When the Spreadsheet Falls Silent: Table Tennis and the Limits of Data

A night in Shanghai, an empty data board, and the biggest question of the analyst's trade.

---

It was 3:47 in the morning when I reopened the footage of a WTT Champions final and launched the point-tracking software. The left column showed two players' names. The right column was blank. No serve-win percentage, no attack-after-backhand-serve index, no heat map of forehand loops. A rendering error at the raw-data layer had wiped out everything I needed to write a post-match analysis.

I brewed another cup of tea. Staring at that white space on the screen, I understood something that seventeen years in the trade had taught me but that I had always postponed admitting: modern table tennis analysis stands on a foundation thinner than we like to think. We build massive towers of conclusion on a layer of sand called tracking data. When that sand disappears, the tower collapses in silence, and the writer like me is left only with a pen and memory.

Context: An era when every touch leaves a trace

Table tennis has come a long way since the ITTF moved the ball from 38mm to 40mm in 2026, cut a game from 21 points to 11 in 2026, banned the hidden serve in 2026, killed VOC speed glue in 2026, and shifted from celluloid to plastic balls in 2026. With every reform, people said the same thing: the match is now more transparent, easier to measure, more scientific.

By the 2020s, the WTT competition system, with its rolling 52-week points mechanism, turned every player into a number machine. High-speed camera tracking records ball placement, estimated spin, and each player's position during every rally. My analytics dashboard can tell you that Fan Zhendong loses 9.3% of his attacking efficiency when forced into the left corner in the fourth game. It can tell you that Wang Chuqin wins 71% of points when serving short to an opponent's forehand side.

This really is a revolution. But I sat at MIT Sloan in 2026, and I know one thing about revolutions: they always sell you more than they deliver.

At Sloan, I encountered a basketball report. A player hit 45.2% from a corner spot but attempted only 1.7 shots per game. The room debated whether he should shoot more. I stayed behind, cross-referenced tracking data with the playbook, and understood that the system had deliberately sacrificed volume to optimise quality. My piece was widely cited. But what I brought home from Boston was not a method. It was a suspicion.

That suspicion became a burden when I moved to covering table tennis for the Chinese market.

Why? Because table tennis is a sport where data looks beautiful on paper but betrays you easily at the table. A 40mm ball crosses the net in about 0.15 seconds. A modern counter-looping rally can last seven or eight exchanges in two seconds. In that window, a player decides by reflex, by muscle feel, by the memory of thousands of training hours — not by a spreadsheet.

And yet we keep building an industry of conclusions on exactly that window.

The Core: Nine analytical dimensions and the white space at the centre

For three years I have built a nine-dimension framework for every big match. Technique and equipment. Player data and head-to-head. Event system and points. The China-versus-the-rest landscape. Rules and governance. Coaching staff and talent pipeline. Risk surface. Public narrative. Industry transmission. Each dimension is a lens. Together they give me a picture thick enough to write from.

But tonight, all nine lenses pointed into an empty room. The first dimension needs a style label — loop-drive, fast attack, chopping, pips, or penhold. None. The second needs a ranking, a head-to-head table, a points ledger. None. The third needs an event name to place it in the Olympic cycle. None. One by one, against each dimension, I wrote the same sentence: insufficient information, cannot assess.

What chilled me was not the empty table. What chilled me was how fast my mind wanted to fill it.

I had felt that before. In 2026, I covered the Western Conference Finals between the Houston Rockets and the Golden State Warriors. Houston led 3-2, then Chris Paul tore a hamstring in Game 5. By Game 7, Houston missed 27 straight three-pointers — the worst streak in playoff history. In a chaotic press room, I stayed behind and rewatched all 27 attempts. Mike D'Antoni's system drew 68.4% of its points from threes or layups. When the Warriors sealed the middle, Houston had no Plan B. I published an analysis the next morning, grouping the 27 misses into five repeating situations.

That piece was right. But while writing it, I fooled myself into believing I had explained the collapse. I had not. I had only described it in tighter language. The weight in a player's chest when he realises his body will no longer obey — that has no column in my spreadsheet.

