The Data Gap in Vietnamese Badminton: What 21 Points Cannot Tell

Last Saturday night I opened the tracking file for the men's singles semi-final of the national badminton championship on my laptop. The "contact point" column was blank. The "rally length" column was blank. The "win rate after changing ends" column was blank too. The file contained fourteen player names, one score line, and nothing else.

At fifty, working as a data consultant for football clubs, I am used to dense files: PPDA, xG, kilometres covered, heat maps for every phase of play. Badminton, the sport that has accompanied my entire career, keeps handing me empty spreadsheets instead. Faced with an empty spreadsheet, the most honest answer a data person can give is four words long: not enough to conclude.

I entered the profession through journalism, but 2026 taught me that numbers can write too.

Context: a sport that refuses to keep records

In 2026, while working as an editor for a new football site in Ho Chi Minh City, I accepted an offer to work as a tactical data analyst for a club in Nha Trang. That season I used PPDA and xG to show something that contradicted the coaching staff's belief: the team only won when it held under 45 percent possession, and lost when it held more. The head coach wanted control football. My twenty-page report ended with one unsoftened sentence: keep going like this and the team will be relegated. The board listened. The team survived.

That experience taught me a principle I have carried for the eight years since: a decisive conclusion is only worth something when the sample is thick enough to carry it.

Badminton does not give me that thickness. Football has dozens of event-data providers, every pass logged by coordinates, every phase tagged. Badminton has three trustworthy sources: the Badminton World Federation's electronic scoring system, Hawk-Eye data at a handful of World Tour venues, the Super 1000 and Super 750 tiers, and the handwritten notebooks of people sitting courtside like me.

The bottleneck sits here: the Badminton World Federation publishes results by set and by point, but not rally sequences. A three-set men's singles match can contain 90 to 120 points, yet all we receive is the final point of each rally. We do not know how many strokes built it. We do not know whether it came from a proactive attack or an opponent's error. We do not know what that player changed after falling behind 11-16.

The rally-point rule, with 21 points per set, gives every point weight. But the weight of a point does not live in the number on the scoreboard. It lives in how the point was made. That is the largest gap in this sport, and it is also why so much badminton commentary drifts toward emotion: with nothing to cross-check against, people tell stories.

Four minimum indicators and one semi-final worth remembering

Over the past seven years I have built a minimum measurement framework for badminton, with four indicators collectable by eye and by notebook: average rally length per set, unforced error rate per set, points won on serve, and decider-set win rate.

The first is the one I trust most. Average rally length reveals almost the entire tactical intent of a player. When a men's singles player deliberately shortens rallies below seven strokes, it signals a plan built on speed and power: short serve, straight smash, finish early. When average rally length climbs above twelve strokes, it signals a plan built on endurance and patience.

Those two numbers lead to two entirely different approaches to managing physical reserves, and misjudging the rhythm by one beat can ruin the whole third set.

The second indicator, unforced error rate, is the most misunderstood. At domestic tournaments people tend to attribute unforced errors to nerves. But when I log them set by set, the pattern usually does not sit in the mind. It sits in the legs. Unforced error rate spikes between points 14 and 18 of the third set, when the ability to place the feet correctly has already degraded. That is a physical problem, expressed as a psychological one.

The third indicator, points won on serve, is almost always undervalued at national level. Serving in modern badminton is no longer a neutral opening stroke. With the below-waist service rule and tight short serves, the server carries more structural disadvantage than the receiver. If a player holds a serve win rate between 45 and 48 percent, that is normal. If it drops below 40 percent in a set, that player is losing points from the very first stroke, and no tactic rescues a start like that.

The fourth indicator, decider-set win rate, separates good players from champions. A player can win 70 percent of matches in a season and still lose 60 percent of third sets. That is the kind of data the scoreboard never shows you.

There is one match whose recording I still keep. A domestic men's singles semi-final, player A winning 21-19, 19-21, 21-18. Reading the scoreboard, it was a balanced contest. When I logged rally length set by set, the picture changed completely. First set: average rally of 8.4 strokes. Second set: 12.1 strokes. Third set: 9.2 strokes.

Player A won the first set with speed. Player B won the second with endurance. In the third, both returned to speed, but B's unforced error rate climbed from 14 percent to 27 percent. B did not lose because of nerves. B lost because he had spent all his energy in a race he should never have entered.

Without rally-length data, the story of that match becomes: "A was braver in the deciding set." With data, the story is: "B misread the rhythm of the match." Two stories lead to two completely different training programmes.

