Vietnam's Swimming Record Returns Empty: Why a Results Sheet Cannot Judge a Swimmer

On the night of 18 July 2026, I re-ran my extraction routine on a Vietnamese swimming meet document. The machine returned a single line: 0 information points. No athlete names, no events, no split times, no reaction times. An empty payload, literal in every sense of the word null.

The first reflex of anyone who works with data is to blame the machine. I opened the logs, checked the parser, re-ran it three times against three different versions of the same text. Nothing changed. On the fourth pass I read it with my own eyes and understood: the document genuinely contained nothing to analyse. It had a meet name, a ranking, and a finishing time for eight lanes — and that was all. Not one line of 50m splits, not one reaction time, not one note on water temperature or pool certification. The results sheet looked complete. It was hollow.

Data does not lie, but the people who read data do. Here the liar is the structure of the record itself: correct format, publicly posted, every field populated, and incapable of answering a single question of value. A results sheet without splits is a financial report with revenue and no costs — you know the outcome, you do not know how it was produced, and therefore you cannot reproduce it.

Vietnamese swimming publishes three things fairly consistently: start lists, final rankings and finishing times. That is the minimum required to hand out medals. It sits far below the minimum required to judge a swimmer. A national championship leaves behind a few hundred lines of time; the same meet, at the level of analysis, needs roughly fifteen thousand data points once you count 50m splits, reaction times, 15m in-and-out turn times and stroke counts.

Four things are almost entirely absent: 50m splits, reaction time, 15m turn times, and stroke rate paired with distance per stroke. A fifth layer — pool conditions, water temperature, automatic versus semi-automatic timing — is rarely recorded at all. Without those five layers, any comparison between two swims in the same event is a comparison between two pixels of a picture that was never printed.

I was born in China and I work in Vietnam, so I have a habit of placing the two systems side by side. China's national championships have published 50m split sheets for every individual event for years; provincial sports training centres maintain split databases across four-year cycles, dense enough to show where a fifteen-year-old stalls and where she breaks through. The gap between the two swimming nations is not in the pools or the training plans. It is in the capacity to read one's own past.

Vietnam is in the phase China passed through roughly two decades ago: facilities and talent identification first, measurement infrastructure later. That phase always produces a comfortable illusion — times improve, medals accumulate, and everyone assumes the system is progressing. Without splits, nobody can separate real progress from a year in which the calendar happened to be denser, or a pool that happened to be faster.

To judge a swimmer I need five layers stacked on each other. Layer one is the finishing time — the only one Vietnam has. Layer two is the split. Layer three is the wall: reaction time, 15m in, 15m out. Layer four is stroke architecture. Layer five is the physical context of the lane. Remove the last four and the first becomes an assertion nobody can verify.

A 100m event has two split points; a 400m has seven; a 1500m has twenty-nine. The difference between a negative-split swimmer and a swimmer who goes out fast and dies is invisible in the finishing time — two people can touch the wall within the same hundredth of a second running completely different physiological models. For the first, the aerobic block is paying off; for the second, it is evidence of a taper placed at the wrong moment. One line of time, two opposite conclusions, and without splits there is no way to choose correctly.

The wall is where Vietnamese swimming loses the most and measures the least. At continental level a good start produces a reaction time of 0.60–0.68 seconds; a good turn and underwater phase delivers a swimmer past the 15m mark faster than their own average swimming speed. In a 100m event with four wall contacts, the accumulated difference between a strong turner and an average one routinely exceeds the entire aerobic gap of a training season. Nobody in Vietnam publishes 15m times. So when a swimmer loses by 0.8 seconds in a final, nobody knows where 0.6 of those seconds went.

Layer four is stroke architecture: stroke rate, distance per stroke, and the stroke index that is the product of the two. It is the cheapest layer to collect — one person with a stopwatch and a count of cycles over 25m — and the most frequently skipped. Two swimmers can both go 51 seconds in the 100m freestyle with a rate of 48 and 2.1 metres per stroke, or a rate of 56 and 1.8 metres. The first needs power; the second needs technique. Hand them the same programme and you will break one of them.

Layer five is context: is the 50m pool certified, what was the water temperature, was the timing automatic or semi-automatic, were the starting blocks regulation. This sounds like paperwork, but it determines whether a national record is comparable to anything. A short-course record cannot be placed beside a long-course mark without a conversion factor; a semi-automatic record carries a thumb on a stopwatch that nobody recorded. When the world stops turning, I build my own data loop — but a loop only means something if every lap is measured by the same clock.

