3 million US dollars and is one of the loudest arenas the sport has. Badminton in Indonesia is not a sport so much as a piece of national history: Susi Susanti and Alan Budikusuma delivered the country's first two Olympic golds at Barcelona 2026, and every Games since has been measured by one question, whether badminton brought gold.

I walk into the cathedral of data to hear the noise, not to pray.

The asymmetry sits right there. A country can pack tens of thousands of people into an arena for a men's singles match and still have no way of knowing how many metres that player covered in the third game, how much his average smash speed dropped across the last ten points, or what his win rate is when he trails at the interval. That data exists. It is simply not handed to the audience.

What Hawk-Eye sees, and who gets to see it

Since 2026, the Instant Review System built on Hawk-Eye technology has been a familiar part of major events. Each player has two challenges per match, retained if the challenge is upheld. When the shuttle lands close to the line, the umpire looks up at the big screen, a simulation of the shuttle landing appears, and the crowd reacts.

What matters here is that Hawk-Eye does not lack data. The system literally reconstructs the shuttle's three-dimensional trajectory, computes speed at the moment of contact with the racket, models the flight path, and determines the landing point to a margin of a few millimetres. All of that volume exists inside the machine. What gets published is one bit of information: in or out.

Based on my experience following these matches, this is the biggest blind spot in professional badminton analytics. In other sports using comparable technology, the surplus data has been mined for broadcast and for fans. Here the surplus is kept closed. Nobody outside the tournament office is allowed to touch it.

The hardest rule to measure is the least measured

In 2026 the world federation introduced a fixed-height service rule: the contact point between racket and shuttle may not be above 1.15 metres from the court surface. The rule is technically sound in spirit, removing the advantage of very tall players and eliminating an overly oppressive style of serving.

The human eye cannot measure 1.15 metres inside a movement lasting a few hundredths of a second. Service judges work on instinct, and most tournaments do not use technology to verify service height. The clearest technical rule in the rulebook is therefore the one with the weakest enforcement accuracy.

The Instant Review System does not make controversy disappear. It moves controversy off the court and into the review room and into the grey zones of the law. The old argument about a shuttle landing near the line is replaced by a new one: which frame counts as the moment of contact, how much the feathers deformed, how many millimetres wide the line is. The same footage, two sides reading out two different truths.

A hand-coded diary

Since no data is available, I do what I have done since eleventh grade: I sit and count. I hand-code every rally from video, type it into a spreadsheet, and log everything loggable. My current database holds 2,412 fully coded rallies from 200 matches between 2026 and 2026, plus roughly 4,800 more added in later seasons. Each rally is recorded across eleven variables: rally length, serving side, who attacked first, the finishing stroke, error type, error location on court, whether the net cord was touched, position within the game, point gap, and whether it happened before or after the interval.

Three findings, with the confidence levels I am willing to claim.

One: men's singles rally length is bimodal, not a single curve. Across the Super 750 and Super 1000 matches I coded, rally lengths cluster into two groups: four to seven shots, and thirteen to twenty-two shots. The arithmetic mean sits at about 9.4 shots. Precisely for that reason, the average becomes meaningless: it describes a match that never happened. On one court, in one match, two different sports are underway, and the deciding factor is whether the serving side can claim the attack on the third shot.

Two: the sixty-second interval does not create the momentum that broadcast narratives describe. Across 318 coded intervals, a player trailing at 11 won the next point 48.2 percent of the time, roughly a coin flip. But the unforced error rate across the three points after the interval was 2.6 percentage points higher than across the three points before it, and the increase showed up on both sides. That is the signature of fatigue, not inspiration. The numbers suggest it rather than prove it; 318 observations is a small sample, and I drew it from matches I could actually watch, not from a random sample.

Three: the net cord is noise, not signal. Forty-three of 2,412 rallies saw the shuttle clip the tape and fall on the far side, or 1.78 percent. I split that group by point gap, by game, and by interval state, and found no pattern strong enough to clear the noise threshold. A net cord is rarely destiny; it is usually the tiny deviation between expectation and probability. But because it happens at the most dramatic moment, a spectator's memory will weight it many times heavier than its true weight.

In men's doubles the picture flips. In my sample, 61 percent of rallies ended within the first six shots. Most modern men's doubles matches are decided by a very short sequence: serve, return of serve, third shot, sometimes fourth. I call the third shot the most undervalued variable in modern badminton: it appears in almost every decisive rally, and no official statistical column exists for it.

On smash speed, the metric broadcasters love to put on a graphic, public data amounts to a few scattered records. The 493 km/h mark recorded in 2026 by Tan Boon Heong was set under laboratory conditions with a completely different measurement method from match play. The highest recorded in genuine competitive conditions is around 426 km/h. Those two figures measure different things, and the fact that they routinely appear side by side in articles is a clean example of data being pulled away from its method.

