The Data Gap in Vietnamese Swimming: When the Numbers Don't Arrive in Time

In the last three SEA Games, the most important data column in my analysis sheet — the 50-metre split time — was empty for the majority of Vietnamese swimmers' finals. I sit far from the pool deck to see a race more clearly than the referee, but that distance cannot fill a different gap. When a swimmer touches the wall, I have the total time. I do not have the acceleration curve at any segment, the percentage of time lost on a turn, or how far a start reaction deviates from a personal peak.

That is the paradox of Vietnamese swimming. We own the results, but we lack the process that produced them. A swim result without split structure is like a goal without a shot map: we know what happened, not why it happened, and least of all whether it can be repeated. The shot appears once. Its trajectory stretches across years.

Context: results are the surface, data is the depth

Vietnamese sport has long been used to measuring output. Medals, national records, final rankings — these are the numbers that get recorded, broadcast and archived. But swimming is a sport of process, not only of outcome. A 200-metre race consists of four 50-metre segments, each with its own structure: start and dive, acceleration, sustain, turn and finish. Error accumulated across segments decides the final result more than any single factor.

When I worked with GPS data from swimmers, the first thing I learned was this: the distance between two wall touches does not tell the story. Stroke rate, distance per stroke, turn count, dive angle after the start — those tell the story. A swimmer can cover 200 metres of butterfly with the same total time across two seasons, yet with a completely different effort distribution: in the first season, front-loading the first 100 metres and fading; in the second, holding rhythm and exploding over the final 50. Two identical results, two different bodies, two different strategies, one forgotten data signature.

Vietnamese swimming has a generation of swimmers good enough to produce high-quality data. Nguyễn Thị Ánh Viên once competed at world level, with swims at the ASIAD and the Olympics. Nguyễn Huy Hoàng is a pillar of the 800-metre and 1500-metre freestyle. But looking back, I realise most of their data legacy is results, not process. The national record book records time. It does not record how that time was made.

This is where I must state my method clearly. A swimming analysis model needs at least four data layers: the biological layer (height, arm span, fitness indices), the technical layer (stroke rate, distance per stroke, turn count), the tactical layer (effort distribution by segment) and the operating-conditions layer (water temperature, altitude, competition schedule). Missing any layer, the model still runs but its judgments turn fragile. When the technical and tactical layers are both empty — as is the case for most Vietnamese swimming data — we are left with a dry results table, enough to rank, not enough to understand.

Core: the evidence chain of a gap

Let me reconstruct a specific case. I followed a young domestic swimmer across three consecutive seasons. In the 400-metre freestyle, his best time improved steadily: season one, season two, season three, each faster than the last. On the results sheet, it was a perfect upward straight line. But when I had the chance to review the footage and hand-time each 50-metre segment, the picture shattered.

In season one, he swam the first 200 metres fast and held to 300, then fell off a cliff over the final 100. In season two, overall speed was faster but the distribution leaned further forward — meaning peak speed rose while closing endurance did not improve. In season three, the best time kept dropping, but this time thanks to a stronger final 50, while peak speed over the first 100 fell. Three progressively better results, three completely different bodies, three different problems. Looking only at the results sheet, a coach would wrongly conclude that everything is on track, when in fact the swimmer is trading peak speed for endurance — a gamble that can blow up on a major stage.

This is the specific physical mechanism I want to point out: in middle-distance swimming, peak speed and closing endurance are two variables that are related but not identical. Improving total time can come from raising peak speed, from reducing late-race deceleration, or from optimising turns and dives. Three different roads lead to the same timeline. Without split data, we do not know which road a swimmer is on, and therefore do not know what to train.

I once built a framework of my own called the decay curve for swimming: comparing the fastest 50-metre segment against the slowest 50-metre segment within a single swim, then tracking that ratio across seasons. The smaller the ratio, the better the ability to sustain rhythm. With full data, this framework distinguishes three swimmer types: the explosive type (very fast opening, steep decay), the rhythm-keeper (stable speed) and the closer (accelerating in the final segment). With an empty data table, the framework cannot run. I keep it in a drawer, waiting for a dataset thick enough.

The simple amateur question here is: if we do not know where a swimmer decays, how do we know what they should train next month? The honest answer is: we are guessing. And guessing in swimming costs more than in many other sports, because the fitness adjustment cycle spans weeks, and a mistake in effort distribution may only surface at the very championship.

The fitness paradox: when data arrives too late

In 2026, when the pandemic froze domestic swim meets, I had a chance to review the fitness data of a group of athletes. What I found was not on the pool deck but in the gym. High-speed workload spiked in the weeks before shoulder and muscle injuries appeared. The model was simple: when the proportion of high-intensity work crosses a certain threshold without a corresponding deload week, injury probability rises exponentially. I proposed splitting training into four load thresholds and rotating them cyclically. As a result, injury cases fell significantly the following season.

But the real story of mine is not that achievement. The real story is: I only discovered that model after the injuries had happened. Fitness data arrived late, just as split data arrives late. I did not predict injuries; I only explained them after they had claimed a swimmer. In Vietnamese swimming, most of the data we have is post-mortem data — it tells us what killed a season, not how to save it before it is too late.

This is the point I want to stress: a data gap is not merely about missing numbers. It is about numbers arriving at the wrong time. A metric that arrives after an injury has only academic value. A metric that arrives before an injury has practical value. The distance between those two moments is the entire difference between a data consultant and an obituary writer.

