When the Spreadsheet Returns Zero: The Discipline of a Table Tennis Data Journalist

Tuesday, 8:40 a.m., Shanghai. My spreadsheet opens with exactly one live cell. The other nine fields are blank: source headline, outlet, article type, one-sentence summary, author stance, article purpose, information points, entities involved, time sensitivity. The only populated cell is the domain label — table_tennis. No athlete's name. No tournament. No date. Not a single value to cross-check against. People outside the trade look at that and ask: why not just write something? Table tennis has tournaments every month, players every month, stories every month. Write on instinct. Readers will still click, still share, still leave comments under the piece. I don't write. On this desk, a blank field is a verdict, not an invitation. When ordinary eyes are asleep, the data stays awake — and it has already seen. This time the data said something else: there is nothing yet to see. Two stages, and one empty cell My process runs on two stages. Stage one reads raw text and extracts four mandatory items: entities (who, which tournament, which association), two to four concrete information points, the source tier, and the publication date. Stage two is where I apply a nine-dimension frame — technique and equipment, player data and head-to-head records, event system and points rules, competitive landscape, rules and governance, coaching staff and talent pipeline, risk surface, public narrative and expectations, and industry transmission. Without those four items from stage one, stage two has no subject. The nine dimensions remain in the frame, but they sit empty, like nine drawers with nothing yet to file. I built this process after paying for it. In 2026 I published an analysis of a well-known foreign striker at a Shanghai club. He scored 18 goals that season. But the team's PPDA with him in the starting eleven was 14.3 — the looseness of a side that has abandoned pressing; with him on the bench it was 9.8. I called him a defensive obstacle at the front of the pitch. The internet called me a bookworm. A month later, that club lost 0-4 to a direct rival, and the first goal conceded came from his own failed press. My old piece was dug up and passed around. The lesson I took was not that I had been right. The lesson was that if I had been wrong, I would have had no way of knowing where, because none of my claims were anchored to any indicator at all. So I set two rules. One: every piece must state which data would refute it. Two: any claim about work rate or spirit must carry pressing figures and distance covered before the sentence is written. A piece that cannot be refuted is not analysis; it is an opinion dressed up in numbers. My desk later standardised this into a recurring format called Star Audit, with a fixed statistical checklist, applied across both football and table tennis. Its principle is simple: reputation does not get exempted from inspection. Table tennis is stricter than football in exactly one respect: the calendar. Ranking in the WTT system runs on a rolling 52-week window. Old points fall out of the system on their expiry date without anyone needing to beat anyone. A player can lose several hundred points in a single week without losing a single match; another can climb the rankings simply by standing still. Based on my experience watching matches across WTT events, I have repeatedly seen a player celebrated for a rise that was really just the system deleting the points of the person above him. An input with no publication date cannot be placed into that calendar slot. And if it cannot be placed into the calendar, ranking analysis becomes nothing more than storytelling. Technique, equipment, and shocks that carry dates The first dimension in the frame is technique, tactics and equipment. To touch it, I need to know a player's playing-style system, or a specific stroke, or an equipment change, or a match with enough tactical detail to dissect. An empty input gives me none of the four. In table tennis, equipment has never been a minor detail. On 1 October 2026, the official ball diameter rose from 38 to 40 millimetres. In 2026, the 21-point game was cut to 11 points per game, best of seven. In 2026, the hidden-serve rule came into force. On 1 September 2026, speed glue was banned. On 1 July 2026, the plastic ball replaced celluloid. Each of those dates changed the statistical architecture of the sport. A larger ball reduces flight speed and the spin that can be generated from the same motion. The 11-point format shortens the window for a comeback, turning the opening service exchanges from a preamble into the decisive phase. The hidden-serve rule took away part of an advantage that servers had treated as a given. The speed-glue ban cut off a source of spin and speed around which an entire generation had built its game. The plastic ball changed contact feel, and therefore changed the average length of a rally. I can recite those milestones because they carry concrete dates and public documentation. But I cannot say which of them is acting on whom this week, when I have no name of a player and no name of a tournament. That is the line between background knowledge and analysis. Background knowledge I keep ready in my head. Analysis requires a subject. Ranking versus real strength The second dimension is player data. What interests me most here is a divergence: the gap between world ranking and real strength. Ranking is a function of the calendar, not of class. A player who enters many small events can accumulate points steadily and sit above a player who enters only a few big ones with better results. Points-defence pressure works the same way: one player walks into an event defending a position, another walks in with nothing to lose. Those two states produce two different behavioural patterns in matches, and I can only tell them apart with an expiry schedule in hand. To see that divergence, I need at minimum one player's name and one ranking snapshot at a specific moment. From there I can build the head-to-head record — overall, last two years, and at the three biggest events alone. Those three layers usually tell three different stories, and the place where they disagree is the place worth writing about. An empty input does not give me the first name. Event system and points rules The third dimension is the event system. What an event is worth lies not only in prize money but in its points structure and the strength of its entry list. A title at an event where all the top players are present says something different from a title at a thin field. The draw works the same way: the difficulty of your half, the chance of meeting a bogey opponent at a particular round, and how the draw rules force players from the same association apart. All of that is a function of the