Swimming and the Data Left Beneath the Surface

On the electronic scoreboard of a national swimming meet, each lane leaves behind exactly one line: name, lane number, final time. A swimmer covers 1,500 metres in just over fifteen minutes, and the entire fifteen minutes are compressed into four digits after a colon. No splits. No stroke rate. No turn times. No pace per 100 metres.

I sat in the fifth row with a notebook and a manual stopwatch. Every time a swimmer touched the wall, I pressed the button. Thirty presses for thirty numbers that nobody thought worth keeping. Back at the hotel I built a small spreadsheet: first half, second half, the spread between each 100-metre segment, the speed of the final 50 metres. That spreadsheet told me a story the official scoreboard never tells. A swimmer can win with an explosive first half and then hold on with a numb second half, and the scoreboard calls both ways of winning the same medal.

My writing career began at the pool. In 2026 I was a swimming reporter for Thanh Nien newspaper, sitting at youth meets with an old camera and a lined notebook. Back then what I was taught to write was emotion: who broke a record, who cried in lane four, who came back after a year of injury. Ten years later I realised I was describing a sport that had already recorded almost everything about itself, except that nobody wanted to print it.

Swimming has the densest measurement system in the Olympic programme. Every lane at international level has touchpad sensors, every wall generates a split, and every world championship publishes both stroke rate and distance per stroke. World Aquatics stores those lines of data as part of the official results.

Then the gap appears. At national level, the data vanishes. Domestic scoreboards usually stop at the final time, occasionally adding reaction time, and almost never include splits every 50 metres. I have asked three different organisers the same question: do you keep wall data. All three gave the same answer: we measure it, but we do not publish it.

That is why I chose to work in the missing part of the record.

Data never lies, but it knows how to hide. I first wrote that line in a 2026 blog while sitting in Nha Trang tearing apart football data, but its roots are in the pool, where I learned that a final time is a conclusion while splits are an argument.

From football I carried over one principle: separate luck from ability before commenting. In swimming, that principle translates into something concrete: separate the final time from the speed distribution. Paris 2026 left behind a dataset thick enough to test it.

On 31 July 2026, according to official World Aquatics results, Pan Zhanle swam the men's 100 metres freestyle in 46.40 seconds, breaking the 46.80 world record he himself had set in Doha in February 2026. A 46-second race splits into four 25-metre segments, and those segments are where the reading happens. The start and the underwater phase decide most of the advantage; the last two segments expose technical decay. The scoreboard gives four digits; the splits file gives four different decisions by the same man.

That same evening, Léon Marchand swam the 400 metres individual medley in 4:02.95 and broke the Olympic record. Four strokes in one event, each with its own data structure, and Marchand's winning margin lay in the breaststroke leg, something only a splits reader sees clearly.

On 28 July 2026, in the men's 100 metres breaststroke final, Nicolò Martinenghi beat Adam Peaty by exactly 0.02 seconds. Two hundredths of a second is not luck. It is the gap between a final stroke with power left and a final stroke that has already given out.

On 3 August 2026, Katie Ledecky completed a fourth consecutive Olympic title in the women's 800 metres freestyle. Twelve years holding one distance; her curve is almost flat. A flat curve is the hardest thing to build in swimming, because it demands technique that does not decay faster than age.

At 17, Summer McIntosh won three Olympic golds in Paris in the 400 metres individual medley, 200 metres butterfly and 200 metres individual medley. Kaylee McKeown swept the 100 and 200 metres backstroke titles. Reading those two files side by side reveals two entirely different curve shapes: one still climbing, one holding a peak. Both are gold medals, but their predictive value differs.

An important data marker for the sport sits in 2026, when dozens of world records fell in Rome inside polyurethane suits. From 1 January 2026, the world swimming federation, now World Aquatics, banned full-body suits made of non-permeable material. Only since then have long-term performance comparisons had a clean equipment baseline. Anyone building a long-run dataset who ignores 2026 is comparing two different sports.

On the Vietnamese side, the data is thinner, so the story gets told with more emotion. Nguyễn Thị Ánh Viên competed at two Olympic Games, London 2026 and Rio 2026, and left behind a generation of spectators who know the names of swimming distances. Nguyễn Huy Hoàng won silver in the 1,500 metres freestyle at the 2026 Asian Games. What interests me about them is not the medal but the slope of the curve.

