The Empty Cell in Vietnamese Swimming Data

At a domestic swim meet, the results sheet was printed and handed to reporters minutes after the final ended. Each swimmer occupied one line: name, event, final time. The 50-metre split column was left blank. Nobody asked why. The press conference went ahead as usual, and by that evening at least three articles appeared with the same explanation: the swimmer started slowly and accelerated in the second half. That explanation sounded reasonable. It also had not one data point behind it. I remember that blank column more than any other tidy number in my career. Over years of working with swimming data, I learned that the frightening thing is not missing data. The frightening thing is empty cells being filled too quickly, before anyone has asked a question. I came to swimming before I came to football. In 2026, aged 23, I began work as a swimming reporter for a newspaper in Vietnam. The job was simple then: go to the pool, record the results, interview the coaches, file the story that night. The more I wrote, the more I realised I was retelling a story whose pieces I could not verify. The results sheet told you who finished first. It did not tell you how that person swam. Swimming is a sport of splits. A 200-metre freestyle race contains four 50-metre segments, and the entire tactical story — who conserved, who surged early, who finished on momentum — lives in how fast those four segments were. Electronic timing systems at national meets still record them. They exist in the hardware, in the software, in the organisers' export files. They are simply not published. The distance between "we have data" and "we have usable data" is the distance my profession lives inside. In football, where I work as a data consultant for clubs, the surface looks different and the substance is identical. Every training session generates tens of thousands of GPS points, yet if the synchronisation process is wrong, the numbers still arrive steadily and they still look good. When data is missing, people know it is missing. When data is wrong, they do not. That is why I started adding one column to every table I build: a confidence column. At the top level, with swimmers such as Nguyen Thi Anh Vien or Nguyen Huy Hoang, international competition data is reasonably complete because major meets publish splits to standard. The gap sits in the domestic circuit — where most young swimmers begin their careers, and where data most needs to be captured. In 2026, aged 25, I was the only female data consultant in the analysis room of a club in Nha Trang. During a match against a strong opponent on round 12, I miscalculated a striker's sprint distance — recording 1.2 km instead of 0.8 km. After the match, a senior analyst said in front of the group that someone sitting at a desk does not understand tactics. I did not argue. I took the club's 14,000 GPS samples from three months and checked every one. The result: three further systemic errors, all originating in the synchronisation software, all producing numbers that looked perfectly reasonable on screen. A small GPS deviation was enough to teach me: verification is everything. A three-round cross-check procedure was born from that and became the club's internal standard. Round one checks the device provenance. Round two checks against an independent source. Round three tests whether the number survives a biological logic check — for instance, a player cannot run 1.2 km above 25 km/h in the second half of a match in which his team held 30 percent of possession. That experience shaped how I read every data table, including a swimming results sheet. When a split column is blank, I do not fill it with inference. I mark it as undetermined and state that in the copy. In the summer of 2026, while supporting analysis for a sports channel during the World Cup in Russia, I collected expected-goals figures for all 64 matches. In the knockout rounds, the side that reached the final generated only 5.3 xG, while its four opponents combined for 7.1. That side scored 8 goals from 5.3 xG — roughly 51 percent overperformance. Croatia 2026 was not a miracle — it was xG written into history. My 2,000-word analysis then was one of the first xG pieces in Vietnamese, and it taught me something simple: numbers let you separate what repeats from what is merely noise. That method only holds when the input data is complete. In 2026, the domestic league was suspended from March to September because of the pandemic. Instead of waiting, I spent seven months building a recovery-index model based on GPS data from 365 players across three seasons. The model combined high-intensity running distance, acceleration counts and injury history. When the league returned, I projected that the three most intense pressing teams faced roughly a 23 percent rise in injury risk. The club I advised cut training load by 15 percent and lost no key players in that period. What I always publish alongside it: the model's limitations. A sample of 365 players, three seasons, one league — the model only holds inside that range. Every projection I make carries a probability band, never an absolute figure. I trust the number, but only after the number has cleared three rounds of checks. Three years later, a club in Ho Chi Minh City asked me to consult on a transfer window. They wanted to spend 500,000 USD on a striker from the Thai league. I analysed 19 of his matches. He had scored 18 goals but generated only 11.2 xG — a conversion rate of 31.4 percent, nearly double the league's 15 to 18 percent average. Seventy percent of those goals came from set pieces. I recommended against the signing. Management signed him anyway. He scored 4 goals in 20 matches and suffered two hamstring injuries. A year later, the club appointed me as an official consultant. I tell these three stories to make one point: every conclusion in my work begins by identifying which cells hold data, which do not, and which hold data that is not yet trustworthy. Data does not tell stories; it records everything so that I can tell them myself. Returning to swimming, I apply that principle to the blank split column. A swimming results sheet has three kinds of cells. The first holds verified numbers — final time, heats, competition date. The second holds unverified numbers — reaction time captured by sensors, for example, but never reconciled with video. The third is blank — splits, stroke counts, breathing rhythm. Of the three, the third is most dangerous, because it invites us to fill it. A swimmer who finishes half a second below expectation gets assigned a "slow start", because that is the popular explanation and it sounds right. Nobody can test it without splits. What worries me most in sports data work is not the blank cells. It is the tables that look flawless. A split column fully populated, neatly printed, with no source note, will go straight into an article and become truth within a single evening. Meanwhile, a blank cell left blank and properly annotated is usually read as unprofessional. The paradox is this: the most honest report is the one that dares to stay empty. I once assumed that collecting more data would automatically improve analysis. After years, I have found the opposite holds just as often. Vietnam does not lack sports data. Many competitions collect it and never use it, or use it to post online rather than to make decisions. A model is only as good as the question it was built to answer. More data without more questions just creates more room to be wrong. And there is one thing I should say plainly about my own trade. Data analysts are prone to a particular failure: when new data contradicts their model, they quietly defend the model. I have been in that position. My recovery-index model once misprojected a specific case, and my first instinct was to hunt for a justification. I forced myself to put that contradicting evidence into the article, in a place it could not be skipped. A model without a counter-evidence section is a model being advertised, not one being tested. The blank split column in that swimming results sheet will stay blank until somebody asks about it. When the data is finally published, we will discover that many of the "second-half surge" stories we keep telling were tidy inferences draped over a gap. The task is not to abandon the stories. It is to add one line beneath each story, stating what we know and what we do not. That is the only way a sporting nation learns from its own numbers.