calendar. Event tier, position in the Olympic cycle, draw timing. Without dates there is nothing to compute. This is the most time-sensitive of the nine dimensions, and the one most quickly disabled by a dateless input. Competitive landscape The fourth dimension is the landscape. I always separate men's and women's singles, and doubles from team events, because the degree of openness differs sharply by event line. A line where three semifinal berths usually belong to three different associations follows entirely different analytical logic from a line dominated by one association for years. Same ranking, two completely different readings. But to separate them, I need to know which line we are discussing. The input does not contain a single association name or player name. Rules and governance The fifth dimension is rules and governance. My historical reference set is thick: ball diameter change, the 21-to-11 scoring reform, the hidden-serve rule, the speed-glue ban, the change of ball material. Every reform creates winners and losers, and usually not the same group. But to say who won and who lost, I must know which rule is at issue, in which direction it was amended, and at what stage. A reform proposal without an effective date cannot be assessed for impact, because its impact still depends on how associations will adjust their training plans. I do not write about a regulation whose effective date I cannot establish. That is a limit I impose on myself, and it has saved me more than once. Coaching staff and talent pipeline The sixth dimension is coaching staff and the talent pipeline. Here I track the age structure of the main squad, the conversion efficiency of the junior cohort — how many players from the under-21 group reach the senior squad, and how long it takes — and the state of the generational handover. The usual triggers for this dimension are a coaching change, a wildcard allocation, a training-camp report, or a remark about internal competition. None of them appear in the input. Risk surface The seventh dimension is the risk surface. I split it into six groups: competitive risk covering form, injury and technique; selection risk covering ranking and entry slots; generational gaps; governance and public-opinion risk; systemic risk covering the calendar and the Olympic cycle; and opponent risk. Each group needs a concrete subject before any item can be screened. Without a subject, all six lie out of reach. The only live risk in this run is the risk of the process itself: the chance that a model further down the chain will fill the gap with a plausible-sounding guess. I rank that above every sporting risk, because it does not live in the table tennis data. It lives in the analyst's head. Narrative, expectations, and industry transmission The last two dimensions are public narrative and industry transmission. For narrative, the first thing I need is the source tier: is this mainstream media, an opinion column, or self-published fan content? The gap between public expectation and measurable reality only means something once you know where the expectation came from. The same result carries a different weight when reported by a long-established sports outlet than when pushed by a supporter account. For industry transmission, I need a trigger at the upstream node — a star's result, a policy decision, an equipment change — before I can trace the path downstream: the equipment market, the training base, the commercial ecosystem of events, the commercial value of players. No trigger, no path. No path, no article. The counter-intuitive part sits somewhere else People assume the hardest part of this trade is finding the answer. The harder part is refusing to give an answer when there is nothing to answer. The entire sports-media ecosystem runs the other way: content must ship steadily, headlines must exist, gaps must be filled. A blank cell is a cost. And for a cost, there is always someone willing to pay with a plausible guess. The plausible guess is the hardest counterfeit to detect, because it is usually right about correlation. Once a match ends, the brain automatically stitches two events together and calls the seam causation. A player changes his rubber and wins — the new rubber is the cause. A player changes his serve pattern and wins — the new tactic is the cause. The only check I trust is to rerun the scenario with the opposite assumption: if the ball had dropped on the other side at the deciding point, would my explanation still stand? If the answer is still yes, then what I have is not an explanation. It is a story told very smoothly. I once published a piece that the trade mocked before a major match, because my model showed the favoured side losing with a higher probability than the market believed. The result went that way, and the piece was shared widely. The Korea shock was never a shock — it was the first time the numbers were listened to. But if I retell that story while stripping out the model and the error bars, I am selling readers the feeling that data is always right. It is not always right. It is only more honest than our memory. Another place where analysts slip is praise. After a correct call, readers come back and call it vision. I have to rerun the scenario in reverse before accepting any praise: if the result had gone the other way, would my piece be judged a failure? If the answer is yes, then my model is being graded on outcomes rather than on the quality of its reasoning. That is how an analyst turns into a prediction salesman, and I do not want that job. In the other direction, I do not answer comments just to jab back. I give one metric and stop. No sarcasm, no exchange of barbs. One row of data is enough to answer, and silence is the rest of the answer. Signals for the next cycle In the next run I will track three operating metrics of the process itself, not of table tennis. First, the fill rate of the information-points field — the share of inputs carrying at least one concrete information point. Second, the fill rate of the source-tier field. Third, the fill rate of the publication-date field. If an article visibly has a source and a date and those two fields are still blank, the problem sits in the extraction stage, not in the article. If two or three consecutive runs from different inputs all return empty, it is a systemic fault, and systemic faults must be fixed at the system, not at the writing. The work for this week is simple and deeply unattractive: stop aggregation, quarantine the empty result, rerun stage one against the raw text. There is no reward for that. Nobody shares an empty spreadsheet. But this trade does not live on being shared. I write dryly, so that the game we love does not get buried under sentimental hands. And in a week when the data cannot yet speak, the only correct move is to stay quiet until it does.