The Empty Cells in Vietnamese Swimming's Data Sheet

The Empty Cells in Vietnamese Swimming's Data Sheet

5:47 a.m., August 12, at a coffee shop on Tran Phu Street, Nha Trang. I reopened the swimming dataset I had been building for eighteen months, and the "50m split" column showed up with 43 percent of its cells empty. No formula error. No typo. Those cells were empty because at the meet I was tracking, the organisers published only final times for the heats, and published splits only for finals and semifinals. For nearly half the swims in my dataset, I had a result but no shape.

Outsiders will say: so what, just compare the results. My career started with a smaller deviation than that. In 2026 I recorded a striker's sprint distance in a V.League match as 1.2 km when the correct figure was 0.8 km, then spent three months re-checking 14,000 GPS samples to uncover three further systemic errors buried inside the synchronisation software. A small GPS deviation taught me this: verification is everything. Since then, my first move when opening any dataset is to count the blanks before reading the numbers.

That 43 percent column is why this piece exists. It is not a story about a technical glitch. It is a story about how a data-collection system gets built in order of priority, and how that order decides what we are allowed to conclude about Vietnamese swimming, which stories we may tell, and where we are obliged to stay silent.

The Three Layers of a Swimming Lane

Swimming is one of the few sports where results are produced by machines rather than by human eyes. But a result and a datum are two different things, and the gap between them is where I work.

The first layer is electronic timing. Touchpads on the wall, automatic timing systems, official result sheets. This layer is almost impossible to dispute: errors are measured in hundredths of a second, there are officials watching, there is a signed record. In Vietnam this layer is complete at national championships, regional multi-sport games, and international meets registered with the world federation.

The second layer is race analysis. This is what actually sustains expert work: reaction time off the blocks, 50m splits, stroke rate, distance per stroke, turn time at the wall, underwater distance after the start and after each turn. This layer does not arrive ready-made. It has to be created, frame by frame, from video.

The third layer is training load and physiology. Inertial sensors on the back, heart rate, blood lactate, heart-rate variability, recovery indices between sessions. In Vietnam this layer exists in a few training centres, is not widespread, and is almost never published.

These three layers do not fail equally. Layer one is stable. Layer two is systematically thin. Layer three is effectively invisible to the public.

The Real Cost of Coding a Race

There is a calculation few people make, and it explains almost the entire gap in my sheet.

A 200m swim at international level lasts two minutes. To code it properly, an analyst must watch the video in slow motion, mark each 50m checkpoint, count stroke cycles, measure underwater distance, and time the turns. An experienced person needs 25 to 40 minutes per swim. A regional games has several hundred swims across finals and semifinals. Multiply it out and you get several hundred person-hours, equivalent to one full-time employee for two months, all to answer one question: did this athlete accelerate on the third length or the fourth.

Regional games organisers do not have the budget for those two months of labour. They have budget for touchpads, for officials, for medal ceremonies. That is a rational choice, not negligence. But the consequence is that most specialist swimming data about Vietnam does not exist in retrievable form. It sits in coaches' phones, on broadcasters' hard drives, and in the memories of people who were in the stands.

Data does not tell stories; it records everything so that I can tell them myself. The problem here is that much of it was never recorded.

What Is Lost When the Cells Are Empty

When all I have is a final time, I can still compare. I can still rank. I can still say who was faster on a given night. But I lose three far more important capabilities.

First, I lose the ability to read pacing tactics. In middle and long distance swimming, the story lives in the third quarter. A 1500m swimmer may go out 1.5 seconds faster than usual, and that decision determines everything that follows. Without splits, I see only one final number, and a final number is the most deceptive object in sports analysis, because it conceals every cause.

Second, I lose the ability to separate technical progress from physical progress. If a swimmer shortens their turn time, that is technique, and it repeats. If they hold speed at the end because their aerobic base is deeper, that is conditioning, and it takes months. These two paths lead to entirely different training programmes. Without splits, I cannot advise a squad which way to go.

Third, I lose the ability to forecast. Performance models need split structure as an input. When the input is only a final time, the model must assume the swimmer paced evenly, and that assumption is wrong for almost every real swimmer.

So I handle empty cells by my own rule: if a conclusion requires layer-two data and layer two does not exist, that conclusion is downgraded to a hypothesis and never promoted to a finding. I believe in numbers, but only after a number has cleared three rounds of checking.

