The 102-Point Dataset and 122 Seconds in Aichi-Nagoya: Where PV Sindhu's Asian Games 2026 Run Ended

In the fourth minute of the deciding game, the stopwatch on my hand froze at 122 seconds. I was sitting in row six of the Aichi-Nagoya arena, notebook open, and I counted 61 strokes in that rally. PV Sindhu won it. She then lost six of the next seven rallies, and lost the match.

I recorded all 102 points of the women's singles quarter-final between PV Sindhu and Chen Yufei. The final score, 21-11, 18-21, 10-21, sits on one page. Three lines of data sit on the next. Those three lines tell a very different story from what the stands had just witnessed.

This match will be remembered for the longest rally of the tournament. It deserves to be remembered for something else.

The Match Frame and How I Measure

PV Sindhu entered the women's singles quarter-final in Aichi-Nagoya as a two-time Olympic medallist who had held a place among the world's elite for nearly a decade. Chen Yufei, across the net, was the Tokyo 2026 Olympic champion and a former world number one. On 27 September 2026, the two met in the quarter-finals — a round that a 31-year-old like Sindhu no longer has many chances to reach.

Badminton has no xG. Nobody assigns a probability to a smash. I had to build my own yardstick, and I built it in twenty minutes before the umpire called for serve.

The first layer is the point-by-point score, recorded by hand, one box per point. The second layer is rally structure: length in seconds and number of strokes. The third layer is movement load, measured as distance covered and average foot speed per rally. The fourth layer — the one I care about most — is tactical choice at pivotal points, what I call the turn point.

I use the Expected Point per Rally Index, or EPP. It answers a simple question: from that court position, at that contact height, after that tempo, what percentage of rallies does a top-tier player win? EPP does not replace the human eye. It only forces the human eye to explain itself.

I still remember the rule I set in May 2026, when I opened my spreadsheet on the V-League match at Lạch Tray and realised that tactics never have a gender. That rule is: data first, emotion after. Today's Aichi-Nagoya match was recorded by exactly that rule.

Game One: A False Signal

Sindhu won the opening game 21-11 in 22 minutes. On the arena scoreboard, that was a demolition. In my notebook, it was a warning.

She closed the game with 11 winners and 3 unforced errors. Chen Yufei had only 4 winners but committed 9 errors. Average rally length was 6.4 seconds, equal to 5.1 strokes — the lowest of the three games. Sindhu's average smash speed reached 369 km/h, and her scoring rate when she came to the net was 71%.

Anyone who has followed Sindhu at major events knows she detonates from the very first point. But one detail did not fit: 22 minutes. A 21-11 game lasting 22 minutes means the match had never touched Chen Yufei's tolerance threshold. The Chinese player had not had to run, had not had to bend, had not had to do anything except push the shuttle over the net and wait.

I wrote in the margin: "Chen hasn't played. Sindhu has played out the whole first game."

A single rally is chance, but a whole tournament is where probability exposes every truth.

Game Two: The Turn Point at 14-13

Stepping into the second game, Chen Yufei changed one thing only. She returned serve deeper. Her rate of service returns landing in the backcourt jumped from 64% in game one to 81% in game two.

The consequence was immediate. Average rally length rose from 6.4 seconds to 11.3 seconds — nearly double. Average strokes per rally rose from 5.1 to 9.6. Sindhu was forced to hit four extra strokes per point, and every extra stroke is one more chance to err.

Sindhu finished the second game with 7 winners and 9 unforced errors. Chen had 12 winners and just 4 errors.

I looked for the turn point. It sat at 14-13 in Sindhu's favour, when Chen served short. Sindhu chose to attack on the third stroke — a cross-court drive to the left boundary. That choice, across the whole match, succeeded 2 times out of 9. She lost that point. The next four went with it.

That was the moment the match changed hands. Not the long rally. Not a smash. Just one choice at one point where a veteran should have known better than anyone.

I do not deny the error margin. Over 9 attempts, a 2-in-9 rate sits inside a wide confidence interval, and one bad choice cannot explain the remaining 29 minutes. But the pattern holds: across the match, whenever Sindhu lost a second consecutive point, the probability she would lose the third rose to 68%. Chen Yufei read that before I had finished filling in the third box.

Chen won the second game 21-18. One game all.

Game Three: 122 Seconds and Its Price

The rally in the fourth minute of game three lasted 122 seconds. 61 strokes. Three times the two players touched the shuttle within about a second of each other. Four times Sindhu was forced back to the left rear corner, where, according to my movement-load data, she burns the most energy.

She won that rally. The stands rose. And here is what the dataset recorded: Sindhu's average smash speed in game three was 341 km/h, down 28 km/h from game one. Her distance covered in game three was 980 metres, against 1,180 metres in game one. Her average foot speed per rally fell from 4.2 metres per second to 3.4.

Average rally length in game three: 15.7 seconds, equivalent to 12.8 strokes.

After that 122-second rally, Chen Yufei won 15 of the remaining 20 points. Across the match, she won 42 of the last 60 points. Game three closed 21-10 in 19 minutes.

Sindhu finished the final game with 3 winners and 12 unforced errors. Chen had 13 winners and 2 errors. Sindhu's winner-to-error ratio flipped completely: from 3.67 in game one to 0.25 in game three.

When the media calls it a miracle, I call it a probability distribution.

The Counterintuitive Angle

The turning point of the match was not the 122-second rally. That rally was a consequence. The real turning point sat at 14-13 in game two, nearly half an hour earlier, in a choice nobody remembers.

And there is an irony the data forces me to write down. Sindhu's first game was too short for her to build match rhythm. She won 21-11 while barely touching the movement threshold of a training session. When Chen Yufei stretched the rallies in game two, Sindhu entered a zone she had never been warmed up for: rallies of twelve strokes, fifteen seconds, repeated over and over. Fitness was not where she broke. Rhythm was where she broke.

A reader familiar with my transfer pieces will ask at once: is this a model error or a human error? The answer is both. My model measures the decline, not the cause. And in badminton, the cause usually lies in a decision made thirty minutes earlier, at a point where the scoreboard had not yet raised any alarm.

I keep the principle: data never tells a sad story, it only points at whoever is lying to themselves. After this match, people will say Sindhu is finished. My numbers say Sindhu was finished in game two, at the nineteenth minute.

What to Watch in the Next Round

Chen Yufei enters the semi-final with something her rivals do not yet have: the ability to win a long match. Over the past six months, she has won 14 of 17 matches that included at least one game lasting longer than 20 minutes.

For Sindhu, the question is no longer whether she has enough technique to beat the world's elite. Her technique is still there, still enough to win a first game 21-11 against an Olympic champion. The question is whether she still wants to rebuild her match rhythm differently — accepting ugly, long, unglamorous games in order to save energy for the third.

I will reopen this spreadsheet in October, when she plays her next event. If her average rally length in the first game is still under seven seconds, I will rewrite this piece. If it rises to ten seconds, it means someone on her team has read the same index I am reading.