Re-reading the Badminton Data Sheet: When Rally Length Tells the Story Speed Cannot

Re-reading the Badminton Data Sheet: When Rally Length Tells the Story Speed Cannot On the night of the Paris 2026 Olympic men's singles final, Viktor Axelsen won 21-11, 21-11 against Kunlavut Vitidsarn. In a cafe near the sports district of Shanghai, most people around me nodded: as predicted, the strongest player won. I reopened the rally map I had built through the tournament. What caught the eye of someone who works with data was not the scoreline. It was that the bulk of the points Axelsen won came from long rallies, where stamina and patience start to overpower pure technique. Fans remember the smashes. I remember the rallies nobody counts. Badminton carries one of the densest data sets in racket sports. Every rally can be measured: shuttle speed off the racket, rally length, distance covered, unforced-error rate, net-point win rate, rear-court win rate. A top men's singles match runs about 50 to 70 minutes, covering 150 to 200 rallies, equal to several thousand raw data points. Yet public conversation about badminton revolves almost entirely around a single variable: smash speed. I have followed international badminton for more than thirty years, from an era when the speed gun was a decorative device beside the court. Back then scoreboards were kept by hand, and the memory of a match depended on who remembered best. Today every BWF World Tour event has automated point recording, but the irony is that we still use it the old way: to highlight the most shocking number, not to understand why a player wins. This piece is a re-reading of the data sheet. When the whole world shouts, I read the numbers again. Not to deny the emotion of the stands, but to check what the data actually says behind what the eye can catch. The context matters. Professional badminton runs on a packed calendar: the BWF World Tour spans many tiers from Super 1000 down to Super 300, plus team events like the Thomas Cup, Uber Cup and Sudirman Cup, and every four years the Olympics. A top-10 player can play more than twenty events a year, each lasting about a week, under constant ranking pressure. This is an ideal environment for data analysis, because the number of observations is large enough to separate signal from noise. But that same density means most media content only has time to chase results, not to analyze process. I was born in Vietnam, work in the heart of China, and report on badminton for a market that is not short of fans but short of the habit of reading data systematically. Here people remember clearly who won, but rarely why. That is the gap a data worker can fill. The first illusion to dismantle is the smash-speed story. In 2026, Denmark's Mads Pieler Kolding was recorded with a smash reaching around 493 km/h, a number repeated worldwide as proof of modern badminton's power. Earlier, Malaysia's Tan Boon Heong was also recorded at a similar level. These numbers are attractive, easy to spread, easy to turn into headlines. But they have a basic problem: the speed measured as the shuttle leaves the racket is not the speed when it reaches the opponent's side, and still less a measure of winning the point. A 400 km/h smash hit straight at an opponent who is waiting becomes a lost point. A 280 km/h smash placed into the far corner, forcing the opponent to cover the full width and lift a weak return, creates a winning point two rallies later. The data shows this clearly: in top matches, the rate of points won directly by smash does not correlate strongly with match outcome. The stronger variable is the opponent's unforced-error rate, and it is often generated indirectly by shots that are not fast at all but well placed. I must tell a professional story here. In 2026 I once confidently presented N'Golo Kante's pressing numbers on a new livestream platform, with figures like 12.4 km covered per match. The audience did not understand, and the commentator cut in to switch to which player dressed well. After that night I realized something that applies fully to badminton: raw data does not speak for itself. To mean anything, it must be placed inside a concrete story on court. Since then, every piece I write opens with a situation, a player, a moment, and only then reveals the number behind it. Notably, the top players themselves understand the gap between speed and efficiency. Looking back at Lin Dan's career, with Olympic gold in 2026 and 2026 plus five world titles, a striking pattern appears. Lin Dan had a powerful smash, but what made him a legend was timing. He did not smash the most. He smashed at the right moments. His decisive variable was the conversion rate from advantage to winning point, a metric that media at the time had no name for. Lee Chong Wei of Malaysia is the opposite case and equally thought-provoking. He held world No. 1 for a long stretch, with movement speed and consistency rare in any era, yet never won Olympic gold or a world title. If you look only at power metrics, his missing major title is a paradox. If you look at the data of big finals, another pattern appears: at decisive moments, a few percentage points of efficiency slipped away, and in elite sport a few percentage points decide everything. Data