Vietnamese Swimming and the Blank Layer of Data

I opened the data sheet on my screen at 6:40 in the morning, before the lane was even filled. It was a twenty-year habit: arrive early, build the frame, wait for the numbers. Later that afternoon, after Nguyen Huy Hoang touched the wall in the 1500m freestyle, I looked at my notes file. The time column was there. The athlete's name was there. Everything else was blank. Not a single 100m split. No stroke rate. No turn time. No average meters per stroke cycle. The data file was almost empty, and it was empty in the exact sense that keeps me awake at night: the problem is not that we swim slowly, but that we have nothing to read. Over many years as a sports data analyst, I have grown used to a football match leaving behind thousands of data points: positions, touches, distance covered, the coordinates of every pass. Swimming is different. A 1500m freestyle race can last nearly fifteen minutes, but if the organizers only record the final time, the entire story inside it disappears. We receive a single number, and from that number we build an entire narrative about talent, about character, about progress. But that narrative is built on an almost empty foundation. I call this phenomenon the blank extraction layer. It is the first layer of any analytical process: the stage of collecting and deconstructing raw information before any conclusion is drawn. When this layer is empty, every layer above it — technical evaluation, performance comparison, future forecasting — becomes a building constructed on sand. You can build as many floors as you like, but they will collapse when someone asks the right question. There is one thing I have learned over the years: strong sporting nations are not strong because they have better athletes at the present moment, but because they have had better measurement systems for a long time before that. Japan did not produce world-class swimmers by accident. They measured from primary school. Singapore, a country much smaller than Vietnam, has also built a continuous swimming data chain over many decades. Thailand has recently begun systematic recording at the high-school level. Vietnam has talent. But we tend to discover talent only when it has already reached the summit, instead of tracking it as it crawls up from the base of the slope. In this article, I want to dissect a question that seems simple but that few people bother to ask: when an athlete finishes a race, how much do we actually know? And how is this data gap quietly shaping the fate of Vietnamese swimming? To answer, I have to recount how I work. Every time I follow a swimming meet, I sit with three parallel sources: the official results sheet from the organizers, cameras placed at both ends of the pool to capture turn times, and a personal notes file marking each stroke cycle. These three sources are cross-checked against one another, and only when all three agree do I trust the number. That is a principle I set for myself long ago, after witnessing too many times a wrong number spreading everywhere because nobody bothered to verify it. But the three-source principle only works if there are three sources to cross-check. At most domestic swimming meets, I have only one source, and that source is usually just the final time displayed on the electronic board. No splits. No stroke rate. No reaction time off the start. No count of valid turns. A fifteen-minute race is compressed into two digits after the decimal point. I once treated models as scripture. Now they are only a compass — but without them, we are lost. And a compass only works when we know where we are standing. If we do not know where we are standing, then every direction seems correct. Let us take a concrete example. Suppose an athlete swims 1500m freestyle in 15 minutes 20 seconds, six seconds better than their previous personal best. That is a beautiful number. The press will call it a dramatic leap forward. But the technical question lies elsewhere: where did that improvement come from? From increasing stroke rate, or from increasing meters per stroke cycle? From the start and underwater phase, or from turning ability? From holding pace evenly throughout the race, or from a burst in the final 200m? Without 100m splits, we cannot answer any of those questions. We only know the result, not the cause. And when we do not know the cause, we make the classic mistake: we attribute improvement to the most visible thing. If the athlete raised their stroke rate and the time improved, we conclude that stroke rate was the decisive factor. But it may be that the higher stroke rate came with a drop in meters per stroke — meaning the movement became less efficient — and the better time came only from an improved start. We have misread the entire story, and then we teach that misreading to the next generation. I remember a season when I followed a group of young athletes. The coach was pleased because their 200m medley times dropped steadily month after month. But when I reconstructed the splits, a different picture emerged: the first 50m got notably faster, the third 50m got slower, and the final 50m only held steady. The overall time was better, but the race structure had worsened. The athlete was relying on an explosive start to compensate for a decline in the middle. That is a kind of improvement that does not last, and it usually collapses at the most important moment. Have you noticed that strong swimming nations tend to talk in splits, while we tend to talk in placements? That is the sign of two different data cultures. One side believes in the process, the other believes in the result. In truth, the problem is not only a lack of