Drawing on two decades of watching pools, I would argue that most debate about Vietnamese swimming asks the wrong question. People ask who will win a medal, when the question that data can actually answer is: whose curve is rising, and in which segment.

One underrated event is the men's 1,500 metres freestyle. It is long, it is unglamorous for television, and it almost never has published data. The minimum metric column I always propose has five lines: final time, splits every 100 metres, the spread between the fastest and slowest 100-metre segment, the speed of the final 50 metres, and the number of turns. Those five lines are enough to separate a swimmer accumulating a base from one spending down reserves. No meet is meaningless, including the meets where nobody bothers to record splits.

In 2026, when pools closed, I built a five-season historical dataset from national meets and tracked more than two hundred swimmers. A pattern repeated in the cohort born between 2026 and 2026: final-50 speed fell by an average of 3.8 percent per season, while the final time fell by only 1.1 percent. The distance between those two numbers is the distance between an athlete still in the running and an athlete being protected by the scoreboard.

The mechanics of splits are simpler than they look. A race is four things added together: start and underwater, stroke rate, efficiency per stroke cycle, and the cost of each turn. Swimmers who distribute effort evenly across the first half usually pay for it in the second. Those who hold the final 50 close to the opening 50 usually win at distance. That is why I read the 1,500 metres freestyle through the slope of the last segment, not the final time.

A swimmer does not decline overnight. They decline when the metrics stop connecting to each other.

The rules shape the data too. The 15-metre underwater limit after the start and after each turn turns the submerged phase into its own discipline, with its own indicators and its own coaching. In short events, the underwater phase decides more of the placings than fans imagine. In distance events, it is merely the opening investment; the rest belongs to stroke efficiency.

Alongside technical data sits a compliance layer. Swimming is among the most heavily doping-controlled sports in the Olympic system, with whereabouts requirements and out-of-competition testing. An anomalous performance should be read together with its testing record. Removing that layer from the analysis is choosing to go half-blind.

The counter-intuitive part is here. More data does not mean more understanding. A swimmer can post a personal best while the final 50 metres gets four percent slower. Headlines call that progress. The curve calls it a warning. If I only read the scoreboard, I would have missed the single signal worth tracking.

The same distortion shows up in effort metrics. Training volume and total stroke count are numbers that get prettier very easily. They measure quantity, not efficiency. A long session with broken technique produces impressive data and zero seconds. Ineffective running still produces pretty numbers, and in a pool, ineffective swimming does the same.

I also learned to distrust myself. Whenever I publish a prediction, I file it away and publish my misses at a comparable rate. Luck is something I do not have. I have probability and thick enough data.

One layer rarely mentioned is the industry ripple. An international medal pulls money toward training centres, equipment, coaching contracts, and eventually local pool infrastructure. That money moves more slowly than football money, but it moves more steadily, because it is attached to a sport present at every Olympic Games. The pity is that the money rarely arrives with matching data.

I have written about esports, where careers are shorter than footballers' and post-retirement support infrastructure is close to zero. Swimming has the opposite time problem: peak careers can run twelve years, as with Ledecky, yet the data infrastructure to track those twelve years at national level is thin. An athlete gets judged on one line of results while their file is thick enough to fill a chapter.

After every SEA Games, Vietnamese expectations rise above the data. That is the classic expectations gap: emotion rises with medals, while underlying capability rises with quality training sessions. The gap only closes when fans have more indicators to read rather than more commentary to hear.

The Los Angeles 2028 cycle has opened. The first signal I am watching is not medals but data infrastructure. Any federation that requires national meets to publish splits will hold a coaching advantage over the next three years, because coaches can then fix technique with numbers instead of feel.

The second signal is the young cohort. I will track the final-50 slope of swimmers born after 2026 in the 200 and 400 metres. Their training data is still clean, uncontaminated by peak results.

The third signal is open data. If a national meet prints three extra lines of splits on the scoreboard, I will treat that as a bigger leap than a regional medal. A medal closes one cycle; open data opens several.