I used this framework to review the career of Nguyen Tien Minh, who once entered the world's top ten men's singles and competed at four Olympic Games. What stands out about him is not the peak but the length. More than twenty years at the top, across service-rule changes, across the scoring switch from 15-point service scoring to 21-point rally scoring, across changes of court surface and shuttle speed. Very few players survive three major rule transitions.

When I reviewed his later-career data, the indicator that fell most clearly was not points scored. It was rally length. He did not shorten matches to compensate for age. He held rally length steady and accepted a decline in quality on the final stroke. That was a tactical choice, and by my reading of the data it was the right one: the player who preserves match structure endures longer and stretches his career further.

In the next generation, Nguyen Thuy Linh carried Vietnamese women's badminton into the world's top 25 and competed at two Olympic Games. She plays the opposite template: shorter rallies, higher speed, heavy dependence on tempo. That template offers a higher ceiling but also higher variance. In other words, the same player, with a wider range of outcomes.

Le Duc Phat, who competed at the Paris 2026 Olympics, represents a third template: men's singles built on endurance and resilient defence, with three-set matches in which decider-set win rate swings from season to season.

Three players, three templates, three different ways of reading data. And here I have to state plainly what I am often criticised for avoiding: with three individual profiles, we do not have a statistical sample. We have three stories.

In badminton, the error in the data is larger than the error in the stroke. A smash twenty centimetres off line can still be a winner. A dataset missing twenty centimetres of information is simply a wrong dataset.

The worry is not the missing data

What worries me most about Vietnamese badminton right now sits somewhere else: the replacement of data with narrative.

Last season I watched a young player win a domestic title after beating three opponents. The media immediately called him the future of Vietnamese men's singles. I took out my notebook and counted: three matches, nine sets, roughly 340 points in total. None of those three opponents sat inside the world's top 200. The sample was nine sets. Nine sets is not enough to conclude anything about a career, not even enough to conclude anything about a season.

That is a trap someone in my profession is just as likely to fall into: using one match to indict an entire system. The transfer market and the reputation market run on the same rule: real value lies in the question, not in the answer.

There is a second trap, more dangerous. When data is scarce, analysts tend to select the indicators that fit their argument and ignore the rest. I fell into this once, using a run of matches with PPDA below 5 to conclude that high pressing would collapse between the 70th and 80th minutes. That conclusion was correct for the dataset I had, but the dataset I had covered only Asian teams in one specific period. I spoke of a rule when I had only a sample.

Badminton faces exactly that trap at a smaller scale. A player's serve win rate dips across two consecutive matches and immediately there are articles about a "service crisis". Two matches, in a sport where each match holds only about 80 to 120 points, prove nothing at all.

Meanwhile a larger subject is being hidden by public data: the calendar.

The current World Tour comprises more than thirty events a year, running from Super 100 to Super 1000, plus the World Championships and team events. That density is not a technical decision. It is a commercial one. And the way it is justified is well known: load management.

I do not deny that load management is a real concept. But when the calendar expands because of broadcast rights, and players are then asked to "manage load" by withdrawing mid-season, load management stops being a scientific method. It becomes a scientific label stuck onto a commercial schedule. What happens is identical either way: the player withdraws, the ticket-buying fan loses money, and the ranking system keeps running as if nothing occurred.

The same logic appears elsewhere. In emerging badminton nations, young players are pushed onto the international stage at 17 or 18, before they have a physical base and before they have their own medical staff. Knee and shoulder injuries then surface between the ages of 21 and 24, precisely when the peak should arrive. We call that "international tuition".

Once again: the data is not innocent. It simply is not published well enough for anyone to see.

Signals to track next season

With the transfer window and the new season calendar, these are the signals I will follow, not to predict who wins, but to check whether Vietnamese badminton data has started to say more.

First, the distribution of rally length by set at national level. If that data is logged consistently across a season, we can tell which players win through match structure and which win through luck.

Second, decider-set win rate per player rather than match win rate. These two indicators often tell different stories, and the second is the decisive one.

Third, withdrawal and walkover rates against calendar density. If that correlation becomes clear within a season, it stops being an individual player's problem and becomes a structural problem of the system.

Fourth, the existence of shot-level data at Asian tournaments. When stroke-by-stroke data arrives, the badminton analytics trade in Vietnam will change shape. And I will have to learn everything again, like a newcomer to the profession.

Every match is a tea session for the data monk: silent, and absorbed slowly.

Numbers are never in a hurry. We are the ones who rush.