The Empty Column: What a Failed Tennis Data Extraction Taught Me

2:47 a.m. The data table for the men's singles match I had been tracking for four hours was full. First-serve points won: a number. Break points saved: a number. Total points, double faults, match duration: all numbers. The last column — the one recording rally-length distribution — came back blank. Not zero. Not a dash. Blank. The player won that match. The stands applauded, the electronic scoreboard ticked over, and the next morning's reports would call it a convincing win. I sat still in front of a blank page, because there was nothing on it capable of defending a conclusion. Twenty-five years watching sport, more than 7,000 articles written, and the lesson that repeats most often is this: an empty cell in a spreadsheet is only a place not yet filled in. It is testimony not yet taken. A writer in a hurry will turn it into a verdict. Tennis is the most heavily measured of all individual combat sports. Wimbledon introduced Hawk-Eye in 2026. In March 2026 the ATP announced Electronic Line Calling Live across its entire tour from the 2026 season, turning every line call into a traceable data point. The 25-second serve clock arrived in 2026. Off-court coaching was permitted from 2026 under an ATP trial. What few notice: every rule change creates a new column. Rules do not merely govern matches; rules manufacture data. When the serve clock appeared, people began measuring preparation time. When coaches were allowed courtside, people began counting interactions. Spreadsheets expand not because of technology, but because of regulation. In Vietnam we consume tennis data far better than we produce it. Fans follow Nguyen Thuy Linh round by round, know her opponents' names, memorise the schedule — yet there is almost no detailed statistical table built by Vietnamese hands for Vietnamese players. We read ATP numbers, WTA numbers, foreign aggregators, and then interpret them by feel. I work as a data journalist in Hai Phong, covering tennis for Vietnamese readers. The daily job is to reconstruct a match from indicators before retelling it in words. When the first layer of data breaks, every layer behind it breaks too — and the reader never sees where the break happened. In a spreadsheet, "empty" carries at least three different meanings, and those three meanings lead to three opposite conclusions. The first is zero: a player faced three break points and saved none, 0/3. The second is null: the data does not exist, the system never recorded it. The third is missing: the data exists but was never retrieved. Sports reporting usually has one sentence for all three: "he was weak at the decisive moment." That is the most expensive logic error in the trade. A player who saves 0 of 1 break points has a 0% conversion rate, but the sample is one. Nobody can conclude anything from a single observation. At 0/0 — a player who faced no break points at all — the rate is undefined, and the correct conclusion is "insufficient evidence." That is the sentence I have written most in 25 years, and the one editors strike out most often. I keep one precedent to remind myself whenever I want to write fast. Midway through the 2026 V-League season I published the first series applying expected goals to Vietnamese football. Hai Phong faced SLNA at Lach Tray; the hosts generated 1.92 xG and lost 0-1 to an individual error. The media called it decline. I called it random injustice: the opposing goalkeeper made 11 saves, 3.8 times his own season average. The piece was mocked for two weeks, until the Hai Phong head coach publicly cited my numbers in a press conference. People remember results. I remember the conditions that produced them. In June 2026, before Germany faced South Korea in the World Cup group stage, I published a pressing analysis: Germany's PPDA had fallen from 8.1 in 2026 to 12.6 in 2026, average distance covered down 6.2 km per match. Germany held 74% of the ball and lost 0-2, eliminated in the group stage. Germany collapsed in my spreadsheet before it collapsed on the pitch. But what I kept from that story was not the correct prediction. It was the way I had presented it: precedent first, indicators second, conclusion only once the margin of error was stated. Apply the same method to Vietnamese tennis and three columns stand empty. The first is first-serve points won split by surface — a player who serves well on hard courts does not automatically hold that efficiency on clay. The second is break-point conversion with a real denominator. The third is rally-length distribution, which decides whether a player wins through power or through the ability to extend rallies. Without those three columns, every judgement about competitive nerve is guesswork delivered in a confident voice. And a confident voice is the cheapest thing in this trade. The blank column from the opening belongs to a different class of error. My system returned nothing, and had I not checked, I could have written an analysis of that player's "suspicious silence." In truth my extraction pipeline had failed at the download stage. No match was silent. Only the person collecting the data was silent. Every shot is a hypothesis. xG is how we verify it. In tennis, every serve is a hypothesis too, and first-serve points won is the verification — but only when the sample is large enough and the provenance is recorded. That is why I set myself an inviolable rule: distinguish clearly between "a match with nothing noteworthy in it" and "my system failed." Those two get merged in most sports analysis, and that confusion is the source of half the worthless conclusions in circulation. The counterintuitive angle sits here: more data does not mean more truth. From the 2026 season, as Electronic Line Calling Live blankets the ATP tour, the volume of ball-tracking data will explode. But correlation is still not causation. Players who win more second-serve points tend to win matches — that does not prove second serves are the cause. They may simply have met weaker opponents in the first round. The greatest risk in data journalism is turning numbers into ritual: citing an indicator to borrow credibility, then concluding by feel. That is data theatre, and it is more dangerous than citing nothing, because it dresses guesswork in clothes it has not earned. I also have to admit my own boundary. A spreadsheet cannot measure the moment a player stands at break point in the third hour of a third set and still chooses the line. Physical data can report heart rate, movement speed, metres run. Spectators can leave the stadium, but physical data never rests. The will itself remains outside every table I have ever built. So my conclusions always carry a margin of error. Not to dodge responsibility, but so the reader knows where they are standing. Data is never in a hurry. People in a hurry are the ones who get it wrong. The next round of this story will be decided by a very small technical detail: whether tournaments in the region publish raw ball-tracking data, enough for outsiders to reconstruct a match, or only processed summaries. If only summaries exist, we will keep writing about tennis by feel — and keep calling blank columns verdicts. The question I leave for myself, and for anyone holding a table of numbers: when the cell is empty, are you missing information about the player, or information about your own system?