One Missing Data Column Strips Every Swimming Result Table of Meaning

Nine analytical sections. Nine identical conclusions: not enough data to assess.

The report sat in the internal files of a sports data group. The professional framework for swimming had been built out in full: technical analysis, performance analysis, competition systems and selection mechanisms, the world map of events, rules and anti-doping governance, athlete career trajectories, risk profiling, public narrative and expectations, and industry ripple effects. The tables were neatly columned, the section headings clear, the glossary complete.

The data was empty.

No competition name. No athlete name. No competition time. No pool length. No date. Not even the title of the source article. If someone forced themselves to fill the page, the only way would be to invent a swimmer, invent a lane, invent a comeback. The group chose the opposite: stop, mark the gap precisely, and ask for the raw data to be supplied again.

Every shock has its own probability. We call it a shock only when we have not yet checked the table. But to check the table, a table must first exist.

In 2026 I started at the results desk of a sports newsroom in Saigon, covering swimming. Back then results arrived by telephone. The person standing at the poolside read out a name, a time, a placing. I wrote it on paper, then typed it up as a three-column table. More than twenty years later, looking at Vietnamese sports portals, the table still has three columns.

The equipment has changed enormously. Electronic timing, underwater cameras, starting-block sensors, software that reconstructs an entire race down to the hundredth of a second. What has not changed is how data is released to the public: name, time, medal. Those three things are enough to file a report. They are not enough to analyse anything.

The nine-section framework that group used is not one person's invention. It is the kind of structure that has settled in professional sports analytics, where every section must rest on a minimum data set. The technical section needs reaction time off the blocks, underwater distance after entry, per-50m splits, stroke rate per minute, distance per stroke cycle. The performance section needs time, pool length, competition name, date. The competition-system section needs to know whether this is an Olympic year, a post-Olympic year, or a mid-cycle build-up year. The power-map section needs a country, an event, a rival. The rules and doping section needs a concrete incident to classify. The career section needs the athlete's age and a performance series over time. The risk section needs a subject to rate. The narrative section needs a publishing source. The industry section needs a triggering event.

Remove any one of those, and the corresponding section drops into an empty state. All nine empty means there is nothing left to say. That report sat exactly in that state, and what is worth noting is that it made no attempt to paper over the gap with soft language.

The simplest amateur question anyone who has watched swimming will ask: why does the same swimmer, in the same event, go 1:58 one day and 2:01 another? The answer lies almost entirely in fields the published table does not contain.

The first field is pool length. A 50-metre pool and a 25-metre pool cannot be compared directly. Pool length is the most commonly omitted data field in swimming, and the one that renders every performance comparison meaningless when it is missing. A 25-metre pool doubles the number of turns, and every turn is a push-off. World records for the two pool types are ratified separately and kept on two separate lists. A results table that does not state pool length cannot be used for any comparison at all.

The second field is the timestamp relative to the high-tech swimsuit ban. From 2026 to 2026, full-body polyurethane suits were permitted. The records of the world aquatics federation show that the 2026 World Championships in Rome produced 43 world records. From 1 January 2026 the suits were banned. That means any performance recorded between 2026 and 2026 must be era-flagged before it is compared with anything. An era of technique dies when nobody reads its data table any more. The data table of the high-tech suit era is still intact, but it only means something if the reader knows what they are reading.

The third field is split structure. A 200-metre race has four 50-metre segments. If the first two are faster than the last two, the swimmer went out hard. If the reverse, they came home late. Two swimmers who touch the wall in the same time can be in completely different physical states, and their next meeting will look different. Without split data, every claim about conditioning is a guess dressed up in jargon.

The fourth field is the qualifying standard. The world aquatics federation sets two time levels for major championships: an A standard for direct entry, and a B standard that depends on quota allocation. A swimmer on a B standard is not guaranteed a place; one on an A standard is almost certain. Read a result without knowing which level it sits at, and you cannot tell whether it is progress or regression. Selection mechanisms also differ by country. The American model takes the two fastest on the day and ignores multi-year records. That model creates very high upset risk, and it is a variable that belongs in the model, not a sentimental story.

The fifth field is the least discussed, and the most painful in women's swimming: the puberty barrier. A female swimmer who peaks at 14 and plateaus at 17 has made no technical error at all. Her body changes, her height-to-armspan ratio changes, her buoyancy changes. If the data table records only times, with no age and no armspan series across years, the reader will draw the wrong conclusion about ability. This is where numerical analysis reaches its own limit, and the correct handling is to state the limit rather than pretend it does not exist.

Ordinary viewers look at a lane to understand a swimmer. I look at a swimmer to understand the years. A swimmer's peak window is far narrower than most people assume. Across many events, that window sits between twenty and twenty-seven. Outside it, any comparison between two points in time has to account for wear, and shoulder wear is the most common kind. Shoulder injury in swimming does not come from a single collision. It accumulates across thousands of strokes with a faulty range of motion, and it tends to surface exactly when training volume rises. Looking at a results table, a viewer sees only a time falling off. Looking at training data, an analyst sees high-speed swimming distance rise twenty per cent in the preceding three weeks.

In Vietnam, some cases show clearly what it means to read numbers as a series rather than as isolated points. Nguyễn Thị Ánh Viên won eight gold medals at the 2026 SEA Games in Singapore, according to the official medal table of the Games. That figure only means something next to the schedule: eight events in a single Games, each with heats and a final, plus the accumulated volume across consecutive days. That is a workload-allocation problem, and it can only be solved when heat data is separated from final data.

Nguyễn Huy Hoàng is a different case. He swims long distance, where split structure decides almost the entire result. In a 1500-metre race, nobody wins with the first two hundred metres. They win by holding the gap between their fastest and slowest segments inside a narrow band. Looking at a table that shows only the final time, a viewer cannot see that band. To see it, split data is required, and it is required across several competitions, not one.

The biggest temptation when data is empty is to fill it with inference. An analytics group might see a swimmer improve across three consecutive meets, see the coach change the training plan at the same time, and conclude the new plan produced the gain. Two series rising together do not prove that one generates the other. To claim the new plan produced the gain, you must point to a concrete mechanism: how high-speed volume changed, how rest ratios between repetitions changed, how the number of sessions per week changed. Without a mechanism, the correct phrasing is that the two things are related, and stop there.

The more serious risk is not with fans, but with empty data pushed into aggregate tables. An empty analysis, if tagged as valid, sits in the same file as real analyses. Months later someone runs a trend report over that file, and the empty record gets counted as a sample. The error multiplies at the aggregation layer, not at the collection layer. The cheapest guard is a check at the gate: if an extraction cannot produce even an article title and a source name, it must be flagged invalid and excluded from every count.

There is a portion of variance that cannot be quantified, and analysts should not pretend otherwise. Swimming is contested in pools packed with spectators, with noise rolling off the water, and in some events a swimmer used to an inside lane can clearly hear the stroke of the rival beside them. When the stands go silent, that home advantage collapses into a number close to zero. Some championships produce a readable performance frame; others carry so much emotional noise that every model has to widen its confidence interval. A good analyst is one who says so in advance, not one who explains it after being wrong.

If I had to propose a minimum data set for every swimming results table published in Vietnam, I would propose five fields: event name, time, pool length, competition name, date. Those five fields do not make a table much longer, but they turn a news line into comparable data, and turn an article into a link in a long-term tracking chain.

What is worth thinking about is that the hardest part is not collection. Electronic timing already exists at most major pools. The hard part is choosing not to put a number into a piece when the conditions are not met. That nine-section report ended with a refusal, and the refusal turned out to be the most valuable part of the whole document.