Numbers speak, but pain is not in the spreadsheet.

So when tonight's table tennis data board came up empty, I had to choose between two roads. The first was to invent. I could write a very convincing analysis of a player's "serving rhythm" without a single data point, leaning on vague memory and good phrasing. Newsrooms still print such pieces every day. The second was to admit the emptiness and write about it.

I chose the second, and I want to explain why that was the harder choice.

Why table tennis data often goes silent

There is an implicit assumption in sports analytics: data always exists; we simply have not mined it yet. That holds in basketball, where every possession is a discrete event with a clear score. It holds in football, where every pass has coordinates. It only partly holds in table tennis.

Table tennis has a property that resists measurement. The score is clear — 11 points a game, first to 11 with a two-point margin wins. But the cause of the score is blurry. A winning ball can come from spin, from speed, from placement, from an opponent's error, from a net cord, or from an opponent losing focus for half a second. Tracking systems capture the trajectory. They do not capture the decision.

And in table tennis, the decision is everything.

Take serving. In a top match, a player may have six or seven different serves, each with several variations in spin and placement. He chooses based on the score, on what the opponent just did, on how his wrist feels after a twisting rally, on which side the crowd is screaming from. The data board records the result of that choice. It does not record the reason.

I once spent a week analysing the serve sequences of a top player at a major event. I built a predictive model: based on the score and the previous rally, I could guess the next serve type with 62% accuracy. I was proud of that number until his assistant told me one sentence: "The score matters less than whether he trusts his forehand tonight."

I saw the future at MIT Sloan, and it had no room for emotion.

The analyst's trap

This is the private tragedy of my trade. We are trained to believe every phenomenon has a measurable cause. When data goes silent, the first instinct is not to stop. The first instinct is to infer. To infer from memory. To infer from small samples. To infer from feeling. And then to present that inference in the confident tone of someone who has data.

I have done that. Many times.

In 2026, mid-Finals, I received a vague tip about a star's calf. Colleagues chased the rumour. I built a verification frame: closed practice schedules, court photos, the degree of rotation in twelve minutes of play. I refused to write until I had three independent sources and a biomechanical risk model. I calculated the load on the Achilles from fourteen sprints in the second half and put the figure at 87%. I published six hours before he went down. The piece was fully confirmed.

People called it a success. I call it the only time I was patient enough not to invent.

Silence is a kind of data. I learned that late, and I still have not learned it fully.

The trap is not that we lack data. The trap is that we refuse to admit it. An analysis built on ten hand-picked rallies, retold in a confident voice, sounds far more convincing than one that says, "we do not yet have enough data to conclude." Readers want answers. Newsrooms want copy. Algorithms want content. And the writer sits in the middle, alone with an empty spreadsheet, feeling the pressure to fill the void with anything.

That is when I understood why I once wrote that probability is the curse of the underdog. Not because probability is wrong. Because probability creates the illusion that we have understood.

The contrarian angle: Models do not fail at the table edge — they fail inside the writer's head

There is a story retold endlessly in sports analytics: the model was right until it was broken by an irrational moment, and the lesson is "humans always beat data." I do not believe that simple version. It is too easy, and it lets the analyst off the hook.

The truth is subtler. The model does not fail at the table edge, where the ball touches the table. The model fails inside the writer's head, at the moment he decides to trust an inference that lacks evidence because he needs a conclusion.

Consider Houston again. Yes, 27 straight misses was an absurd event. But my analysis of it was not absurd. I pointed to the lack of attack variation, and that was correct. The problem was that I told the story as if the lack of variation explained the entire streak. It explained only part. The rest — tired bodies, tight nerves, luck, shaking hands — lay outside the model, and I quietly folded them into the part I could explain.

If that game were played ten times, Houston might win seven with the same strategy. My model described one playing. It did not predict fate.

The Houston shock of 2026 taught me that probability never speaks in the final minute.

