The Swimming Lane and the Empty Data Problem: When the Split Board Is Blank, a Record Is Only a Hypothesis

2:47 in the morning. The spreadsheet in front of me was blank, and the coffee beside me had gone cold long ago. The national swimming championship had ended just a few hours earlier; the results were published, the numbers dancing on the electronic board. But when I scrolled down to the split section — where the first-50m time, the back-50m time, stroke rate and stroke frequency ought to be — there was only white space.

In swimming, a final number tells you nothing. It is like looking at the finish line of a marathon runner without knowing which stretches of road he crossed. I learned this after years: data never lies, but it knows how to hide itself. And the most dangerous thing in sports analysis is not a wrong number, but a missing one.

That night I did not write. There was nothing to write when there was nothing to compare.

Swimming is a sport of numbers, but not every number is transparent. In the men's 100m freestyle, the current world record is 46.40 seconds, set by Pan Zhanle at Paris 2026. In the women's 400m freestyle, Katie Ledecky holds the record at 3:56.46 from Rio 2026. Those are numbers everyone knows. But what analysts actually need lies deeper: the split chain.

A 200m freestyle swimmer does not swim four 50m laps the same way. He distributes his energy along a curve. The first lap is fast to build momentum, the second holds rhythm, the third is where character is tested, and the fourth is where the body pays the price. If I only know the final time, I cannot distinguish someone finishing with a sprint from someone finishing by clinging on in desperation. The same number, two completely different stories. People look at the price board; I look at the curve. Many results die before they are published, simply because nobody bothers to read the speed of each segment.

That is why I always begin any analysis by checking whether the data is labelled correctly. A result swum in a 25m pool cannot be compared directly with one in a 50m pool. The two environments differ in the number of turns, in the push-off from the wall, in the inertia of the water. A 200m swimmer in a short course can be several seconds faster than in a long course, but that does not mean he has improved. It is a consequence of the geometry of the lane, not of ability.

In Vietnam, this problem is even more serious. I have followed domestic meets for years, and what I found is that split data almost never exists systematically. A few meets publish it; most do not. Organisers measure the finish time, print it on the board, and stop there. The 50m and 100m segments — which any developed swimming nation treats as a minimum standard — are left blank.

When Nguyen Huy Hoang rose to prominence with medals at the SEA Games, the first thing I wanted to know was not how fast he swam, but how he distributed his energy. In the 1500m, the gap between segments changes very little at the elite level. The champion is not the one who swims the first segment fastest, but the one who keeps the deviation between segments smallest. Looking at total time, we honour luck. Looking at the stability of the splits, we see ability.

I remember the first time I applied this thinking seriously. COVID closed the stadiums, and I reopened the V-League directory. No league is meaningless. When the swimming lanes also fell silent because of the pandemic, I started rebuilding the historical dataset of several domestic swimming seasons. I tracked each athlete, recorded their acceleration times, estimated stroke rates from video, and built a certainty scale — where I set clear thresholds for what counts as signal and what is merely noise.

In swimming, the real signal lies in the stability of the segments, not in the peak of a single swim. A single personal best can be luck. Several seasons of keeping split deviation below a threshold is ability.

Another metric I follow closely is the trade-off between stroke rate and distance per stroke. When a swimmer raises stroke rate, distance per stroke usually falls. In a short race, a higher rate can create an appearance of speed. But if distance per stroke drops too fast, real efficiency declines. That is a form of wasted swimming — sweat and water that buy no speed. People look at pace; I look at efficiency per stroke. A champion's line-up is not in the wallet, but in how time is compressed into an index.

I once analysed a young female swimmer with a leap in times at the age of 15. The media called it innate talent. But when I placed her data on the age-development curve, I saw something else: her body was in puberty. In women's swimming, this is a sensitive phase. Changes in body structure, in the ratio of height to stroke length, can temporarily slow progress — or create an unsustainable leap. Ignoring this variable skews every forecast. Athletes do not collapse overnight. They collapse when the indices stop connecting with one another. Form does not collapse in a single swim; it collapses when scattered signals no longer line up.

Injury is another variable that pure time data cannot capture. The shoulder of a freestyle and butterfly swimmer, the knee of a breaststroker — these are the structural weak points of the sport. A swimmer can hold the same time for months, while acceleration frequency declines. The final number does not change. But the curve has changed. Someone who reads only results sees nothing. Someone who reads signals sees ahead.

And here is the part I want to save for those who believe data is always there, simply waiting to be mined.

The absence of data is itself a signal. When a federation does not publish splits, when a meet does not label long course or short course, that carries no neutrality. It is a message about the maturity of the system. I have learned to read those blank spaces the way I read a number. An empty dataset does not make me conclude wrongly. It makes me know that I cannot yet conclude.

There is a temptation I always remind myself to avoid: mistaking correlation for causation. A swimmer with a fast split who finishes first in one race does not mean the fast split produced the win. In many cases, the victory comes from an opponent distributing energy badly, from a good turn, or simply from a familiar pool. Assigning the cause to a single variable while ignoring all others is storytelling, not analysis.

The market angle is also worth raising. In sports, the commercial value of a swimmer is usually priced by narrative, not by splits. A medal at a major meet can multiply sponsorship value many times over, while a stable split chain across seasons is not priced accordingly. That is a market anomaly — where money flows with the headline, not with the data.

Luck is something I do not have. I have probability and thick enough data. A single elite performance is not enough to build a model. It is only enough to build a headline. A model appears only when I have enough sample to separate noise from signal, enough seasons to distinguish trend from short-term fluctuation.

That is also why I am always wary of milestones tied to the swimsuit era. The years when swimsuit technology exploded created a tier of records that cannot be compared directly with the tier of the later period. A record set in that era is not wrong, but it belongs to a different frame of reference. Ignoring the era variable is the fastest way to make a meaningless comparison.

The signal for the next cycle, then, lies in three points.

Standardising split data is the prerequisite. In many developed swimming nations, splits are published automatically for every round. That is the minimum standard for serious analysis. When this data is published fully and labelled by pool type, forecasting models for Vietnamese swimmers can be rebuilt reliably.

Alongside that is a calm view of the young generation during puberty. A leap in times at 14-15 is notable, but not enough to conclude. At least two or three consecutive seasons of stable split deviation are needed before speaking of real potential.

Behind it all is the physical signal hidden behind the time figure. Stroke rate, distance per stroke, acceleration frequency, and injury history — these are the pieces that finish time alone never tells.

That night, while the spreadsheet stayed empty, I did not write. I saved a note: data missing, wait for the next round. There are times when the most honest answer from a data analyst is not a number, but an acknowledged blank space.

The most frightening thing in this profession is not a wrong prediction. It is a correct prediction built on data that does not exist.

Data never lies. But it only speaks when there is enough data to speak.