The Empty Data Sheet and the Temptation of Table Tennis Analysis

The Empty Data Sheet and the Temptation of Table Tennis Analysis

Saigon, a rainy August afternoon. I am sitting in my usual cafe in District 3, a laptop open in front of me with a spreadsheet from a recently concluded WTT Contender event. The player-name column is complete. The nationality column is complete. The ranking-points column is complete. The three most important columns — service-point win rate, receive-point win rate, and win rate in rallies lasting more than seven contacts — are blank.

I take a sip of coffee, look at the screen, and wait.

Numbers know how to hold their breath, and I wait for them to exhale. But this time they do not exhale. They just sit there, silent, like a player standing at the far end of the table waiting for an opponent's serve without knowing which way the spin will go. In table tennis, you are taught never to guess the spin — read it from the opponent's wrist. But when the opponent has not served yet, the only honest thing is to stand still.

A newcomer to the trade will type a few lines to make the deadline. Someone who has been in the trade long enough understands that silence here is a professional decision, not a dead end. My data cafe is busiest when the arena is empty, and that afternoon I was its only customer.

Context: The table tennis data ecosystem, where it is full and where it is empty

Table tennis has a public data ecosystem that is better than many team sports in the region. The ITTF publishes a world ranking every week, refreshed on Tuesdays. WTT runs a clear tournament tier system: WTT Finals, WTT Champions, WTT Star Contender, WTT Contender, WTT Feeder. Each tier carries different points and prize structures. Fans can look up any player's points, win-loss record, and head-to-head record.

But that is the surface layer. Deeper down — where it is decided whether a player wins through structure or through luck — the data becomes strangely sparse. Service-point win rate by spin type. Rally-length distribution. Win rate in rallies exceeding nine contacts. Conversion rate at 9-9 or 10-10. These are metrics that analytics departments of major federations track, but they are rarely published in full, and they barely exist at domestic tournaments.

I once worked as a data consultant for a football club in Ho Chi Minh City, and the first lesson I learned had nothing to do with software. It was about the question: does this data actually exist, or am I imagining it?

In 2026, I wrote a long analysis of the team's winning streak, pointing out that the opponent's PPDA was only 8.2 — they deliberately abandoned pressing to counter-attack. I used V.League xG to show that the streak was more luck than quality, with a gap of 4.7 goals between actual goals and xG. The piece spread widely. But what I remember most is not the shares — it is the fear I felt when I rechecked the spreadsheet and realised I had nearly cited a metric with no traceable source.

Since then, I have set one rule: every conclusion must be anchored to a specific data point. No data point, no conclusion. It sounds simple, but most of the sports analysis trade operates the opposite way.

The core: when the sheet is empty, the analyst faces four temptations

The first temptation is to fabricate. With no service data, people write "this player's serve is very awkward to handle." That is not data. That is a feeling wearing a data costume. I call it disguised data — a form of professional debt the writer borrows against his own credibility.

The second temptation is to borrow from another sport. Taking football's PPDA and applying it directly to table tennis. Taking football's xG and attaching it to a table tennis rally. Every sport has its own data structure. Table tennis has no "possession" in the football sense. It has rally length, direct service winners, and the share of points won while trailing. Applying a metric born in another sport to table tennis is methodologically wrong at the root.

The third temptation is to dress the eye test in data clothing. You watch a match, you feel player A played better, and then you go looking for numbers to justify that feeling. This is the most dangerous error for anyone with match-watching experience, because the eye is convinced before the brain checks. Based on my experience watching matches, a player can win a match with three lucky points at decisive moments and look like "character" on television. In the point-by-point dataset, that is random variance, not ability.

The fourth temptation is to extrapolate from too small a sample. A WTT Contender gives a champion three or four matches. A domestic event can give fewer. From four matches, you cannot conclude anything about "form across the year." You can only conclude about those four matches.

These four temptations share one mechanism: they fill the gap with confidence, and confidence is not data. In table tennis, when a player cannot read the spin, they should not swing blindly — they should push the ball safely and wait. The analysis trade works the same way. When the data will not let you read the spin, the most honest thing is to say you have not read it yet.

That sounds like an admission of weakness. In practice, it is a capability. Mature analytics departments all have someone I call the "empty-sheet gatekeeper" — a person with the authority to say no to an attractive report with no foundation. In many federations, that person is a statistician, not a coach. And their voice is often drowned out by pressure from media, management, and the fans themselves, who want a story to tell.

Three metrics worth reading, and three gaps worth admitting

If I had to pick three metrics to assess a player, I would not pick win count.

One is service-point win rate, split by spin type. This metric tells you whether a player has a genuine serving weapon or is simply serving safely to reach the rally. The difference between these two player types rarely shows in the scoreline, but it shows very clearly in the metrics table once the sample is large enough.

Two is the win rate in long rallies. In modern table tennis, with a bigger ball and higher speeds, short rallies account for most points. But long rallies are where physical foundation, consistency, and decision-making under pressure are exposed. A player who wins many short-rally points but loses the long ones is usually someone relying on early luck, and will be figured out once an opponent extends the match.

Three is the win rate at decisive moments — from 9-9 onward. This metric is small, its sample is usually small, and that is precisely why it is dangerous when misused. But tracked across many events, you will see that players with a stable rate here differ clearly from those who shine for one tournament.

All three share a common feature: they require point-by-point data, not just final scores. And because they require point-by-point data, they do not exist at most table tennis events, especially domestic ones. This is a genuine gap in the industry, not a gap caused by laziness. To have metrics, someone must sit and record every point. To record every point, you need a system. To have a system, you need investment. And investment in table tennis data in Vietnam remains modest.