equipment. Vietnam has enough cameras, enough electronic boards, enough software if it wanted them. The problem lies in the fact that nobody treats data as an asset. We treat medals as assets. We treat records as assets. We do not treat the recording of a twelve-year-old's stroke rate as an asset. But that is precisely the thing that decides who will stand on the podium ten years from now. In swimming, there is a group of metrics I call the hidden metrics — numbers that do not appear on the scoreboard but determine the scoreboard. They include: reaction time off the start, underwater distance after the start, meters per stroke cycle, stroke rate per minute, time and number of cycles at each turn, and split structure. These six metric groups, combined, explain most of the difference between an athlete who swims 15:40 and one who swims 15:20. Without them, we are only guessing. I once tried to apply Western metrics to Vietnamese data and failed. Not because the metrics were wrong, but because the foundation for comparison was not the same. An American athlete grows up in a dense competition system, a meet every month, continuous data. A Vietnamese athlete may compete in only a few meets a year, with long gaps between them. That rhythm completely changes how the body adapts. When I took a model from a dense international calendar and applied it to sparse domestic data, the model produced systematically skewed forecasts. That is the lesson I paid a price to learn: a good model is not the most complex one, but the one suited to the actual data rhythm you have. A simple model calibrated by local reality will beat a sophisticated model borrowed from elsewhere. Here I must speak to an uncomfortable truth. Vietnamese swimming is at a stage I call the pre-data stage. We are still debating whether to record at all, while the leading nations have finished recording and are now analyzing. The gap is not in the pool. It is in the computer. But I am not writing this article to be pessimistic. When the field is empty, every model collapses. I rebuild from the charred remains of the data. And Vietnamese swimming, with its blank data layer, is actually in a strangely advantageous position: we have lost nothing, because we never had anything to lose. We can build from scratch, the right way, without having to demolish old systems. There is a pressure no one sees, but every athlete feels. I name it: the measurement gap. It is what keeps an athlete from knowing whether they are progressing or regressing, keeps a coach adjusting training plans by feel rather than by data, keeps an entire generation of children swimming while nobody truly sees how they are swimming. To fill this gap, expensive technology is not needed. I tried and succeeded with three cheap tools: a phone on a tripod shooting in slow motion at one end of the pool, a simple spreadsheet recording 50m splits, and a notebook counting stroke rate at three fixed points in the race. These three tools, used correctly, gave me about eighty percent of the information I needed, at almost zero cost. The problem is that these three tools demand something more expensive than money: discipline. You have to record every training session, every race, every month, for years, even when there is no major meet. That discipline is what we lack, and it is also the core reason our extraction layer is empty. Here I must be careful. The three-source metric is not a guarantee of truth. Sometimes I had three sources that agreed and still drew the wrong conclusion, because all three sources viewed the race from the same angle. The agreement of three sources does not create truth; it only makes error harder to catch. I must always remind myself of that, because my nature loves numbers that agree. And this is where I want to challenge the analytical community itself. We easily fall into the illusion that more data automatically yields better answers. Not true. More data, from the same viewpoint, only creates false confidence. I once built a model with hundreds of variables for a group of swimmers, and it failed disastrously at the very moment that mattered, because I had disguised a single viewpoint with countless variables. If the extraction layer is empty, then every layer above it is fabricated, no matter how complex it looks. This is something I have learned many times, and it still irritates me whenever I repeat it. Numbers do not lie, but people always find ways to lie with numbers. When there are no numbers at all, people lie with emptiness. Emptiness is harder to question than any wrong figure, because nobody goes to check a void. There is an example I often tell in workshops with young coaches. An athlete swam the 100m freestyle and achieved a better time than expected. Everyone praised his competitive spirit. I was curious, so I reconstructed his start with a slow-motion phone. It turned out he reacted off the start a tenth of a second slower than usual, but compensated with an underwater phase that was deeper and two meters longer than normal. That underwater phase saved more energy and speed than the start itself. But nobody saw the underwater phase, because it happened below the surface. The whole stand praised his spirit, while the technical truth lay somewhere entirely different. This is the essence of what I call hunting anomalies. You are not looking for a striking number. You are looking for a number that surfaces where there should be nothing, a small phenomenon hidden by the general narrative. Swimming is full of such anomalies, and they usually vanish because nobody records them. I once witnessed an entire trend being missed for lack of data. A group of young athletes in a southern province