Rally Rhythm and the Third Game: Where Badminton Data Stays Silent

In May 2026, the annual general meeting of the Badminton World Federation voted on a proposal to change the competition format: five games to 11 points. The proposal fell short of the required two-thirds majority and was rejected. The three-game, 21-point format introduced in February 2026 remains in force to this day.

What I wrote in my notebook that day was not the vote count. What I copied down were the two sets of documents presented to the assembly. The supporters showed a chart of average match duration in men's singles. The opponents showed a chart of the number of rallies in a match. The two tables sat side by side, both correct, and both failing to answer the question audiences actually ask: what determines how long a badminton match lasts.

Numbers are not wrong. I simply forgot to ask where each number was standing. A figure removed from its context is nothing but a lie that has been polished.

Context: when the system changes, the breathing changes

Before February 2026, elite badminton was played under a 15-point system in which points could only be won while holding serve. That system produced a very particular kind of match: long rallies, interspersed with service exchanges during which the scoreboard did not move. When the Badminton World Federation switched to 21-point rally scoring, every rally became worth the same. That was the biggest change to this sport in half a century, and it had nothing to do with duration.

Under the old system, a missed shot could be a deliberate way of handing back serve while waiting for another opening. Under the new system, that missed shot is a point for the opponent, with no way back. Risk became more expensive, and the way players accept risk changed with it. This is the contextual layer most format debates skip, because it does not show up on a duration chart.

I have followed badminton since 2026, through radio broadcasts, through studio floors, through men's and women's team events and major multi-sport Games. My profession is reading data, but my habit is to stay behind after the stands have gone dark. Only when the arena is empty do I hear the background data clearly.

How I measure rally rhythm

My logbook is nothing modern. For every match I record four things: the duration of each rally in seconds, the number of court transitions from the back court to the net and back, shuttle speed at three points in the match, and the rest gap between rallies.

From those four columns I build three indices. The rally-cadence index is the distribution of rally durations within a game. The transition index measures the density of movement along the length of the court. The volatility coefficient measures how far a match can be turned around by a short run of points.

I built the volatility coefficient after 2026, when a prediction model based on raw data embarrassed me in front of colleagues and in front of myself. The lesson that year was not to abandon models. It was that the mistake lies not in trusting a model, but in never asking what the model left out.

The chain of evidence: the third game is where the truth shows itself

Read only the scoreline, and the third game of a men's singles match looks like a compressed first game. My logbook says otherwise.

In the elite men's singles matches I have logged, the first game usually has the longest rallies. The second game has the highest shuttle speeds. The third game has the fewest court transitions, the longest rest gaps between rallies, and the highest rate of self-inflicted errors.

In other words, the third game decides matches not because it is the most tense. It decides matches because that is when the body stops pretending.

I use the 2026 World Championship final at Wembley between Lin Dan and Lee Chong Wei as a reference sample. The three games finished 20-22, 21-14 and 23-21. In my log, the third game contained the longest rallies but the lowest average shuttle speed of the three. Both players were still performing at the highest level; they were simply performing in a different way.

That is why I distrust conclusions of the kind that says one player is stronger because his smash speed is higher. Smash speed is an index of the second game. The third game is decided by recovery capacity between rallies, something that appears in almost no official statistics table. The same logic applies to Viktor Axelsen's speed in the Paris 2026 Olympic final he won against Kunlavut Vitidsarn, or to An Se-young's composure on her way to the women's singles gold. Neither can be explained by a single speed figure.

In doubles, the picture inverts. Rallies are shorter, the transition index is higher, and the volatility coefficient is lower because each rally contains fewer decisions. Curiously, audiences tend to assume doubles is more exciting because it is faster. My logs suggest doubles contains fewer turning points; the turning points simply happen too fast for the human eye to register.

There is one detail I must state plainly, because hiding it would turn everything above into a fraud. My logbook is not part of any federation's official data system. It is a personal observation sample, skewed towards the matches I was assigned to follow, and therefore possibly skewed towards players who appear on television more often. My sample does not represent world badminton as a whole. It is only enough to open a question.

The counterintuitive angle: a shorter format does not make a match less random

The five-game, 11-point proposal was sold to the public with two promises: shorter matches and more drama. Both promises are true in their crude form, and both are false in their real form.

A shorter format reduces the number of rallies needed to finish a match. It also reduces the number of rallies needed for a lucky run to overturn the result. When the number of observations falls, variance rises. A system with fewer points is a system more easily decided by a lucky net cord, a line judge's call, or a cramp that arrives at exactly the wrong moment.

I have written before about allowing five substitutions in football. The same argument appears there, only in the opposite direction: deeper squads benefit, but the final twenty minutes become a war of attrition. In badminton, the short format works against popular intuition. It does not reward a technical foundation built over a decade. It rewards whoever can tolerate uncertainty.

For a sport whose youth-development pipeline demands a full decade, devaluing technical foundations is a commercial decision, not a technical one. And it is presented as a technical one.

What I missed during that debate was the commercial layer of the story. I spent two weeks on rally-distribution tables without noticing that the real argument was happening somewhere else: in broadcast rights prices. Streaming platforms are paying heavily for sports rights, accepting losses to capture audiences, and repeating the very mistake pay television made two decades ago. When a sport has to shorten its matches to hold viewers, the problem lies in the calendar and the distribution model, not in the number of points per game.

There is one further layer I always raise in conversations with colleagues. Live in-match data, the kind sold to betting companies, is the darkest side effect of sport's digitisation. It does not make the sport more transparent. It turns every rally into a short-term financial asset and turns the viewer into a line of code.

What I will track in the next round

In the coming rounds I will not count points. I will measure the rest gap between rallies in the third game, then compare it with the player's last three matches. If the rest gaps lengthen while the number of long rallies does not rise, that is a signal that the body is pushing back against the technique.

Elite badminton is entering a phase in which the winner is no longer the hardest hitter, but the player who knows how to ration energy across three games. My logbook has not answered that question yet. But in this line of work, a correct question is usually worth more than a fast answer.