The Empty Spreadsheet: The 'Not Enough Data' Discipline in Swimming Analysis

The spreadsheet opened on screen close to midnight. Twenty-three rows. The 50m split column was empty. The stroke-rate column was empty. The push-off count column was empty. The lane-condition note column was empty too. I looked at it for about four minutes, then closed it. In this trade, an empty sheet is harder to stomach than a wrong one. A wrong sheet can be fixed, and it can be argued with. An empty sheet allows exactly one choice: no conclusion. And that is the hardest choice of all, because everyone around me is writing. I started tracking swimming data in 2026, when I took the swimming beat at Thanh Nien Bao. Back then I thought the job belonged to the stopwatch. That is partly true. A stopwatch tells you who touched first; it does not tell you why. A swimming session leaves behind a great deal that can be measured. Reaction time off the blocks. Underwater distance after the start, measured in metres and in seconds saved against swimming on the surface. Stroke rate, the number of arm cycles per minute. Distance per stroke, the metres covered in one cycle. Turn efficiency, measured from the 5m mark before the wall to the 5m mark after it. And the fade rate over the last 50m, something the official results sheet never prints. But to have those numbers, the organiser has to publish 50m splits. Most domestic meets publish only final times. That means, for most lanes, I have one line and nothing else. That is why I have kept one rule since I was 17: a conclusion needs at least three independent data sources from three different contexts. The official results sheet. An independent video timing. And the athlete's long-term competition record. Those three do not stop at cross-checking a number; they test what the number means. That rule came out of a football match, not a lane. August 2026, V-League round 18, Hanoi FC hosting FLC Thanh Hoa at Hang Day Stadium. I was 16, just starting to read numbers from the VPF site. Hanoi held 68% of possession and took 21 shots. Thanh Hoa took 9 and won 2-1, through two counter-attacks by Uche Iheruome. I remember feeling cheated. The raw numbers I had read before kick-off said one thing; the score said another. Only later did I understand: those numbers were not lying, they were answering a different question from the one I thought I was asking. Possession is a beautiful lie; the score is the blinding truth. In swimming, the equivalent of possession is leading at the 100m mark. I have watched lanes where the swimmer led from the first 50m, held the highest stroke rate through 100m, and finished third. The results sheet records third. With 50m splits, you would see that the fade over the final 50m was the cause, and the high stroke rate in the middle of the race was the earliest warning. Leading is not controlling; leading on a high stroke rate is borrowing. The Hang Day shock taught me: strong teams know fear too. The numbers forgot to record it. June 2026, the World Cup in Russia. I was 17 and had built my own dataset for the tournament. Before Germany played South Korea in the group stage, I compared two metrics: Germany averaged a PPDA of 12.1, allowing opponents to pass fairly freely; South Korea sat at 9.1, a far higher press. I wrote a tweet warning that Germany could go out, with a chart comparing the two teams' xG. South Korea won 2-0. Predicting Germany's exit was not courage. It was a number that could not find a place in the crowd's model. Carry that principle into swimming and PPDA has no direct equivalent. The closest thing is press density in a lane: how many times a swimmer forces an opponent to lift stroke rate over the middle 50m. I once hand-counted that metric for a 200m medley final and found it predicted better than the qualifying ranking. But I could only count it for six lanes with clean video. Six lanes is not a sample. In 2026, when the Bundesliga returned to empty stadiums, I was 19 and realised I was holding a natural experiment. I collected 72 Bundesliga matches from 2026/19 with crowds and 26 behind closed doors in 2026/20. Home win rate fell from 44.4% to 36.2%; the away side's average points rose by 0.3. Empty stadiums do not erase football. They only erase one layer of the game's costume. Swimming has a crowd variable too; it just sits somewhere else. Roaring over the final 15m does not make anyone swim faster physiologically. It changes the decision: whether to lift stroke rate over the last two arm cycles. At meets without crowds, personal bests still fall, so many people conclude the crowd does not matter. I do not. A personal best is a speed variable; the decision at 15m is an outcome variable. Different things. Then came June 2026, Euro 2026. I was already a betting-analysis contributor and too confident in my model. I stated Denmark would exit early, because their pre-tournament xG average was only 0.9. In the opening match against Finland, Christian Eriksen collapsed on the pitch. Denmark beat Russia 4-1 and reached the semi-finals. I lost 12 million dong on a parlay. I deleted the unquantifiable variable from my model and the model demanded an explanation from me. Since then, every analysis I write carries its own section: unquantifiable variables. Shoulder injury. A female athlete's menstrual cycle. Two events scheduled in one session. First-final psychology. I assign a risk adjustment factor between 0.8 and 1.2 and remove the word certain from my vocabulary. And when the sheet has 23 empty rows, I write the Method section and stop. An analyst's duty is not to be right. It is to say what the data wants said. Every match sends a signal. The analyst does not decode it; the analyst listens. Here is where I have to contradict my own trade. The three-source ritual easily becomes three echoes. I have seen analyses cite three sources, all copied from a single press release. That is one source multiplied by three. In swimming, the common version is quoting three results sheets that all came from the same timing system. Nothing has been verified. The second source must come from an independent timekeeper; the third from long-term competition history. Otherwise I am just repeating. The second pressure comes from the opposite direction: this trade rewards people who conclude. Nobody shares an article that ends with not enough data. I once had an editor ask me point-blank what I was writing that had no prediction. I said I was writing about why no prediction was yet possible. That piece had the lowest readership of the month. There is a deeper trap, and it sits inside quantitative thinking itself. Not everything measurable matters, and not everything that matters is measurable. Stroke rate is measurable; the decision to hold rate over the last two cycles is not. If I only write about what carries numbers, I will describe an athlete built from whatever is easiest to capture, and skip the rest. Vietnamese swimming is at a stage where data is not yet treated as infrastructure. Domestic meets publish final times, rarely splits. Without splits there is no tactical analysis, only a leaderboard. And anyone can read a leaderboard, so it adds no value for the writer. The paradox sits here: the thinner the data, the more easily writers slide toward feeling, because feeling is always available. Three signals are worth watching over the next two years. One is whether national swimming meets publish 50m splits as a default part of results, rather than as an internal document. At the same time, I am watching for an independent video-timing group, strong enough to form a second source genuinely separate from the organiser. And the hardest signal: whether an editor will run a piece ending in not enough data without demanding one more line of prediction. If all three happen, Vietnamese swimming will have what football has had for years: a data layer thick enough to argue with in numbers. If only the last one happens, that is enough for me to reopen that 23-row sheet and write something else.