Table Tennis and the Data Void: When the Analysis Board Returns Zero

Last Friday night I opened the analysis file for a WTT tournament round and found the same sentence repeated in every cell: insufficient information. The column for point-win rate on serve was empty. The column for touches inside the opponent's danger zone was empty. The column for the landing distribution of the sidespin serve was empty. The column for average rally length was empty. The file was not corrupted. The system reported no error. It simply had nothing to hold.

In forty years of watching this sport I have grown used to data arriving late, incomplete, or wrong. I had never met a table tennis dataset that returned zero across every field at once. What struck me was that I was not confused. I felt lighter.

Some things only become visible when the cell is left blank.

The professional table tennis calendar runs on the WTT model, tiered into Grand Smash, Champions, Star Contender and Contender events, alongside continental and national circuits. World ranking points are calculated on a rolling 52-week window using a player's best results. Table tennis therefore owns a fairly complete administrative data layer: who beat whom, at what score, in which round, for how many points, and how far they moved.

Beneath that administrative layer sits an almost total void.

No independent data provider labels every table tennis rally at global scale the way Opta or StatsBomb does for football. No public dataset records whether the serve went short or long, topspin or backspin, where the first and second bounce landed, where the receiver stood, whether the third ball was a loop or a flat hit, and on which stroke the rally ended.

Football has thousands of matches tagged pass by pass. Basketball has play-by-play to the second. Tennis has Hawk-Eye recording the landing point of every ball. Table tennis, a sport where the landing point is decided in less than the blink of an eye, has almost nothing.

That is the foundational paradox: the more micro-variables a sport has, the less micro-data it keeps.

I want to separate this void into layers, because merging them leads to wrong conclusions.

The administrative layer is the thickest. Match results, game scores, ranking points and raw head-to-head records are all searchable. Anyone who wants to know how many times Wang Chuqin has faced Tomokazu Harimoto, and how many he won, can find out. This layer serves mass media well, and it is why result bulletins keep running.

The micro-technical layer is different. To know how a player won a point, a human must sit in front of a screen and press a button for every rally. No algorithm replaces that today. High-speed cameras can capture the ball's trajectory, but they cannot classify spin automatically. A forehand topspin and a backhand topspin share the same parabolic path; they differ on an axis of rotation the naked eye cannot see and off-the-shelf software cannot read.

This is the point I want to press. Table tennis lacks data not because people are lazy note-takers, but because labelling one table tennis rally costs more than labelling one football pass, while the commercial value of that rally is many times lower.

A football pass gets labelled because tens of millions of people will pay to read about it. A short sidespin serve followed by a backhand flat hit ending on the fourth stroke matters only to a small audience, and that audience has never been measured well enough to form a data market.

I once tried to do the work myself. In 2026 I hand-labelled twelve men's singles matches at a WTT Champions event. Each match took about four hours, meaning nearly fifty hours for a dataset too small to support any statistical conclusion. I counted each player's point-win rate on short serves, the rate at which they were attacked on the third ball, and the landing distribution by quarter of the table. With twelve matches, the margin of error exceeded every difference I found.

Data does not lie; readers misinterpret it. Twelve matches prove nothing certain about a player, and I came close to writing a wrong conclusion from them. Since then I changed my method: note-taking is a tool for asking questions, not for closing them.

There is another layer of the void that few people discuss, and it is heavier than the technical one. That is the secrecy layer.

Table tennis is a sport where tactical advantage lives in very small details: a wrist angle off by a few degrees, a contact point a fraction of a second earlier, a landing pattern that hides intent. Those details have value only when the opponent does not know them. Publishing them as open data means handing an advantage to someone else.

At leading national teams, opponent analysis is conducted seriously. But that data stays indoors, is written by hand, is coded by internal convention, and never leaves the system. It does not exist publicly because there is no reason for it to exist publicly. Every tactical system is a confession, and no team wants to confess to outsiders.

So the table tennis data void is the product of two forces pushing the same way: a market too thin to sustain an independent data provider, and a secrecy culture that keeps data indoors.

I should be clear that this is an area where I have been wrong, memorably so. In 2026 I criticised a European football team's tactical shape based on an impression of how far one midfielder ran. The match went entirely against my judgement. Readers pushed back hard, and they were right. The lesson was not to stop commenting, but to stop commenting without a measurement framework. Since then, whenever I want to praise or criticise, I force myself to answer: which indicator backs this sentence?

In table tennis that question frequently has no answer. And when there is no answer, I choose to say plainly that the data cannot answer it.

That is why an empty dataset does not bother me. It is honest. A dataset full of skewed numbers would be far more dangerous. But stopping there is not enough. We need to look at the work the void leaves behind.

There is a counter-reading of this situation, and I think it is truer than the conventional one.

The assumption is that table tennis is falling behind because it lacks data. Football is ahead, basketball is ahead, tennis is ahead, and table tennis must catch up. But the results of data abundance elsewhere deserve comparison. Football has xG, and xG has been abused to the point of becoming decorative labelling in most analysis. People cite indicators instead of watching matches. What xG cannot explain — a coach's decision, an individual's form on one particular night, a referee's threshold — is pushed out of the discussion because there is no number with which to discuss it.

Table tennis was spared that phase. Here, an analyst must sit and watch, must count by eye, must remember. No indicator shields laziness. An empty arena says more than thirty thousand spectators, and a match without data still says a great deal to anyone willing to look.

Put differently, the data void acts as an immune system for the analytical craft. It filters out those who can only read tables. It keeps those who can read gaps.

I am not naive enough to think missing data is good. Some conclusions can never be reached without a large labelled dataset, and I accept that part of what can be known about this sport will remain out of reach. But between a world full of numbers with wrong conclusions and a world short on numbers with cautious ones, I choose the latter.

The more interesting question is which way this void is moving. High-speed phone cameras are everywhere. Machine-learning models that recognise ball trajectories have advanced quickly. One day, automated spin classification will be cheap enough for a small club. Then the issue will no longer be whether data exists, but who owns it and who is allowed to read it.

What I want to verify over the next twenty-four months is concrete. Whether any independent provider publishes a rally-by-rally labelled dataset covering a full WTT season. If it does, table tennis enters the phase football entered fifteen years ago, with all its good and bad consequences. If it does not, this void remains a structural feature of the sport rather than a temporary defect.

The tactical board has no room for noise. Sometimes it also has no room for data. Anyone who reads gaps has to learn to live with that.