In table tennis, this is harsher still. A match can turn on a single net cord at 9-9. That ball is not the output of a modellable process. It is a random physical event. Every model I have about that match becomes meaningless the instant the ball clips the edge. And an honest analyst must have the courage to say: most of what I know about this match does not apply to the decisive moment.

I once believed in models. The Rockets taught me that humans break every model. But seventeen years later, I realised I had taught myself a new illusion: that I could write about humans in the language of models. I cannot.

At Sloan they sold me a revolution. I bought only a part of it — the rest is human.

Old underlines: where I was wrong

I keep a fixed section in my notebook called "old underlines," where I record the predictions I got wrong. That list is longer than the list of correct ones, and I keep it for a simple reason: every time I want to write a certain sentence, I reread it.

Once I predicted a young player would win a major based on a superior attack-after-serve index. He lost in the second round to a much lower-ranked opponent, unable to handle the pressure of a home crowd. His index said nothing about his never having played before five thousand people.

Once I wrote that a former champion was finished, based on declining movement speed. He reached the semifinal of the next event, winning three straight matches by switching to a control style. I measured foot speed. I did not measure the will to adjust.

The Durant investigation was never about finding a culprit, but about understanding how pain gets hidden. I keep that line as a reminder: my job is to understand, not to indict data or people.

Transmission: when table tennis analysis becomes a commodity

There is a dimension I rarely write about, but tonight's blank board forced me to look at it. Table tennis analysis has become a transmission chain from upstream to downstream. Upstream is equipment, youth development, training. Midstream is events, associations, clubs. Downstream is broadcasting, commerce, derivative markets.

Data flows through that chain. When data breaks at one link, the whole chain suffers. An event without full tracking data sells less to broadcasters. A player without good metrics struggles to negotiate sponsorship. A club without a selection model buys the wrong people.

But the more worrying direction is the reverse. When that market is starved of data, it does not stop. It buys cheaper data, blurrier data, data manufactured to fill the gap. And people start making decisions on numbers no one can verify.

I have seen a player's wage contract negotiated on a metric no one in the room understood. I have seen an article cite "tracking data" from an unverifiable source. The white space on my spreadsheet tonight is a miniature of the white space spreading across the industry.

A story: A Korean, a Chinese, and two ways of meeting silence

I was born in Korea, work in China, and stand between two coaching schools. The Korean school leans toward systematising every stroke, breaking technique into repeatable units, measuring and optimising each. The Chinese school leans toward collective emotional intensity, turning discipline into a shared current of the whole team, where the individual dissolves into the group.

Tonight I realised something. Both schools collapse at the same point: where data goes silent.

The Koreans, facing a blank board, tend to return to process. They say: rerun the basic drills, measure again, the system will fix itself. The Chinese, facing a blank board, tend to return to emotion. They say: trust the group, trust history, trust spirit. Both avoid the hardest question: when we know nothing, should we say we know nothing?

I do not write to answer that question. I write to keep it open.

Every victory is a hypothesis not yet falsified. And every defeat, by the same logic, is a hypothesis falsified by data — but only if the data is still there to falsify it.

What it takes to fill a blank board

If a newsroom ever asked me what I need to write a serious table tennis analysis, I would give a short list.

I need a named player, with a style description.

I need a head-to-head table, or a set of recent results.

I need an event name and its date, to place it in the Olympic cycle and map it to the WTT points table.

I need a coach's name and at least one signal of change.

I need a verifiable source with a publication date.

And I need the hardest thing: a little courage to say this list is not yet filled.

In seventeen years, what I learned was not to read data better. It was to recognise when data is standing still, and when I am fooling myself into filling a white space with words. My spreadsheet tonight is blank. That is a lesson, not a failure.

I once believed in numbers and was abandoned by them in the final minute. I once believed in emotion and was misled by it. Between the two lies a small gap my trade keeps trying to fill. In that gap, a 40mm ball departs, spins exactly as many times as no spreadsheet can measure, and flies toward a person with 0.15 seconds to decide.

The fate of the match lives in those 0.15 seconds. That is why I am still awake at nearly four in the morning, staring at an empty spreadsheet, finding it more beautiful than any number.