At the top international tiers, the picture is brighter. Players such as Ma Long, Fan Zhendong, Wang Chuqin, Tomokazu Harimoto, Truls Moregard and Hugo Calderano are tracked by multi-camera systems, and many of those events now release detailed point data to the public. At Vietnamese domestic events, where Nguyen Anh Tu, Dinh Quang Linh, Mai Hoang My Trang and Nguyen Khoa Dieu Khanh compete, the data mostly stops at the game score. The distance between these two layers is not only a technology gap. It is a gap in recording habits.

Rankings, draws and head-to-head: the three most misread places

There is one specific trap I see repeated in table tennis media: the world ranking is mistaken for actual strength. A player ranked tenth in the world is not necessarily stronger than the player ranked fifteenth at every moment. The ranking is the output of a points system weighted by event tier, with a retention window and pressure to defend points. A player can drop in the ranking simply because old points expired, not because they lost repeatedly. This is verifiable from the official ranking table, yet it is routinely skipped in quick commentary.

When the ranking is mistaken for strength, everything downstream goes wrong. You predict draws based on seeds. You judge form based on rank. You build narratives based on hierarchy. And you forget the basic question: which events, over which period, produced this ranking?

Head-to-head is the second easiest place to err. A 5-2 record in favour of player A sounds persuasive. But you must ask: over how long did those seven matches take place? Were any of them before player B changed rubber or changed playing style? Were any of them at small events while one side was injured? Head-to-head only means something when both sides are at the same developmental stage and in the same physical condition. Otherwise it is a pretty number with no meaning.

The draw is the third easiest place to err, and the most ignored. In a knockout event, a player landing in a half with fewer strong opponents is not tactics. It is tournament structure. But it directly affects the result, and it never appears in the "transformation" story. A player can reach the semi-finals without facing a single seed, while another is eliminated in the quarter-finals after beating two seeds. Looking only at the final result, the first looks more successful. Looking at the data, the story reverses.

The counter-intuitive angle: correlation is not causation

When a player wins a title, the media tends to build a story: this person has transformed, has found a winning formula, has overcome themselves. These are easy stories to sell. But they usually ignore a simple variable: draw structure and schedule.

The correlation between winning a title and being in top form does not prove causation. It only proves that the two appeared together. And in knockout sport, two things appear together for many reasons — including reasons that have nothing to do with form.

This is the biggest blind spot of the sports analysis trade, and table tennis is not outside it. When you see a player reach the semi-finals, you must ask: how many seeds were in their half? How many of their opponents had beaten them at previous events? Did they have to play seven games while their main rival played five in the previous round? These are structural questions, and they are usually brushed aside because they are less appealing than a story about character.

I once had a microphone moment. In 2026, I was invited to work as a data commentator for a major international event. In the opening match, I was so excited that I mispronounced the name of Russia's striker three times in the first half. Viewers reacted sharply. I was so ashamed I wanted to vanish from the broadcast. But instead of quitting, I spent the following month reviewing every match tape, taking notes on each team's pressing metrics. In that process, I found that Croatia's defensive model had gaps behind both full-backs — something almost impossible to see with the naked eye at live broadcast speed. I used to fear the microphone; now I let the data speak for me.

Old footage is a mirror, and only those who dare to look will see themselves. Had I not gone back to review what I had said wrong, I would not have found that gap. And had I not found it, I would have remained the man who mispronounced a name without understanding why.

I tell this story not to talk about football. I tell it to talk about table tennis, because the mechanism is the same. The crowd looks at a rising player and assigns them a story. The data person looks at the point-by-point sheet and checks whether the story left footprints in the numbers. If it did not, it is a story — not a fact.

What is changing, and what to be careful about

Table tennis's data market is showing small but steady movement. WTT is publishing more metrics for top-tier events. Match-tracking apps are beginning to provide point distribution by game. Streaming table tennis with more camera angles makes motion data more feasible — we can measure travel distance, serving position, and reaction time after the ball leaves the opponent's racket.

But this is also the danger point. As data becomes more plentiful, temptation grows with it. People easily assume that having data means having conclusions. It does not. Having data only means having ingredients. Cooking is a different matter, and a poor cook can still ruin good ingredients.

There is another trap I see in transfer windows, when noise overwhelms signal. In professional table tennis, clubs and federations disclose contracts less than in football, but the pressure for results is no less. When a young player is promoted to the national team, the right question is not "is this person talented," but "how many matches has this person played at what level, against what opponents, and over how long." An eighteen-year-old winning a junior event is not the same as an eighteen-year-old winning three matches in the qualifying rounds of a WTT event. But in the newspapers, both are "young talents."

That is why I always return to the question of data sources. Every number is a puzzle piece, but I do not assemble them out of habit. Before assembling, I must know where that piece came from, how it was recorded, by whom, and under what conditions.

Conclusion

Table tennis data is at the stage Vietnamese football passed through about fifteen years ago: enough people care, not enough people record. The biggest gap is not in technology. It is in habit — the habit of tolerating an empty sheet without rushing to fill it with a story.

A ranking can tell you who is leading. It cannot tell you who will win the next match. A head-to-head record can tell you who has beaten whom. It cannot tell you who is improving faster over the past six months. The truly important questions often lie where the data has not yet been recorded, and that is exactly where the professional must learn to say "I do not know yet" before learning to say it more elegantly.

When someone asks me to predict a table tennis match for which I have no point-by-point data, my most honest answer will be: I do not know yet. This industry needs to learn to say that a little more often, before rushing to tell beautiful stories about numbers that never existed.