Nine Blank Pages: What Table Tennis Data Cannot Fill In

Nine Blank Pages

Last week I received a table tennis analysis report nine sections long. Section one covered technique, tactics and equipment. Section two covered player data and head-to-head records. Section three covered the event system and ranking points. Section four covered the competitive landscape between Chinese table tennis and the rest of the world. Section five covered rules and governance. Section six covered coaching staff and talent pipelines. Section seven covered the risk surface. Section eight covered public narrative and expectation. Section nine covered industry transmission.

The report had tables. It had a six-row risk matrix. It had a comprehensive assessment with a five-star scale. It had a signals-to-watch section and a glossary at the end.

Not a single cell contained a number.

Every row of the risk matrix was filled with the same three letters. The technique section read “not assessable”. The head-to-head section read “no entity identified”. The conclusion explained that the Stage-1 deconstruction input was empty, that no conclusion could therefore be drawn, and that baseless speculation was prohibited.

My first reaction was to delete it. My second was to keep it.

In five years working with spreadsheets, I have never read a document as honest as that one about my own profession.

Why a blank report exists

That report was produced by an automated analysis pipeline. The first stage was tasked with deconstructing a source article: extracting the title, source, article type, core arguments, information points and the list of entities mentioned. The second stage took that output and ran a deep analysis across nine dimensions. When the first stage returned an empty set, the second stage still ran the whole process. It still set section headings, still drew tables, still numbered pages, still printed nine pages.

That is the part worth discussing. A system can generate a document that looks complete from an empty input. Technically, it is not wrong. Professionally, it is a warning.

Table tennis entered the data track about fifteen years later than football and basketball, but its speed has been far higher. World Table Tennis was founded in 2026, taking over the professional tour from the International Table Tennis Federation. The world ranking is updated weekly. Grand Smash events were redesigned for television and data. Every match now comes with a parallel stream of numbers: game scores, service counts, third-ball conversion rates, average rally duration.

Based on my own experience watching matches across three recent World Table Tennis seasons, one thing repeats suspiciously. When I review footage after a match ends, my impression usually drifts from reality in the same direction: I remember the winner as more proactive than they actually were, and the loser as more passive than they actually were. Memory is not wrong on detail. It is wrong on frequency. That is why I started recording everything, including things that seemed meaningless.

The more we automate analysis, the easier it becomes to mistake a filled-in form for a valuable conclusion.

The form always looks complete. It has no shame. Only the reader has shame on its behalf.

And sitting opposite those nine blank pages, I realised they describe the current state of table tennis analysis quite accurately: plenty of frameworks, very little real data, and a growing belief that enough framework will eventually produce a conclusion.

Nine dimensions, nine blank cells

What is interesting is that the nine dimensions in that report are not the product of some eccentric imagination. They are close to a complete list of what a professional table tennis analyst must answer. The problem is that each dimension has a very specific reason to be blank.

One: technique, tactics, equipment. This is the dimension the public assumes is easiest to digitise, yet in practice it is the emptiest. There is almost no public database recording when a player changed rubber, changed blade, or changed sponge thickness, and on what date. Meanwhile, equipment is a variable sitting squarely inside the decisive zone of a match.

In 2026, the ball diameter was raised from 38 mm to 40 mm. In 2026, celluloid balls were replaced by plastic balls. Both changes reduced spin and altered average rally length at a level anyone who has trained can feel. But you will not find an open dataset measuring that change per player, per event, per month.

At the individual level the story is clearer. A player switching from a tacky rubber to a tensor rubber typically needs six to eight weeks to adapt. During that window, short-game control drops and away-from-table counter-looping rises. It is a grey zone visible to the eye but invisible on a spreadsheet. Nobody records the switch date. So the first dimension is blank not because nothing happened, but because nobody wrote it down.

Two: player data and head-to-head records. At Paris 2026, Fan Zhendong won men's singles and completed a career Grand Slam. Truls Moregard took silver, Sweden's first Olympic men's singles medal since Jan-Ove Waldner in 2026. Felix Lebrun took bronze, the first Olympic men's singles medal in French table tennis history. Wang Chuqin, the top seed, was eliminated in the round of 32 by Moregard himself.

Looking at that, head-to-head records sound like a strong indicator. They are not. Two top players meet four to eight times a year. A 6-4 head-to-head tells you nothing beyond the fact that they have met ten times. At that sample size, a two-match gap sits comfortably inside the noise.

The world ranking works the same way. With a rolling twelve-month window, it does not measure form. It measures success at defending points already earned. A player in excellent form with no old points to lose will stand still, while a weaker player who won a major last year will fall fast. Those two opposite movements look identical on a table.

Three: event system and ranking points. World Table Tennis points are fairly clear: a Grand Smash champion earns 2026 points, WTT Finals 1500, WTT Champions 1000, Star Contender 600, Contender 400. The World Championships also sit at 2026.

