When the Data Sheet Returns N/A: Verification Discipline in Swimming Analysis

2:17 in the morning. The spreadsheet in front of me has four tabs, and three of them are blank. I was reconstructing a 200 metre individual medley swim. What I needed was very specific: reaction time off the start, the first 15 metres underwater, average stroke rate over the third 50, and the final turn. Four variables, enough to build a force distribution curve. The split column came back blank. The reaction time tab came back blank. The technical notes tab came back blank. Only the fourth tab, the one I wrote myself, had words on it. Ten years in this trade taught me one thing: the most dangerous moment is not when a model predicts wrong. The most dangerous moment is when it says nothing at all, and the person at the screen has to choose between silence and invention. In 2026, aged sixteen, I watched V-League round 18 at Hang Day Stadium. Before kick-off, Ha Noi controlled 68 per cent of possession and fired 21 shots; FLC Thanh Hoa managed 9. I believed the ratio. Thanh Hoa won 2-1 through two Uche Iheruome counter-attacks. That night I learned raw metrics can lie. The Hang Day shock taught me that strong teams also know fear. The number forgot to record that. It took four more years to understand the reverse side of that lesson. If raw metrics deceive by saying too much, emptiness deceives by saying too little. The second deception is far more dangerous, because it invites the analyst to fill the gap by hand with something that sounds plausible. That night I did not fill it. I built a table of nine categories, each marked with a single sentence: insufficient information, cannot assess. Every match sends a signal. The analyst does not decode it, but listens to it. A blank dataset sends a signal too, and it is the hardest one to hear. A little architecture is needed here so the reader knows what I am actually talking about. A serious sports analysis pipeline runs in two stages. The extraction stage works on the source text: it must return information points, core viewpoints, named entities, time sensitivity and source quality. The analysis stage, where I sit, asks the heavier questions: technique, performance, competition system, world landscape, rules and anti-doping governance, career trajectory, risk profile, public narrative, and industry ripple effects. This time the first stage returned zero. No information points. No viewpoints. No entities. No source. No timestamps. Article title: unidentified. Article type: unclassified. In a newsroom this is a familiar situation. An editor receives a report with no date, no names and no numbers. He does not publish it. He calls the reporter back and asks for the original. Swimming raises the bar further. This is a sport where every conclusion is anchored to an absolute time axis, and a swim report without splits is not a swim report. It is a paragraph containing a person's name. My own discipline has long been three-source verification, and the three sources must be independent in their data path. For a swimming result those paths are: the official World Aquatics results database; the electronic timing sheets supplied by the event's official timekeeper; and the technical record of the organising committee, listing equipment checks, false-start warnings and referee decisions. The three do not overlap in nature. One is an archive. One is a raw electronic signal. One is a human decision. Only when all three agree do I allow myself a declarative sentence. Based on my experience of following competition records across nine years, I have never seen a case where the absence of the third source produced a correct conclusion. This time all three paths came back blank. That is the only fact I have, and it is a sufficient fact. What I could do, and did, was describe precisely what is missing and why the gap is large enough to halt the entire analysis. Missing first: performance coordinates. Any serious swimming analysis must position a swim against three reference points — world record, all-time list, current-season ranking. Without an official time, all three are empty and there is nothing to compare. A swimmer going 1:46.00 in the 200m freestyle is a medal contender at a national meet, ordinary in a world final, and entirely meaningless beside Paul Biedermann's 1:42.00 set in Rome in 2026. The same digits, three completely different meanings, depending on the coordinate. Missing second, and the part I regret most: split structure. A 200m swim can be broken into reaction time, the 15-metre underwater segment, and four 50-metre segments. Only then can the distribution rhythm be derived — negative split, front-half fade, or flat distribution. Those three rhythms tell three entirely different stories about conditioning, tactics and psychological maturity. Without splits, I have a final number and an unfounded belief. Missing third: era screening. This is the trap newcomers fall into most often, and it is specific to swimming. At the 2026 World Championships in Rome, 43 world records fell in a single meet. The cause was not that humanity suddenly swam faster; it was swimsuit technology. World Aquatics banned high-performance suits from 1 January 2026. Every record set between 2026 and 2026 therefore carries a label that must travel with it. Remove the label and place a 2026 result beside a 2026 result and the arithmetic is methodologically wrong even though the subtraction works. Biedermann's 3:40.07 in the 400m freestyle, Zhang Lin's 7:32.12 in the 800m freestyle, and Federica Pellegrini's 1:52.98 in the women's 200m freestyle all belong to the labelled group. Missing fourth: pool parameters. Long course 50 metres and short course 25 metres are structurally different sports. A 200m race in a long course pool has three turns. The same distance short course has seven. Each turn is one loss of speed and one recovery, and athletes with strong turning technique benefit disproportionately short course. Converting short-course times to long course with a fixed multiplier is a game I refuse to play. Without knowing the pool, every comparison is worthless. Missing fifth: rule boundaries. Swimming has very fine lines, and part of the value of analysis lies in locating which line a performance is hugging. Underwater travel is limited to 15 metres from the start or turn in freestyle, backstroke and butterfly. In breaststroke the rules permit exactly one dolphin kick during the pull-down after the start or turn. The backstroke start device was approved from 2026 and significantly changed the structure of the opening 15 metres in backstroke events. Each such rule change creates a new data generation, and mixing two generations in one spreadsheet is an