The Empty Data Layer of Vietnamese Swimming: Why Post-Race Analysis Still Stops at Emotion

On the electronic board of a SEA Games 1500m freestyle final, the only thing that lingers in most people's memory is the finishing time: one line, four digits, two colons. No splits, no stroke rate, no distance per stroke, no underwater time off the turn. Fifteen minutes of a human being wrestling with water compressed into a single string of characters, with the rest of the story handed over to emotional language.

I have sat down many times after sessions, reopened the results board, and tried to find a second data layer behind the final time. Every time, the second layer does not exist. The organisers publish the result. Nobody publishes the process.

Here is the comparison between two data landscapes, based on what I collected from official results boards of regional Games and from technical documents published by European swimming federations:

| Data layer | International standard publication | Actual publication at regional meets | |---|---|---| | Final time | Yes | Yes | | Splits at 50m or 100m | Yes | Rare, unsystematic | | Reaction time off the start | Yes | Almost never | | Stroke rate and distance per stroke | Yes | No | | Turn time and underwater distance | Yes | No | | Opponent analysis by lap | Yes | No |

The table is not a complaint about a sport. It is a description of the raw-material landscape. When the raw material is a single line, any post-race analysis must begin from an empty space, and an empty space cannot produce a verifiable conclusion.

Context: the paradox of a regional power

Vietnamese swimming carries a measurable paradox. At regional level it is a genuine force. Nguyen Thi Anh Vien collected more than twenty SEA Games gold medals between 2026 and 2026, according to organisers' medal tables across those Games. Nguyen Huy Hoang, born in 2026, stepped onto the continental stage with a bronze in the 1500m freestyle at the 2026 Asian Games in Jakarta. Tran Hung Nguyen, Pham Thanh Bao and a generation born in the early 2000s have kept their places in medley and short-distance freestyle events.

At world level the gap remains. That gap is not my question here. My question is this: if a regional swimming nation wants to close it, it must know exactly where it is losing time. To know that, it needs splits. And splits do not exist in the publicly available data.

The consequence is a narrative economy with a single currency: medals. People count gold, silver, bronze. They rank by medal table. Nobody ranks by improvement slope, by consistency across swims, or by speed decay over the final two hundred metres. Swimmers are described with adjectives, not with coefficients.

I began working in 2026, covering swimming for a newsroom. Back then I wrote from observation. I described the lane, the breathing rhythm, the moment a swimmer lifted their head out of the water. Twenty years later I looked back at all those pieces and realised one thing: I had never once verified what I had just written. I wrote "faded over the last two hundred metres" without a single line of data proving that swimmer was slower over that exact stretch.

That is why I moved to spreadsheets.

A single time line is not enough to tell a race

Take a 1500m freestyle race apart into its layers.

The first layer is the start. Reaction time is measured from the starting signal to the moment the foot leaves the block. At Asian level, most swimmers fall between 0.65 and 0.80 seconds. The spread between the fastest and slowest in a final can reach 0.15 seconds. Over 1500m that is negligible. Over 50m it is an entire ranking.

The second layer is splits. This is the most important layer and the one most often missing from the region's public data. A 100m split gives you a curve instead of a point. That curve answers three questions a final time can never answer: how long the swimmer held speed, where they collapsed, and by how much.

There are two distribution patterns. The first is even pacing, with each 100m within roughly one second of the others. The second is a first-half surge followed by progressive fade. Arriving at the same final time, the two patterns mean entirely different things. The even pacer still has headroom to accelerate. The surger is spending capital, and that capital runs out within one or two seasons.

With only a final time, the two patterns look identical. A commentator looks at it and writes: magnificent willpower.

Data never lies, but it knows how to hide. Here it hides by disappearing entirely.

The third layer is stroke rate and distance per stroke. Stroke rate counts arm cycles per minute. Distance per stroke measures metres covered per cycle. The two are inversely related up to a critical point, then reverse. Pushing stroke rate past the threshold makes distance per stroke fall faster than the compensating gain, and total speed drops. That is the point most swimmers without metric-based coaching cross before an opponent crosses them.