My Three Rounds of Checking

Round one is source checking. Does the number come from the official result sheet, from raw video, or from an article citing another article? A great deal of distortion in Vietnamese swimming data comes from citation chains with no terminus.

Round two is conditions checking. How deep was the pool, what was the water temperature, was there an anti-wave system, what was the altitude, which suit generation was in use. I have seen times a year apart compared directly, when one swim was in a shallow pool and one in a compliant one. Pool depth affects the turbulence bouncing off the floor, and it affects even world-class swimmers.

Round three is sample size. Three races do not make a trend. One explosive night does not make a career breakthrough. I usually have to wait a full cycle, meaning a season or a year, before calling a change real.

Vietnamese Swimming and the Question of the Start

Of all the segments in a race, the start is the easiest to measure and the most consistently ignored.

Reaction time is recorded automatically and appears in result sheets at many meets. At regional level it typically falls between 0.60 and 0.80 seconds. That means the spread between the fastest and slowest starter in a final can reach two tenths of a second. In a 50m race, two tenths is the difference between a medal and fifth place. In a 1500m race, two tenths is close to meaningless.

This sounds simple, but it has a large practical consequence. If a Vietnamese swimmer loses 0.4 seconds to a regional rival in the 100m, and 0.15 of that comes from reaction time, the remainder is only 0.25 seconds. That 0.25 can be addressed through stroke mechanics and turns. The 0.15 requires a dedicated start programme, with force-plate blocks, and not many facilities in Vietnam have them.

I have spent considerable time on this data column, and my conclusion is methodological: most analyses of Vietnamese swimming defeats skip the start because it looks small, yet precisely because it is small it is the most improvable part. People tend to hunt for causes where the fixes are hardest.

Underwater Distance and Invisible Technique

After the start and after each turn, a swimmer may travel up to 15m underwater. At world level this is where the biggest advantages are generated, and it is the least analysed area in emerging swimming nations.

The reason is practical: underwater distance is not recorded automatically. No touchpad measures it. To know whether a swimmer went 9m or 13m, someone has to watch video and count the markings. At regional meets, cameras are often placed at angles unsuited to this, and bubbles sometimes obscure the swimmer's legs for the first two seconds.

In my dataset, the underwater-distance column has the highest blank rate, above 70 percent. Which means I can describe a swimmer's stroke rate but cannot say how much time they lose at the turns. A technical analysis that omits this is like a football analysis without set-piece data: still useful, but it overlooks precisely the part that leading teams use to create separation.

This is the point I want to press with anyone building youth programmes in Vietnam: the competitive edge in regional swimming will not come from swimming more, but from measuring what rivals do not measure. Underwater distance is the clearest example, because it is tactically free and demands no extraordinary physical gift, only technique and patience.

The Problem with Comparing Across Eras

Whenever a new Vietnamese performance lands, I get messages asking where it stands historically.

The honest answer is that cross-era comparison in swimming is one of the riskiest operations in sport. At least four variables shift between two swims a few years apart: pool depth, suit material and buoyancy, timing technology, and lighting and temperature conditions that affect feel for the water. Polyurethane suits were banned from 2026, and that era left behind a stratum of records that even the world federation acknowledges cannot be compared directly.

With Vietnamese swimming the problem compounds: many national records were set in pools that do not meet international standards for depth or anti-wave systems. That does not devalue the performance. But it turns the question "what is this record worth on the world list" into something that can only be answered as a range.

I always answer with a range. For example: a performance might correspond to somewhere between 40th and 70th on that year's world list, depending on how you adjust. Giving a single number is lying, even when the speaker does not mean to.

The Counter-View: Data Scarcity Is Not a Scandal

This is where I have to argue against myself.

People who work with data share an easily spotted professional bias: when numbers are missing, they assume someone erred, that there is negligence somewhere, that with enough money everything would be measured.

I used to think that. Now I only partly do.

A country with a finite sports budget must choose between hiring a race analyst and paying for a month of board and lodging for eight young athletes at a training camp. Policymakers choose the second option, and mathematically that is the right call: athlete sample size matters more than data sample size, at least during a foundation-building phase.

The cost is not that we lack data. The cost is that we do not know we lack it, and therefore substitute intuition for data unconsciously. A coach who knows there are no split records will say "I need to watch more." A coach who believes the data is sufficient will say "he lacks speed," and that statement may be right or wrong and no one can test it.