had an unusual way of turning — they approached the wall at an angle rather than perpendicular. In some cases, this method saved time, though it demanded higher technique. But because nobody recorded turn times, the trend disappeared when the athletes grew up and were retrained in the traditional style. We may have lost a valuable technical direction, simply because there was no cell in the spreadsheet to record it. I do not want to blame any individual. The problem is the system. When a federation rewards only the final result, nobody has an incentive to record the process. The economics are simple: you measure what you reward, and you get what you measure. If we measure only placement, we will get only placement, and everything that creates placement will be abandoned. This is the counter-intuitive point I want to stress. People often think that a lack of data is a problem of weak sporting nations, and that having data is a sign of success. But the reality is partly the opposite. Precisely because we have little data, we can easily achieve large gains if we start recording correctly, because the marginal benefit of the first information is the greatest. Water poured into a shallow lake rises faster than water poured into a full one. However, I must warn of another trap. Many people, when they first get data, rush into correlation and mistake it for causation. They see an athlete raise stroke rate and performance improve, then conclude that raising stroke rate is the key. But correlation is not causation, and in swimming, the relationship between stroke rate, meters per stroke, and speed is a complex equation, not a one-way arrow. I fell into this trap. One season, I convinced a group of coaches to raise stroke rate for their athletes, based on a correlation I found in a small sample. The result was flat performance for three months, and several athletes lost their foundational technique. I had forgotten a basic principle: a small sample can produce a perfect correlation that is entirely meaningless. I read too much into too little data, exactly the mistake I always warn others about. Since then, I have set a rule: state the sample size in every conclusion. If the sample is small, I call it a hypothesis, not a conclusion. A hypothesis is allowed to be wrong, and must be tested with new data. A conclusion is not. The difference between these two words is the difference between an honest analyst and a seller of predictions. This brings me to a larger question about the nature of swimming. Swimming is the sport of absurdity. You can swim beautifully, efficiently, with perfect technique, and still lose to someone who swims worse but has a better physique or a stronger mentality on competition day. Swimming does not forgive models. It reminds us that the human body is not an equation. But precisely for that reason, I learned to count the absurdity as well. I added a variable for what the model cannot explain. I do not try to remove the residual from the model; I record it, name it, and track it as its own metric. Sometimes that residual is the best predictor of the next race. This is what separates a data analyst from someone who reads a scoreboard. The scoreboard reader believes the final number is the truth. The analyst knows the final number is only the shadow of a much more complex truth. So what are the signals for the next cycle? I believe there are three signals to watch in the coming years in Vietnamese swimming. First, the emergence of young coaches who begin recording data systematically, even if only in a notebook. When I see a coach in a remote province writing down their students' splits after every session, I know something is changing. That is a small signal but more important than any medal. Second, the emergence of pools with camera systems and automatic recording software at the grassroots level, not only in big centers. When data becomes cheap and accessible, it will spread down to the base of the slope, where the real talent lies. Third, and most importantly, the emergence of a generation of parents and athletes who understand that performance is the result of a process, not a miracle. When a parent asks a coach about their child's stroke rate instead of about placement, that is a sign of a data culture taking shape. I do not know exactly when these signals will mature. It may be five years, it may be ten. But I am certain of one thing: the data gap will not fill itself. It fills only when someone decides to record the first thing. Back to that morning, when I looked at the empty data file on my screen. I could have closed the laptop and gone home. Instead, I opened a new file, named it with the date, and began rebuilding the recording frame from scratch. I wrote the first line: 50m splits. The second line: stroke rate. The third line: meters per stroke cycle. Then I sat and waited for the next competition. Maybe that day I recorded nothing worthwhile. Maybe in five years, those notes will still be just a small file on the personal computer of a middle-aged man in Binh Duong. But I believe that every great data system in the world began with one line of notes by someone, somewhere, on a morning no one noticed. When the field is empty, every model collapses. I rebuild from the charred remains of the data. And sometimes, those charred remains begin with a completely empty file, and a person who decides to fill in the first cell. The question I leave for those working in Vietnamese swimming is not whether we have talent. The question is: are we recording that talent, every athlete, every session, every lap, or not? Because if we are not, we will forever know only the result, and forever guess about the cause.