The fourth layer is turns and underwater segments. In distance swimming, the turn is the cheapest place to accumulate advantage. A swimmer who leaves the wall 0.1 seconds faster and holds one extra metre underwater accumulates advantage across dozens of turns. Over 1500m in a 50m pool, that is twenty-nine turns. Multiplied out, that advantage can exceed the entire gap in pure conditioning.

I once tried to measure this layer by hand. In 2026 I built a spreadsheet interpolating expected indices from match footage of early rounds of a domestic league, timing each turn manually. It took three days per race, the error was large, and it could not scale. But it taught me something no book could: most of my "faded at the end" judgements were wrong about the location. The collapse point was not in the final two hundred metres. It was in the middle stretch, where speed had already fallen quietly and nobody was watching.

The case of a curve outside the prediction zone

There is another kind of distortion that an empty data layer cannot protect us from.

At a European championship, a striker scored far more goals than his own expected-goals figure. The press called it a moment of genius. I calculated the expected value of every shot and found the total expected value was roughly half the actual goals. That does not mean he was poor. It means the record sat outside the sustainable prediction zone, and anyone spending money on that record was buying a peak, not a curve.

Swimming has an identical version. A swimmer produces a personal best a few percentage points better than their own baseline, usually in a heat with no direct rival, in a fast pool and low pressure. That mark is framed and hung on the federation's wall. The next funding decision is built on it. The following season the swimmer returns to the old baseline, and the whole plan collapses with it.

I call it the sacred-peak error. It appears because there is no benchmark metric comparing one swim to that same swimmer over time. Without a moving average, every peak looks like a step up.

Curves instead of medal tables

My profession is transfer-market administration. In other words, my job is to price people before the results are announced. In football people look at the valuation table. I look at the curve. Many deals die before they are announced, and they die exactly there: a metric showing the trajectory heading down while the record still looks good.

Vietnamese swimming has no real transfer market. But it has an equivalent: allocation of investment slots, overseas training slots, international competition slots and scholarship slots. That allocation rests on the medal table. And the medal table is a lagging indicator.

Imagine two swimmers winning a regional gold in the same year. The first achieved it by even pacing, a flat split curve. The second achieved it through a single surge, with everything else below their personal average. On the medal table they are identical. On the curve, the second is at the top of an unsustainable cycle.

The second will be invested in more heavily, because their medal is newer, and because nobody has the data to know that peak has already passed.

I call this the halo-allocation error. It does not come from corruption or favouritism. It comes from the absence of the cheapest, most available data layer of all, requiring only someone sitting at each lane marking times.

Measurement thresholds and the volume trap

While pools were closed during the pandemic, I did something many colleagues considered pointless: I rebuilt a multi-season historical dataset, tracking a group of swimmers, focused on acceleration speed and the ability to hold speed once fatigued. It was exactly what I had done when stadiums closed and I reopened the domestic league directory. COVID closed the stadiums, I reopened the V-League directory. No league is meaningless, not even a provincial swim meet nobody broadcasts.

The result was a small but useful finding. One swimmer in the study group kept winning medals, but their acceleration index had dropped sharply against the previous season. The record was living off an old season, while the machine underneath had changed. When swimming returned, that swimmer fell back and lost their starting place.

I posted the warning on a community page with a season-by-season comparison table. The reply was one short line: don't sit far from the pool and talk about the water. I did not argue. I simply attached the raw dataset and the method to reproduce the calculation.

Most debates about sporting performance in Vietnam stall for want of exactly one thing I had here: a comparison across time. Without a previous season and a following season, all that remains is the impression of the current one.

At the same time I noticed another trap. Training volume is an appealing effort metric because it can always increase. Add hours, add kilometres, add sessions. A few years ago a load-management project made waves by publishing player running distances: over twelve kilometres in a match. The problem is that the largest distance usually belongs to the player running most to cover a team-mate's mistake. Distance covered and sprint counts are packaged as effort metrics, but ineffective running also produces pretty numbers.