The more important number is behind it: points-defence pressure. A player who won a Grand Smash last March loses the full 2026 points this March if they fail to repeat the result. That variable shapes schedules, shapes withdrawals, shapes a player's willingness to fly halfway around the world within two weeks. And it almost never appears in commentary.

Four: the competitive landscape between China and the rest. This is the dimension most easily misread, because it changes slowly. Chinese table tennis still holds most seats in the top ten, but the density has changed. In Paris, both Chinese mainstays struggled at different stages. Moregard, Lebrun, Hugo Calderano and Tomokazu Harimoto are all aged twenty to twenty-five. Ri Jong Sik and Kim Kum Hyang of the Democratic People's Republic of Korea took mixed doubles silver, a result that sat inside almost no predictive model.

At junior level, Lin Shidong rose very quickly through 2026. That is a signal worth watching, but a signal is not a trend.

Five: rules and governance. The 40 mm ball, eleven-point games, the ban on hiding the ball when serving. Those changes all had a purpose: more rallies, more television appeal. They also created a grey zone.

Umpires can penalise illegal serves, and that decision sits at the millimetre level of ball placement and the height of the toss. A service fault in the seventh game at 9-9 in front of a packed arena cannot be measured by any index. And it is not the same at home and away. How an umpire treats a home player and an unknown player is not a conspiracy. It is real pressure, from the stands and from the media, and it exists independently of the will of anyone sitting in the chair.

Six: coaching staff and talent pipelines. China's selection system runs through provinces, cities and then the national team, with internal competition so dense that a place at an international event can be harder to win than a medal. Europe took another road: clubs in the European Champions League system, centres such as Saarbrücken or Düsseldorf, where players train together between national league fixtures.

The two models produce two kinds of data. The Chinese model produces a concentrated block of data that is almost entirely private. The European model produces scattered, more public data that is hard to standardise. No dataset connects the two.

Seven: the risk surface. Wrist, shoulder and lower-back injuries are the occupational risk of the modern spin-heavy game. The World Table Tennis calendar thickens that risk. Points-defence risk is administrative rather than physical, but the consequences are identical. And generational risk: a cohort graduating at the same time creates a silence lasting four to five years.

Eight: public narrative and expectation. Fan culture in table tennis has shifted fast in a few years. Pressure on young players no longer comes only from the scoreboard. Chen Meng won Olympic women's singles in two consecutive cycles, in Tokyo and in Paris, yet most media storytelling circled another player. The gap between crowd expectation and objective result is a real, measurable index, and almost nobody puts it in a report.

Nine: industry transmission. The equipment market runs on product launch cycles, not performance cycles. A new rubber needs a top player using it, and sponsorship contracts usually run months ahead of results. On the rights side, table tennis is repeating a familiar error: platforms pay high prices for broadcast rights, then have to recover them through advertising and subscriptions, while loyal audiences concentrate in a handful of markets. The rights bubble in table tennis has not burst, but it is inflating on the same formula.

What data cannot fill in

This is where I want to be direct. An indicator does not generate meaning simply because it appears on a table.

A simple example: a player wins seven of ten matches after changing rubber. Reading that, people conclude the rubber change helped them win. But between those two variables sit at least three confounders: an easier schedule, a healed injury, and the arrival of a familiar opponent. A sample of ten matches is not enough to separate those three, let alone assign causation.

The second problem is imported analytical frameworks. Football has expected goals, has passes allowed per defensive action. Basketball has shooting efficiency, has usage rate. Table tennis has no possession, and an average rally lasts a few seconds. Indicators of that kind, transplanted into table tennis, usually become a mechanical addition with no theoretical foundation. People call it an advanced metric because it is complicated, not because it is correct.

We do not hunt treasure, we hunt the way to read the map.

And there is one decisive skill in table tennis that no statistical table records: the ability to read a serve. That means judging spin direction, spin volume and placement in less than the blink of an eye. There is no column for it. There is no index for it. But it decides most short rallies, and short rallies decide most games at close scores.

In an empty arena, that variable becomes even harder to measure. When the arena is empty, data sits and cries alone. Service rhythm shifts, the sound of the ball is no longer swallowed by the crowd, and every player hears their own contact more clearly — which helps some and hurts others. The model does not have that variable. The model has never had that variable.

Data cannot save a match, but it shows why the match died.

Conclusion

Back to the nine blank pages. I think we should read them differently. A document that declares it knows nothing is a good document. The problem is that most documents in this industry do not declare it.

The signal I will watch in the coming period is not about any specific player. It is about whether table tennis data opens up to the level of each rally, each serve, each umpiring decision. If it does, we get our first chance to verify what the naked eye has always asserted without being able to prove. If it does not, we will keep reading nine-page reports, fully formatted, and blank in exactly the cell that matters most.

Numbers do not lie, they only keep secrets. The problem is that we usually forget the second half of that sentence.