Regional context: when rivals have moved ahead

At the SEA Games, Vietnamese swimming often has to compete with Thailand, Singapore and Indonesia — countries with more systematic data-recording systems. Singapore stands out in particular with a swimming programme quantified from school level. I do not say this to compare results, but to point out an information asymmetry. When two swimmers race side by side, the one with full split data knows which segment to attack. The one without data only knows to try their hardest. In a race decided by fractions of a second, the distance between those two states is the distance between a medal and fourth place.

I once analysed a 4x100-metre freestyle relay final on the regional stage. The Vietnamese team had a good total time, but when I separated each leg's start time, I found one opening leg notably slower than the rest. That was a technical signal, not a fitness signal. Had start data been recorded regularly, the issue could have been fixed in a few training sessions. Without data, it becomes a destiny that repeats every championship.

I want to restate one thing about the local character of analysis. We cannot transplant European models wholesale into Vietnamese swimming. Climate, body type, domestic competition calendar and training resources all differ. A model built from data at large overseas training centres may not transfer to a provincial pool. Analysis must grow from its own soil.

Contrarian angle: more data does not automatically produce medals

Here I must argue against myself. There is a widespread belief in sport that once enough data is collected, results will follow. I do not believe it, and I want to show why it is dangerous.

First, correlation is not causation. The fact that a swimmer has a higher stroke rate and swims faster does not mean the higher stroke rate creates the speed. Both may be consequences of a third factor: a better fitness base. If we force a swimmer to raise stroke rate without building the fitness base, we do not create speed — we create shoulder injury. Before saying X causes Y, I must point to the physical mechanism linking the two variables: propulsion in water, arm angle, breathing coordination. Without that mechanism, I only use the word related.

Second, data does not replace judgment. A model can say a swimmer has a 70% chance of meeting the A-cut. It cannot say whether she should rest this week. That decision belongs to the coach, who sees the swimmer's eyes in training in a way no table records. Data narrows the blind spot; it does not erase it.

Third, there is a portion of variance that cannot be explained. In a swimming final, the stands may roar, but that emotion cannot be quantified into a stable number. At major championships, I accept a wider confidence interval for every judgment of mine. When crowd conditions exceed historical thresholds, I note clearly that my model is stepping outside its familiar data range.

Therefore, the data gap in Vietnamese swimming cannot be filled by importing equipment and software en masse. It must be filled by a habit: recording the process, not just the result. A hand-timed notebook at every lane, capturing all four 50-metre segments, can be more useful than an expensive sensor system that stores only total time. Every shock has its own probability. We call it a shock only when we have not checked the numbers. And in swimming, the numbers often do not exist to be checked.

System and people: where data must begin

There is a truth about swimming I do not want to skip. This is a sport of youth. A swimmer's peak often comes early, and the competitive life cycle is far shorter than in many other sports. That means every season lost to missing data is a significant share of an entire career. There is no time to start over. Football and esports do not differ in essence, only in reflex tempo — and swimming differs in that it does not give us a second chance at the same age.

So the data system of Vietnamese swimming must begin at grassroots level. A fifteen-year-old swimmer needs their stroke rate, distance per stroke and effort distribution recorded, not just their finish time. By the time they are twenty, we will have five years of data to understand how their body is changing. Without that foundation, all analysis at national-team level is merely scattered pieces of a picture never painted.

I have spoken before about contracts in swimming, but even at athlete level, an investment in a young swimmer is not a signature. It is a hypothesis signed into being. And every hypothesis needs data to be tested. When we invest in an athlete without tracking data, we are betting on belief, not on a model.

Ordinary people look at goals to understand a match. I look at a match to understand the years. In swimming, those years are the chain of data accumulated across seasons, fitness phases and technical adjustments. A tactical era dies when its data table no longer has readers. I do not want the next generation of Vietnamese swimmers to grow up in an era where their table is empty from the start.

Conditions for the forecast to hold

I always close each analysis with a section like this, because a forecast without stated conditions is an empty promise. For Vietnamese swimming, I set three conditions for a data model to work.

First, split data must be recorded at every official swim, including heats. Without it, there is no decay curve, and without a decay curve there is no real tactical analysis.

Second, fitness data must be collected continuously, not only when there is an injury. The goal is to turn data from a post-mortem tool into a prevention tool.

Third, there must be a person responsible for reading the data and translating it into coaching language. Data sitting on a hard drive is useless. Data lives only when someone turns it into a decision on the training floor.

These three conditions do not require expensive technology. They require discipline. And discipline, in sport as in analysis, is the cheapest thing to start and the most expensive thing to sustain.

A forward-looking reflection

I do not write this to complain about what Vietnamese swimming still lacks. I write it because a data gap is not a death sentence. It is a blueprint waiting to be built. Every empty table today is a chance to build the habit of recording for tomorrow. And if there is one thing I have learned after years of hand-timing in front of a screen, it is this: the truth does not lie in the final number. It lies in the distance between two numbers we never recorded.

Vietnamese swimming will enter the next season with a new results sheet. The question I want to raise is not how many medals they will win. The question is: will we have enough data to understand why they won them, or why they did not? If the answer is still no, then every triumph and every failure will keep carrying a baseline probability we can never look up. And when that happens, people will call it a surprise again — only because our table never existed.