When the Swimming Dataset Returns Empty: The Discipline of Reading Numbers

Last Tuesday I reopened a report file on my computer. Nineteen columns, nearly three hundred rows. Start time, stroke rate, turn count, high-speed distance. Every one of them returned the same line of text: insufficient information to assess. The fault came neither from the spreadsheet nor from the server connection. The source file had been pulled from an internal fitness test at an aquatic training centre, but the extraction stage at the head of the pipeline had finished with an empty result. Every analytical layer behind it therefore froze in place.

The first reflex of someone who has worked with data for twelve years is to fill the gaps. The trade teaches you that. Spectators want a conclusion, the coaching staff wants a number, and an empty table looks like a personal failure. I have filled whitespace before with interpolation, with last season's averages, with instincts politely renamed expert estimates. This time I sat still.

In swimming, data does not come from a single source. It is three layers stacked on one another. The first is electronic timing from the touchpad, producing the driest and most trustworthy figure. The second is imagery, where I count stroke rate, measure each stroke length, dissect the start and the underwater phase. The third is forms and athlete files, recording age, arm span, injury history and weekly training load. These three layers lock together like three door bolts. Lose the first, and the last two become decorative numbers.

In Vietnam, most aquatic training centres still run the third layer by hand. A few have electronic touchpads, a few have analysis cameras, but rarely does any site synchronise all three inside one system. So a single failure at the extraction stage erases weeks of observation, and erases the very ability to compare athletes against one another.

Every shock has its own probability. We call it a shock only when we have not checked the table yet.

But an empty table has no probability at all. That is the fundamental distinction outsiders usually miss. When I opened last week's file and saw every column blank, I was not facing an unexpected result. I was facing a silence. And silence in sports data is not information — it is the sign of a broken chain of operations.

Three kinds of breakage are common. First, a failed extraction: the software reads the source file but maps fields incorrectly, so every value drains into an empty cell. Second, mislabelling: the data exists but is tagged wrongly, so an endurance test gets recorded as a recovery swim. Third, a source that never existed: people assume a measurement file is somewhere, when in fact the measurement was never performed. All three leave the same trace on screen, but the causes and the fixes are entirely different.

This is why I write the conclusion insufficient information instead of filling in a pretty number. In my trade, the most honest answer is sometimes the empty one. It does not earn many reads. It does not please someone waiting for a forecast. But it keeps the rest of the model from being poisoned by a fabricated assumption.

Once I nearly made a bigger mistake. In 2026, reviewing positional data for twenty-nine athletes at a club in Saigon, I saw two curves running parallel so neatly it was suspicious: high-speed distance rising and muscle injury frequency rising. Twenty percent. A relationship as perfect as a textbook. The natural reflex is to declare causation, that high-speed running causes muscle injury.

I almost wrote that sentence. Then I reconstructed the third variable and found both curves were pulled by the same thing: a spike in training load after the pandemic break. High distance and injury did not create each other. They were siblings with the same parent. The physical mechanism linking the two variables — accumulated fibre overload when volume grows faster than adaptive capacity — was the only thing I could defend with an argument. Had I looked only at correlation, I would have advised the club to cut high-speed running, possibly weakening the team while the real problem sat in the training calendar.

Correlation is an invitation. Causation is a verdict. Anyone reading numbers must know which document they are signing.

Based on my experience tracking matches and fitness tests, the most dangerous thing is not missing data. Missing data incriminates itself. One glance tells you something is wrong. The truly dangerous thing is a flawless table, so complete that nobody bothers to check it again.

A report with all its sections filled, a clear headline, a tidy conclusion, can convince people the analysis is finished. But if every data cell inside is the product of carefully presented guesswork, then that handsome frame is only a cover. The real content sits where nobody is looking.

I have seen this in a report sent to viewers after a major games. Every metric was present, every judgment flowed, and not one line explained that most of the underlying figures had never been cross-checked. Readers understood it as fact. The writer understood it as estimate. The gap between those two understandings is where the danger lives.

When the stands fall silent, home advantage collapses into a number near zero. The same holds for data. When the source falls silent, every model collapses to an empty value, even if the outer shell still looks dignified.

Ordinary people look at the goal to understand the match. I look at the match to understand the years.

In swimming, a sprint miracle over the final twenty metres is not the magic of one evening. It is the convergence point of weeks of fuelling, of a pacing plan suited to each phase of the cycle, of reading the opponent to choose the moment to accelerate. To prove that, one swim is not enough. We need repeated data. We need the same athlete, with the same pacing pattern, across three or four different starts, to know it is skill rather than luck. And when that data is empty, I can neither praise nor criticise. I can only say: not enough to conclude.

At one SEA Games, Nguyen Thi Anh Vien won eight gold medals — a figure officially recorded in the organisers' files. That was not a lucky swim. It was eight repetitions of one standard. The lesson lies not in the medal count, but in the fact that each repetition came with data attached: time, split structure, direct opponent. The table turns a miracle into a verifiable sample.

A contract is not a signature, it is a hypothesis signed. And an empty dataset is the same. It is a hypothesis not yet written, waiting for someone to return to the head of the pipeline and fix the extraction.

What I want to leave behind is not a forecast for next month. I want to leave behind a habit. Fans, coaches and journalists alike should start asking one simple question whenever they meet a beautiful number: what was it measured with, and if we remove it, does the conclusion still stand. For me, keeping every number traceable is the only way to tell the truth.