When the Swimming Lane Loses Its Data: A Pipeline Failure and a Lesson in Data Discipline

One morning in mid-August, I opened my tracking sheet for the national swimming championships and saw a completely empty column. Not the points column — the raw data column. Everything I needed to analyze a swimming performance was supposed to be there: splits for every 50m, stroke rate, turn times, underwater dolphin-kick efficiency. But this time, there was nothing. I stared at the screen for three minutes, then did what nine years in the profession had taught me to do: closed the spreadsheet and went to check the data pipeline.

This is not a story about a missed feat. This is a story about what happens when your analysis system returns an empty payload — and you have to choose between fabricating numbers or admitting you have nothing to say.

Context: When swimming becomes a data problem

I came to swimming from an unexpected angle. In 2026, I was sixteen, sitting in the stands at Hang Day Stadium watching Hanoi FC control 68% possession, fire 21 shots, and lose 2-1 to FLC Thanh Hoa through two Uche Iheruome counterattacks. I was shocked. I felt cheated by the very numbers I trusted.

From that day, I learned to read sports data the way an investigator reads a crime scene. I studied xG, PPDA on Understat and FBref. I built my own spreadsheets to track every match. And when I shifted to specializing in swimming, I realized this sport is actually a paradise of quantitative data: every lane is a continuous string of numbers, every wall touch a data point measurable to the hundredth of a second.

In swimming, you cannot say "this athlete swims beautifully" and call it analysis. You have to point to stroke rate, distance per stroke, reaction time off the blocks, number of underwater kicks after the dive, and how speed is distributed across four 50m laps. That is the language of this sport. Without it, you are commentating, not analyzing.

Core: The structure of a swimming performance

A competitive swim is divided into four separately measurable segments. First is the start and underwater phase — for butterfly, backstroke, and freestyle, this is where the biggest advantage is created because athletes move faster underwater than on the surface for the first 15m. In breaststroke, the rules allow only one underwater kick cycle, so this advantage is strictly limited.

Second is the middle phase — where stroke rate and distance per stroke decide everything. An athlete can increase stroke rate to swim faster in the short term, but the cost is reduced distance per stroke, leading to early energy depletion. This is where data analysis separates itself from visual perception: the eye sees an athlete "swimming hard," but the 50m splits show he is steadily slowing down.

Third is the turn phase. Every wall touch and rotation is an opportunity to lose or gain time. In freestyle, the flip turn allows a turn in roughly 0.6-0.8 seconds. In breaststroke and butterfly, the two-hand touch and rotation technique takes significantly longer.

Fourth is the finish — the final 5m, where the decision often comes down to whether the athlete maintains a legal wall-touch technique.

Contrarian angle: Correlation is not causation

This is where I have to state plainly something many swimming analysts avoid. When you see an athlete with a high stroke rate winning, you cannot conclude that high stroke rate is the cause of victory. That athlete may have a naturally long stroke, superior fitness, or simply a weaker opponent on the day.

I have made this mistake before. In 2026, in a Euro accumulator bet, I declared Denmark would exit early because their pre-tournament average xG was only 0.9 — among the weakest. In the opening match against Finland, Christian Eriksen suffered cardiac arrest on the pitch. Denmark played with emotional force, beat Russia 4-1, and reached the semi-finals. I lost 12 million dong. The lesson: my model had no non-quantifiable variable.

In swimming, that variable exists too. An athlete can have every best-in-class metric and still fail because of a shoulder injury (swimmer's shoulder), because of psychological pressure in a final, or simply because the pool has different water turbulence than where he trains.

Takeaway: When data does not speak, do not force it to

Back to that August morning. I checked my data pipeline and found the problem: the extraction step from the source had silently failed. The returned payload had full structure — every field existed — but all values were empty.

I had two choices. One, fabricate numbers: assign athlete A a stroke rate of 42 strokes per minute, give athlete B a first-50m split of 24.3 seconds. No one could verify, and the article would look perfect. Two, admit I had nothing to analyze, and write about that very failure.

I chose the second. Not out of morality — but out of discipline. An analyst who fabricates data once will fabricate data forever, and eventually the fabricated number will collide with reality.

In swimming, this is even more serious. Because swimming is a sport where the gap between top athletes is measured in hundredths of a second. A small error in split analysis can lead to a completely wrong conclusion about who will win. And in a sport where every movement can be measured, the inability to measure is itself the strongest signal.

I rewrote my process. Added a mandatory check: if the payload is empty, stop. No analysis. No speculation. No filling gaps with imagination.

The truth is, sometimes data says nothing at all. And the duty of the analyst is not to force it to speak. It is to listen to the silence as well.