quantifies the match, but cannot quantify the heart of the fan, and sometimes cannot quantify the psychological weight on the player either. Now to the most important but least discussed layer: rally length. The length of each rally is the indicator that most clearly reflects a player's style and stamina. In modern men's singles, average rally length in top matches usually falls around 7 to 12 shots, but the distribution is very skewed: most points end early, while a few rallies stretch to 30, 40, even more than 50 shots. These long rallies consume disproportionate stamina and often decide the course of a whole game. Japan's Kento Momota is a textbook case of using rally length as a weapon. At his peak he won the world title in 2026 and 2026 with a game not based on the highest smash speed but on the ability to extend rallies and minimize unforced errors. Data showed his non-mandatory error rate among the lowest in the field, while his distance covered per game was among the highest. He won by making opponents run more, endure longer, and finally err first. Axelsen at Paris 2026 followed a similar logic with optimized physical tools. He still owns one of the hardest smashes in the game, but the way he won gold was not by smashing more. In the final he controlled tempo, pushed Kunlavut into long rallies, and forced the young opponent to keep making decisions while tired. This is a tactic based on data about the opponent's stamina, even if the player himself might not call it that. Tactics are not on the diagram, they are in the way data arranges itself. When I plot a player's rally-length distribution game by game, I often see a rule: the games they win have a higher average rally length than the games they lose. This holds for most control-style players, and reverses for some pure-attack players. Each pattern tells its own story about how that player builds points, and that story never appears in a summary report. The next data layer is the net game. In modern badminton, the net is the area that decides match tempo. A player who controls the net controls the match, because from the net they can force the opponent to lift and create attacking chances. Net-point win rate is a metric I track closely, and it usually predicts match outcome better than smash count. In women's singles this is even clearer. South Korea's An Se-young, who won the Paris 2026 women's singles gold, embodies a model based on stability and control. She is not the hardest smasher, but owns movement, defense and placement among the best. Her data shows a low unforced-error rate, fast transition from defense to counter, and above all patience in long rallies. In an era when media seeks spectacular shots, she wins with what is hardest to turn into a clip. Her top rival, Japan's Akane Yamaguchi, twice world champion in 2026 and 2026, follows a similar model: enduring stamina, resilient defense, few errors. Spain's Carolina Marin, Rio 2026 Olympic gold medallist and three-time world champion, represents a powerful attacking model based on speed and relentless pressure. Both models can reach the top, but data shows the control model has higher stability over time, while the attacking model has a wider range of fluctuation. Taiwan's Tai Tzu-ying, former world No. 1 and Tokyo 2026 silver medallist, is a special case with an unpredictable style. She constantly uses deceptive shots and sudden tempo changes, making her data harder to analyze than almost any other player. But that unpredictability is itself a tactic: preventing opponents from preparing for a fixed pattern. Here a question of culture and system must be raised. Chinese badminton once dominated the world for decades thanks to a centralized training system, high discipline and large resources. But in the data era, systemic advantage is no longer enough if paired with old habits. I observe that big training centers increasingly emphasize opponent analysis, but the speed of adopting new tools varies greatly between groups. Some use data to change training plans; others still treat it as a side part of the session. Shi Yuqi, China's leading men's singles player, is a notable case. He owns a comprehensive skill set and once reached world No. 2, but his journey shows that the gap between data potential and actual results depends heavily on stability, both physical and mental. This is the zone where old data tells the story of a dead era, while new data needs time to accumulate enough sample. Other standout players of this era include Malaysia's Lee Zii Jia, Indonesia's Anthony Ginting and Jonatan Christie, and Taiwan's Chou Tien-chen. Each represents a different data model, and comparing them with a single metric always leads to a wrong conclusion. That is why I do not trust emotion, I trust the time series. One match can deceive, but a series of twenty matches is much harder to deceive. In doubles, the data picture is even more interesting, because two people and their coordination create a variable that cannot be reduced to individuals. India's men's doubles pair Satwiksairaj Rankireddy and Chirag Shetty is an example of a badminton nation rising outside traditional powers, based on build, speed and chemistry. In doubles, the first-serve-point win rate, the