The Nine-Part Empty Table Tennis Report: When a Blank Cell Is Misread as a Verified One

The report ran eleven pages and was divided into nine sections. Every section had tables, a one-to-five star scale, its own “Evidence” block, its own “Hidden Information” block, and a risk-flag matrix with little squares waiting to be ticked.

I read it twice in forty minutes and made seven margin notes.

On the third pass I realised I had not been able to make a single one.

All nine sections were filled in. Not one blank cell. And almost every cell carried the same line: insufficient information to assess. Technique, tactics and equipment: insufficient information. Player data and head-to-head record: insufficient information. Event system and points rules: insufficient information. Competitive landscape: insufficient information. Six risk categories in the risk matrix: all six insufficient information. Four information-value ratings: one star out of five, annotated “effectively zero”.

Exactly one thing was identified as a genuine risk, and it sat outside the nine sections: the data pipeline itself had returned an empty payload.

I had just read a flawless analysis of something that does not exist. And I think it deserves to be written about, because the error it exposes is not a table tennis error. It sits in the way we read tables.

An elite table tennis match lasts forty-five minutes. In those forty-five minutes the ball crosses the net several hundred times, each crossing carrying a different combination of spin, speed, placement and tempo. The human eye captures a small fraction of it, and the fraction it captures is usually the part that has already ended — the ball has landed on the other side, the point has been awarded.

Most of today’s table tennis analysis technology exists to capture the rest. Live scoring systems break each point into a data row. Review systems at major events turn edge-of-the-table rallies into decisions that can be rewound. Trajectory cameras try to reconstruct the flight path and estimate the spin. Every event in the WTT system — from Grand Smash down through Champions, Star Contender, Contender and Feeder — is tied to a different block of ranking points, and the ITTF world ranking runs on a rolling twelve-month window that collects a player’s best results.

That window creates a treadmill. A player who wants to hold a position has to play densely, travel far, and accept that a small injury in March will surface as a dent in the ranking in March of the following year. Nobody designed the treadmill to hurt anyone. It simply runs like a machine, and a machine cannot tell the difference between a player who chose to rest and a player who was not invited.

At sixty I still write my own statistical software and still hand-code each rally into a pattern. I do it because I do not believe any tool can read a set for me. But I also know I am in the minority.

What matters is that the data from a match does not stay in the arena. It leaves very quickly and travels in several directions at once: onto broadcast graphics, into federation reports, into sponsor decks, into selection files, and onto a trading desk in another time zone. The same feed, the same latency, but the consequences are not the same. A broadcast graphic that is wrong gets fixed within seconds. A trading desk that reads it wrong gets fixed by nobody, and nobody knows it needs fixing.

Which is why the pipeline deserves closer scrutiny than the ranking table it feeds.

Back to the nine empty sections.

I do not consider that report a failure. It was a correct output. It refused to invent a player, an event, a head-to-head record or a rule context in order to fill the table. Every conclusion in a serious analysis must trace back to a specific information point. No information points, no conclusions. That is right.

But that correct output has a shape that is very easy to misread, because it looks exactly like a finished analysis. Nine sections present. Tables present. Star ratings present. Even the “hidden information” block, complete with confidence labels.

A fully populated table is the easiest thing in the world to read, and the easiest thing in the world to misread.

In my work I meet three different kinds of empty cell, and we routinely read them as one.

The first kind is the measured zero. This is real data. A player serves eleven times in a set and faults on none of them — if the system classified each serve type, recorded the placement and recorded the outcome, then that zero is a finding. It says the player chose the safe option at a moment when the match demanded safety.

The second kind is the zero that was not measured. Same column, same zero, but filled with a default value because the classifier did not recognise the serve type, or the camera angle missed the placement, or the data entry operator skipped the field. A coach reading that column will conclude his player served safely, and will adjust the wrong thing. He has just repaired a part that was not broken while ignoring a part that is.

The third kind is the zero nobody asked for. That field was never designed. Nobody sat down and decided that the win rate on the fifth stroke of a counter-loop exchange was worth recording. When a field does not exist, it does not appear as a gap in the report. It simply is not there, and the reader never learns that it could have been.

Those three zeros look identical on a screen. They differ in exactly one respect: the first two can be verified, the third cannot, because there is nothing to verify.

Apply the same test to the ranking table.

The world ranking runs on a rolling twelve-month window. That mechanism has a consequence few table readers notice: two players on the same total can have completely different fragility. If sixty per cent of one player’s points come from a single event, that player’s position is a bet on one week of the year. If the other player’s points are spread across seven events, the same total is a far more stable asset.

The total does not say that. It only says total.