Take Nguyễn Huy Hoàng. In the 1500m freestyle, the bronze at the 2026 Asian Games and the Olympic berths for Tokyo 2026 and Paris 2026 are public landmarks. Landmarks say nothing about whether his third 500m is a fade or a deliberate float. For a distance swimmer the middle is the whole story: it tells you where the aerobic threshold sits and how far it has moved across a cycle. Without splits, a good training block and a wasted one produce the same line of results.

Trần Hưng Nguyên in the 400m individual medley is harder still, because the event splits into four stroke segments with four separate logics. The gap for a 400m IM swimmer usually concentrates in two places: the butterfly-to-backstroke transition, and the final breaststroke leg. The breaststroke leg is where underwater technique wins or loses 0.3–0.5 seconds at every wall. A sheet that reads 4:20 tells you nothing about where the 0.4 seconds went, and therefore nothing about what to fix on Monday.

Phạm Thanh Bảo in the 100m backstroke is a wall case, since backstroke carries the longest underwater phase of the four strokes after the start and after each turn. Võ Thị Mỹ Tiên in the 400m IM is a pacing case: a butterfly leg opened too fast takes from the breaststroke leg exactly the time the breaststroke leg cannot give back. Both are problems with solutions written in splits, and both are being judged on a single unknown.

Nguyễn Thị Ánh Viên is the only exception with a time series long enough to regress — more than a decade at the top, more than twenty SEA Games golds. Even that series is thin at the split layer. We have a beautiful performance curve and no data explaining why it is beautiful. A curve without an explanatory variable is an untested hypothesis, not a lesson.

In my own database I once regressed Southeast Asian swimmers' seasonal bests against three variables: sessions per week, long-course races per year, and the timing of the taper relative to the target meet. With a large enough sample, the second variable tends to explain more variance than the first. I keep the confidence label at medium, because the sample is still thin and because Southeast Asian swimming has too few high-quality long-course meets to generate clean variance. I do not believe in luck; I believe in the error bar. A swimmer who goes two seconds faster at one meet may have improved, may have found a faster pool, or may have raced a weaker field. Those three only separate with layer-two and layer-five data.

That is why I refuse the word miracle for a breakthrough. A miracle is just a data point that has not been regressed yet. In 2026 I filed an internal analysis warning that a foreign striker would regress because his goal count sat far above his xG; the board dismissed it in favour of instinct in front of goal, and he scored twice in twelve matches. The lesson was not that I was right. The lesson was that when a phenomenon runs far beyond expectation, the work is to check what instrument is measuring it.

The counterintuitive part: more sensors is not the answer. A Vietnamese pool can buy an automatic timing system and still produce unusable data, because there is no shared protocol — whether a split is measured from the feet leaving the block or the head entering the water, whether 15m time is measured from the wall edge or the 15m mark, at what moment water temperature is logged. If those three questions have no common answer across meets, each meet generates its own format and no two meets can be joined. The problem is that the data does not speak one language.

The most neglected variable is the calendar. A Vietnamese swimmer typically has two peaks a year — SEA Games and either the Asian Games or Olympic qualification — and the two peaks are unevenly spaced. The performance curve is therefore not a training curve but a sawtooth produced by scheduling. Read a sawtooth as a training curve and you draw the wrong conclusion about both coach and athlete. The correlation between training more and swimming faster is close to meaningless in this kind of data, while the correlation between racing more long course and swimming faster is surprisingly stable.

One more blind spot: results at school and student level, and in age-capped events, tend to predict senior performance poorly, because they measure biological maturity faster than they measure potential. A fifteen-year-old regional champion may simply be the earliest maturer in the lane that day. Weighting selection towards that medal group is a bet on a variable that does not hold. When that swimmer stalls at eighteen, the system calls it a plateau; the data calls it regression to the mean — something that should have been forecast from the start.

The signal to watch in the next cycle is specific: whether the result sheets from the 2026 Asian Games in Aichi–Nagoya, opening 19 September 2026, arrive with 50m splits attached. If they do, a Vietnamese swimmer sits on the same vertical axis as the rest of the continent for the first time, and we can read where the time is lost. If they do not, the limits of Vietnamese swimming will keep being explained by things nobody can measure.

Data only dies when we stop asking questions. An empty results sheet says more about the record keeper than about the swimmer. And the record keeper can always start again from the first row, with a splits column added.