conversion rate after a return, and the average positional distance between the two players matter more than individual smash speed. Another data layer I consider undervalued: workload and injury. With a packed calendar and Olympic ranking demands, top players must compete continuously to hold their ranking, and their bodies carry heavy loads. Tendon, ankle, knee and back injuries are characteristic of the sport. Injury data shows a significant share of long absences relate to playing too much in a short period, rather than a single on-court incident. This is where I see a parallel with team sports. In football, loans with obligations to buy tend to benefit big clubs and create risk for small ones. In badminton, a similar structure exists in the form of a ranking system that forces players to race for points, in which players and their teams bear most of the physical risk while commercial benefit is distributed back to the system's center. Analyzing workload data is not just the story of one player, but of an entire structure. Alongside this, the training and talent-transfer market in badminton also operates on a logic that must be read through numbers. Academies often overvalue young players' potential based on a few standout technical metrics, while undervaluing hard-to-measure factors like discipline, pressure tolerance and team integration. Every investment in a young player is a gamble, but the win rate is in the spreadsheet, not in the gut. Now for the counter-intuitive part I want to spend time on. The common belief is that modern badminton is a game of speed, that whoever smashes harder and moves faster wins. Data at the elite level says the opposite in an important sense: speed is a necessary condition to compete, but not the factor that separates winners from losers within the group of players who are already fast enough. Imagine two players both averaging a 380 km/h smash. One smashes into mid-court; the other smashes into the corner and moves to the net to meet a weak return. The data sheet credits both with the same speed, but their point-win rates differ clearly. The metric that separates them is not speed, but the rate of sustained pressure, a metric visible only when you look at the next rally, not the first shot. The meta changes weekly, but the rule stands outside time. Measuring tools may be new, player patterns may shift, but the basic structure of scoring in badminton, controlling the net, controlling tempo and reducing errors, has stayed the same for decades. I do not treat old data as useless; I treat it as a mirror. Old data is not wrong, it just tells the story of a dead era, and the analyst's job is to place it in the right context. Another trap to avoid is mistaking correlation for causation. A player with a high win rate usually has many pretty metrics, but that does not mean every pretty metric contributes equally to victory. Some metrics merely accompany success rather than create it. Correct analysis requires checking at least two independent metrics that confirm each other, and always asking whether the relationship holds when the sample is widened. The 2026 pandemic taught me this lesson unforgettably. When global tournaments stopped, every prediction model based on historical data became useless overnight. I tried to gather data from online training sessions of a Shanghai club, but received only a few data points a week, not enough to run a model. For the first time I admitted data is not an omnipotent god. Since then I add a section at the end of each analysis: the limits a model cannot cover, such as psychology, weather and luck. In badminton those three appear more often than anyone wants to admit. So what comes next? For badminton, the next data layer will come from sensors in the racket and real-time motion tracking. When data on impact force, shot angle and position is collected automatically in every event, we will be able to compare players across eras with common measures. Then the gap between legends and contemporaries will be measured in numbers, not just in memory. But I also believe there will always be data we cannot measure. I have witnessed matches where every metric tilted one way, yet the result went the other. Sport contains a dark zone the spreadsheet never touches. That is why I write rather than only run models: to remind those who read the data sheet that behind every number is a human under pressure in a moment that cannot be repeated. Back to the Paris final night. The 21-11, 21-11 scoreline tells a story of dominance. The rally map tells another story, of rallies stretched, mistakes seeded from the seventh shot, of a champion winning with patience rather than only power. Fans buy tickets to see the shots. A data worker like me stays behind to count the rallies that follow. I will keep following the next events on the BWF World Tour with a specific question: will the tempo-control trend keep overtaking the pure-attack trend, or will a young generation find a new way to turn speed into lasting advantage. The answer will not come from one match, but from a time series long enough to separate signal from noise. And when that signal appears, I will open the spreadsheet again and re-read it from the start, as I have done for thirty years.