In other words, the problem is not too little data. The problem is confidence out of proportion to the data available.

The Correlation Trap

In recent years I have seen one argument pattern repeat in discussions of Vietnamese swimming: an athlete moves to train abroad, their times improve, and the conclusion is drawn that the foreign training environment was the cause.

The correlation is clear. The causation is not.

Athletes who move abroad are usually at an age when physical capacity is developing naturally. They also change coach, change programme, change diet, change training partners, and change psychological motivation because expectations rise. Isolating "abroad" from those four other factors would require a study design nobody in sport conducts and never will, because you cannot randomise human beings into groups.

What I can say with more confidence is different: among the cases I follow, the largest improvements usually arrive within the first eighteen months after a coaching change, regardless of where the coach is based. If that holds, the real variable may be programme quality and individual attention, not geographic coordinates.

I write this knowing it is unattractive. It does not make a good headline. But it is what my dataset is saying, and I have a duty to read that before reading what I want it to say.

Lessons from a Rejected Transfer Analysis

In 2026 I advised a club against spending 500,000 US dollars on a foreign striker whose conversion rate was nearly double the league average, with 70 percent of his goals coming from set pieces. Club leadership overruled the recommendation, saying data cannot replace a scout's eye. The player scored four goals in twenty matches and suffered two hamstring injuries.

I tell that story not to prove I was right. I tell it because it taught me something that transfers to swimming: a model persuades nobody if people do not understand how it was built. A dataset presented without method, assumptions, and error margins gets read as an opinion, and anyone is entitled to ignore an opinion.

For Vietnamese swimming, that means publishing data matters almost as much as collecting it. A federation that publishes split tables with notes on pool conditions creates far more value than one that keeps those tables in a drawer and announces only results.

Recovery Indices and the Pandemic Lesson

In 2026, when the domestic football season was suspended for seven months, I built a recovery-index model from GPS data on hundreds of players across three seasons. When the league returned, the model predicted that the three teams applying the highest pressing intensity faced a 23 percent higher injury risk. The club I worked with cut training load by 15 percent and lost no key players.

I retell it here because its logical structure is exactly the structure swimming needs. A disrupted season is a natural experiment in how bodies lose and rebuild a base. Vietnamese swimming has been through at least two major disruptions in the past decade, and I have never seen an analysis of their effect on youth age-group structure. That is the most serious data gap I know of, because it affects an entire generation of athletes, not one season.

If there is one place I want data collection to start right now, it is here: training diaries for age groups 12 to 16 at provincial and municipal swimming centres. No expensive sensors required. A spreadsheet, a consistent data-entry person, and one rule that bad data must never be deleted.

What the Next Cycle's Signals Will Be

I am watching four concrete signals for Vietnamese swimming's next cycle, and I say plainly that these are signals, not predictions.

First, the appearance of split tables at the national championship. If the federation publishes splits for finals, that is a sign the second data layer has entered the official process rather than remaining the private work of a few individuals.

Second, results in the 200m and 400m individual medley. These events demand four different technical disciplines within a single race, so they expose technical weakness more clearly than any single-stroke distance. A rising swimming nation usually improves in medley more slowly than in its specialty events, and the pace at which that gap closes is a good indicator of overall coaching quality.

Third, the number of athletes aged 15 to 18 meeting qualification standards for continental junior meets. This is a lagging indicator, meaning it reflects work done four to six years earlier. If the number rises, the cause is not the current season.

Fourth, and this is the signal I care about most, the number of Vietnamese coaches who can read a split table. This skill spreads more slowly than money, but once it spreads it stays. A cheering culture lives not in the volume of the shout but in the frequency of patience.

What I Still Do Not Know

I must be honest about this article's limits.

The dataset I used is my own, assembled from public sources, not from a federation's internal database. The 43 percent blank rate applies to my dataset, not to any official statistic about Vietnamese swimming. The models I describe have not been validated on samples large enough to call evidence, and every causal conclusion in this piece should be read as a grounded hypothesis.

I do not know whether Vietnamese swimming's data gap will narrow in the next three years. I only know that the cost of closing it is far lower than the cost of continuing to make decisions on feel, and that in sport the most expensive thing is always a correct decision made too late.

At 5:47 that morning I did not delete the 43 percent column. I left it, coloured it differently, and wrote a note beside it: data not available here. To me, that is a kind of result. A result that says there is work to do.