In swimming, the equivalent is weekly kilometres. A swimmer covering sixty kilometres a week is not necessarily better than one covering forty-five, if most of the first one's volume is swum below threshold. Without a speed distribution by session, we are counting water, not measuring effect.

The data column of a neglected meet

There is a group of competitions Vietnamese swimming almost never looks at: provincial youth meets and internal time trials. No crowds, no broadcast, and therefore no data.

I still track them by hand. For each group of young swimmers I record four columns: age in months at the time of the mark, event, time, and the gap against their best previous swim of the same season. Four columns, no more. Those four columns are enough to separate a swimmer improving steadily from one popping up on the back of a single swim.

The real value of a sport sometimes lies where there is no money. Regional swimming holds a large quantity of quality signal inside unbroadcast meets, and we throw it away every season.

The counterintuitive angle: more data is not the answer

The first instinct of most people hearing about a data gap is to demand more data. That is an architectural mistake.

A set of thirty metrics collected by hand, entered into three unconnected spreadsheets, is worse than a set of five metrics measured consistently and stored in one structure. Value does not come from the number of metrics. Value comes from the ability to link one metric to another across time.

Swimmers do not collapse in a single evening. They collapse when metrics stop connecting to one another, when underwater distance shortens, stroke rate rises, and nobody notices that those two lines have just crossed at a warning point.

The second paradox is the regional medal. For years Vietnamese swimming has been encouraged by a signal that is correct but measured on the wrong ruler. A nation leading an event at the SEA Games can still be seconds off the Olympic standard, and those seconds do not close themselves simply by winning more regional golds. Nguyen Thi Anh Vien once came close to the final threshold in a medley event at Olympic level; according to the heats results, she stopped just short of a final berth. That is evidence that defeats at world level are usually decided by margins in segments nobody publishes.

Luck is something I do not have. I have probability and sufficiently dense data. At the level of hundredths of a second, luck does not exist. There is only technique and distribution of effort.

What it takes to patch the data layer

If I were asked to build the data layer for a regional swimming federation, I would not start with expensive equipment. I would start with one form and one timekeeper per lane.

Four minimum layers: reaction time off the start, splits at twenty-five or fifty metres, time off the wall after each turn, and stroke rate counted at two to three fixed segments. All are measurable by hand, with tolerable error around two per cent, at near-zero cost. The entire layer only has value if the data is stored in one structure, with one athlete ID key, one unit of measurement, one definition across every meet and every season.

Internationally this infrastructure has existed for a long time. European federations publish technical documents with splits for every round, allowing anyone to reconstruct a swimmer's effort distribution curve. The regional gap sits in exactly this layer, and it is not a technology gap. It is a process gap.

There is another reason this is more urgent than it appears. In sports with short professional lifespans, post-retirement support systems are near zero, and I have seen it in both esports and in short-distance swimming events. An athlete gives ten years to the lane, and when they stop there is no data record of themselves to carry into a coaching or analytical role. Building a data layer is not only about selecting people. It is also about keeping the profession.

The signal for the next cycle

The coming cycle brings another regional Games, another continental championship, and another medal table read aloud as if it were the whole story. What I am waiting for is not the medal that will be won.

I am waiting for a split table published openly. When that exists, for the first time in Vietnamese swimming we will know exactly where we are losing: in the turns, in the middle stretch, or in the final sprint.

If the next Games still produce only a single time line, then the only thing we will have measured is the audience's excitement. That is a media metric, not a water metric.

One last qualitative note, the part spreadsheets never touch: after every final I still stand at the rail above the stand and look down at the pool. There is a silence lasting a few seconds after a swimmer touches the wall, before the announcer reads the winner's name. In that silence the losing swimmer is still breathing, still recounting their own turns in their head, and is often counting wrong. Data will tell them where they counted wrong. Our job is to keep enough raw material for that answer to exist.