error. Missing sixth: qualification status. A and B cut standards, qualifying windows, and the domestic competitive landscape are the three variables that determine what a result means. A qualifying result is a different story from a medal-standard result. Without those three variables, I do not know whether I am reading about performance or about eligibility. Missing seventh: injury data, and here I want to be blunt. Schedule density is the single largest cause of injury in elite sport. No medical team saves an athlete contesting two meets in one week. In swimming the two signature injuries are swimmer's shoulder and breaststroker's knee, both volume-driven rather than collision-driven. Any analysis without injury history and weekly session load is ignoring the largest variable. Missing eighth: career trajectory. The athlete's age, position on the performance curve, puberty-barrier risk in women's events, season-on-season improvement slope, and the team's training model. A coaching change can reverse a career, and I always want to know that coach's track record. Without a named athlete, there is nothing to assess. Missing ninth: landscape and personnel movement. World swimming has a dominant tier, a challenger tier, a chasing tier and a potential tier, and those orderings differ by event. Some events have a stable ruler; others have a throne that is wobbling. Alongside that runs a quieter current: sporting nationality switches, training base changes, sports-science staffing moves. Those movements often predict a turning point years in advance. Missing tenth, and this one belongs to the transfer window I currently live in: market noise. During a transfer window, rumour drowns signal. My filter is simple: rank rumours by evidence, and only by three traceable things — money, contracts, and agents' movements. Where none of those three exist, there is only noise. Within that noise, one trend has irritated me for years. The valuation of young talent is inflating beyond reason. A player with fewer than 50 top-flight appearances is priced at a quarter of an entire club's budget. In swimming the equivalent exists in another form: personal sponsorship deals, scholarship slots, and performance bonuses paid in advance to teenagers who have never carried a final. The bubble will burst, and when it does, the payroll is not the only thing it drags down. Alongside that sits a problem I always keep central in my case selection. Commercialisation of women's competition is largely not taken seriously. Most sponsorship deals for women's events are signed as a line item in a corporate social responsibility report, not as a return-generating investment. I do not need to say that sentence in print. I only need to look at contract structure, duration and broadcast presence. All ten categories above are empty. And the fact that they are empty simultaneously is the most valuable information of that day. I removed the source-quality assumption from the model and the model demanded an explanation from me. There is a very human temptation in this trade: when data is missing, go and find a ready-made story to fill the hole. A breakout teenager, a continental triumph, a national record. Those stories are available, easy to write, and always true to some degree because they assert nothing specific. Writing one takes forty minutes. Writing a table of nine empty categories takes two hours, and nobody reads it. But I do this work because I believe readers need a filter, not another story. Three things must be said clearly about the contrarian angle here, because it is easily misread as mechanical objection. First, blank does not mean low quality. An empty input file almost certainly reflects an upstream capture failure, not a genuinely content-free article. The source may have been unreachable, the original text may have been withdrawn, or the link may have broken. This is a pipeline fault, and it must be handled as a pipeline fault, not interpreted as a sporting conclusion. Second, correlation is not causation, and in a blank state not even correlation exists. I always ask the reverse question: what if the crowd is right? If the source really was empty, then constructing a complete analysis out of nothing is not courage. Predicting Germany's elimination was not courage. It was a number that could not find a seat. The distance between those two acts is the entire ethical content of the profession. Third, the sports analytics industry is selling a dangerous product: manufactured precision. A forecast table with two decimal places looks more trustworthy than one with a single decimal, even when both are built from the same volume of data. I once lost 12 million dong by being overconfident in a model before the 2026 European Championship, judging a team purely on pre-tournament expected-goals averages. An on-pitch event in the opening match reversed the entire tournament. Since then every piece I write carries one mandatory section: non-quantifiable variables, listing injuries, psychology, cards and sudden events. I apply a risk adjustment coefficient between 0.8 and 1.2, and I have dropped the word certain from my vocabulary, replaced by low risk or high risk. Some will say a piece about having nothing to write is a useless piece. I disagree. In an industry where everyone races to manufacture numbers, the person willing to leave a cell blank is the person keeping the whole field from deceiving itself. An analyst has no obligation to be right. An analyst's obligation is to say what the data wants to say. When the data wants to say it has not arrived yet, the only correct action is to record that it has not arrived, note the date and time, and wait. So what is the signal for the next cycle? One: re-run the extraction stage with the recovered source text and check whether it returns at least one information point and one named entity. Without a named entity, there is no analysis. Two: establish source identity before reprocessing, with the original URL. Source quality is a mandatory parameter, not an optional one. Three: verify that the swimming domain label is genuine rather than an automatically defaulted value. The check is simple: does the text contain stroke, distance or rule terminology? If not, the entire analytical framework is running on the wrong domain. I closed the spreadsheet at 4 a.m., left the three blank tabs untouched, and filled nothing in. The next morning I sent the group a single line: no source, no conclusion, awaiting re-run. In a transfer window where hundreds of new numbers are released every day, the ability to say there is no data yet may be the hardest skill to train, and the most valuable one. Readers do not need me to be faster. They need me not to invent.