Three tiers of data, three tiers of audience

A Super 1000 quarter-final has eight to twelve cameras, an Instant Review System, on-screen smash speed graphics, and courtside data operators. A first-round Super 300 match in the same country may have no review system, no speed graphics, no commentary, and sometimes no live stream at all.

At the top tier, players are tracked by technology. Below it, players are tracked only by human eyes. And most of a young player's career, particularly one from a developing badminton nation, unfolds at that lower tier. The people who most need data to convince a national coaching staff, to secure sponsorship, to prove they are improving, are the least likely to have access to it. This is a tiered data system, built parallel to the tiered tournament system, not by coincidence.

As a working analyst I have to admit my sample is skewed for exactly this reason. I can code the matches I can watch. I can watch the matches with good video. Good video belongs to the big events. My database, and that of nearly every independent analyst, is a mirror of the sport's power structure rather than a map of the sport.

Where the data stream flows

At every event with live data, someone sits courtside and keys in each stroke on a device. Where does that stream go, and who pays for it?

A small part flows into the statistics graphics shown on broadcast. The bulk flows into live sports data platforms, and the biggest customer, the fastest payer, the one demanding the lowest latency, is the betting market. The heartbeat of the sport, converted into data, flows to a place the fans packed into Istora will never set foot in.

I have worked on the other side of that stream, and I know exactly how it operates. An analyst sitting in Surabaya can read the smash speed of a player in Copenhagen before Indonesian fans learn it from a commentator. That latency gap is not an incidental detail. It is the product.

Alongside it sits tournament integrity monitoring, which exists to detect unusual betting activity. A player is watched by the very data generated from her own sweat, through a system she has no access to. I do not write this to indict anyone. I write it as a structural feature: when sports data is produced, it flows toward money, and that order is set long before the player steps on court.

The white space is not a technical fault

What I want to push back on is the default reflex of the analytics industry: the assumption that the data gap is a fault to be fixed, a backwardness to be corrected, a sign that this sport has not grown up.

That white space operates as a business model. Scarcity creates value. If everyone had shuttle trajectory, the only thing left to sell would be the speed of distribution, and the highest bidder for speed is the party with a betting incentive. The current structure keeps most of the data with a small group, and that group has no incentive to open it. The hole is not in the source code; it is in the eyes of the person reading the source code.

My second objection targets the habits of people in my own trade. Adding data does not make controversy disappear; it relocates controversy to a layer where viewers no longer have the tools to check for themselves. When I made a major public prediction and got it wrong, and then spent sixty hours reviewing footage to write a self-critical piece, the lesson was not that I lacked data. I had the most attacking data in the field. I had asked the wrong question. I asked who scored most instead of who controlled the most dangerous space.

I repeated a smaller version of that mistake in badminton. One season I built a knockout-stage prediction model for men's singles on average smash speed, because that was the only metric with numbers attached. The model failed. The eventual champion sat mid-table on average smash speed. The deciding variables, once I rewatched the tape, were the quality of the third shot and the ability to retain the attack across two exchanges, none of which appear in any official data table.

The third objection concerns collective memory, and it worries me more than the other two. Indonesia entered an Olympic cycle expecting a badminton gold and left without one, something that had never happened since the sport joined the programme. The story told immediately afterwards was a story of decline. The underlying data I hold does not support that telling. In the same cycle one Indonesian player won the All England and another reached the final, producing the first all-Indonesian men's singles final at that tournament since 2026. An Indonesian woman won an Olympic bronze. The foundational structure did not collapse. A probability distribution simply landed its outcome in a different branch. When crowds read a result as destiny, they are reading a sample of size one.

Signals for the next cycle

Five signals I will be tracking.

First, whether the world federation publishes rally-level data for at least one top-tier event. If that happens, the entire independent analytics sector reshapes within eighteen months.

Second, whether the surplus data from the Instant Review System is licensed for broadcast and research use. The shuttle trajectory has already been recorded. It is a question of a signature on a contract.

Third, the review process for the fixed-height service rule. This rule has the widest gap between text and enforceability, and any change here reshapes the scoring structure of an entire generation of players.

Fourth, tracking data at the talent-development level, where national federations are starting to build their own measurement systems. Whoever owns the data of a fifteen-year-old decides where that player goes at nineteen.

Fifth, the growth rate of the live data market. If money spent on badminton data keeps rising faster than the pace of public data release, the distance between insiders and audiences widens by another notch.

I am not hoping this sport gets more data so it becomes easier to predict. I am hoping it gets more data so fans understand more precisely the price a player pays for every point. The more precise the metric, the wider the distance between the human being and the match, and the writer's job inside that distance is to pull the human being closer, not to push the spreadsheet further away.

Tonight another match will cross 11. I will sit and count through sixty seconds of silence, with thirty-nine empty columns, and ask whether my readers need to know anything beyond the score. After nine years, my answer is still yes. That is why I am still here, typing every stroke into a spreadsheet nobody pays me to maintain.