And the rolling window has an automatic valve that nobody operates. When Ma Long — twice Olympic men’s singles champion, at Rio 2026 and Tokyo 2026 — stepped away from the international circuit, his enormous block of points had to be released from the system after twelve months, and an entire region of the ranking suddenly emptied. Nobody decided that. The valve opened by itself. At the same moment, another Olympic men’s singles champion, Fan Zhendong, who won at Paris 2026, stood on the opposite side of the same valve.

A points column can look like a conclusion when it is really nothing more than a blank cell wearing a number’s clothes.

The same mechanism runs inside selection criteria. When a federation sets out quantitative criteria — ranking, points, number of semifinals — those criteria are treated as objective because they are read off a table. But if that table has just swallowed a gap without announcing it, the objective criterion has quietly become a subjective one, and not a single word of the rulebook has to change.

This is where I think table tennis review systems deserve to be faced squarely. They move the argument from the edge of the table onto a screen. The argument does not disappear. It simply relocates, and in its new home it sits precisely in the grey zone of the law — where the question is no longer whether the ball touched the edge, but whether the system has enough resolution to answer that question at all. That grey zone has no referee. It has only an equipment operator.

Something similar happens with challengers outside China. A player like Tomokazu Harimoto can carry the same point total as a Chinese player while being structurally different: event density, spacing between deep runs, win rate against peers in the same tier. The ranking collapses those two structures into one position, and the reader at home sees a single line.

Of the nine sections, the one I studied most closely was the risk matrix. Six categories, six cells, all six marked insufficient information. And the report warned against its own trap: do not let downstream logic read “no flags raised” as “assessed and clear”.

That is the most important sentence in all eleven pages.

Because in table tennis analysis this is the most expensive error and the hardest to see. An empty risk cell carries two opposite meanings. Meaning one: it was examined carefully and nothing was found. Meaning two: nobody has looked. On a screen those two meanings render identically. Only an audit trail of operations can separate them, and almost nobody keeps one.

Every tactical scheme is an organised lie told in the face of the chaos of a match. So is a risk matrix. It is only useful when the reader knows which cell was filled by measurement and which was filled by default.

This is the part I want to say most plainly, and it has nothing to do with table tennis.

The sports data pipeline is designed by people with an incentive to fill. Nobody can sell the sentence “I do not know”. No vendor signs a contract with a blank cell. No bulletin airs the line “not enough data”. So the system default is to fill. And when the system fills, it fills with whatever looks most plausible.

That is why an honest empty cell is more dangerous than a cell full of rubbish. An empty cell summons a human. A full cell stays silent.

Shenzhen taught me that haste in reform only produces a well-irrigated graveyard. I have seen it often enough in sports digitisation projects to have stopped being surprised: people buy the tracking system first, and only then go looking for someone who can read it.

When people replace the grass, they forget to replace what feeds the roots.

And there is one more paradox I have to state, even though it is not pleasant to hear. Refusing to fabricate is cheap in this case, because there was nothing to fabricate. When you hold not a single line of information, honesty is easy. The real test lies elsewhere: when you hold eighty per cent of the data, do you have the courage to leave the remaining twenty per cent alone and say you do not know? Or will you interpolate, will you reason conditionally, will you write a sentence that sounds entirely reasonable and then forget that you just filled in a blank yourself?

There are seasons in which we must learn to live with defeat before the ball rolls. For an analyst, the equivalent is this: there are questions we must learn to leave unanswered, and to state clearly that we have left them unanswered.

Data Limits

What I have: the shape of a nine-part analytical process, and an empty result.

What I do not have, and must state clearly so the reader can weigh it: no player name tied to a specific match, no event name, no head-to-head record, no points-conversion cross-reference, no time-sensitivity assessment, no source-quality assessment. The original report did not have them either. This piece therefore discusses mechanism, not any particular match.

A data payload fit for real analysis needs at minimum four things: a list of information points citable by number, at least one named entity — player, coach, federation or event, a populated source title, and assessments of both time sensitivity and source quality. Without those four, every table downstream is decoration.

I am leaving this section as it stands. Readers have a right to know what they are reading and what they are not.

Next time somebody hands you a table tennis dataset with no empty cells in it — a complete ranking table, a clean risk matrix, a tightly sealed nine-part report — the only thing worth doing is asking one question.

Which cell here was filled by measurement, and which was filled by default?

If the person who handed you the table can answer that for every cell, you are holding something valuable. If they cannot answer it for even one, then however handsome the table is, it is only a sheet of paper that has been watered.

And the thing worth doing before next season is not buying another layer of equipment, but building a single checkpoint at the entrance: do not let an empty payload pass through in the shape of a full one.