Vietnam's Table Tennis Data Void: When the Model Has Nothing to Read

Hook

2:14 a.m., a day in mid-June, in Hai Phong. I reopened the CSV file I had built for the national table tennis season that had just ended. Twenty-seven columns, neatly named: player, round, game-by-game score, points won on serve, average rally length, win rate in the 9-9 stretch, number of tactical switches between games. Everything exactly as designed. Below the header row, not a single line of content.

I had not forgotten to enter it. I had entered it for eleven days, from paper scoresheets, from screenshots of livestreams, from three phone calls to an acquaintance working at the venue. Then I deleted all of it. Not one row could be traced back to a source solid enough that I would defend it in front of an editor. An empty data file is an event, and it says more than any full statistics table I have ever published.

Context

Vietnamese table tennis has results. SEA Games editions, national championships, Southeast Asian head-to-heads all leave behind medals and names. Nguyen Anh Tu, Dinh Quang Linh, Tran Tuan Quynh, Mai Hoang My Trang, Nguyen Khoa Dieu Khanh — names that appear steadily in the press.

What Vietnamese table tennis does not have, and does not hide that it lacks, is a match-level data layer. I want to state that precisely: it is not a shortage of results, it is a shortage of structure inside the results.

In Europe or China, a professional table tennis match generates three data tiers. The first tier is the result: who won, by what score. The second tier is the metric layer: share of points won on serve, share of points won on receive, points claimed inside the first three shots, distribution of points by table zone. The third tier is movement data: stance position, reaction time, racket trajectory. The ITTF's WTT system, running since 2026, has standardised the first tier and most of the second for its own event circuit.

Vietnam has the first tier, scattered across news items. The other two tiers are effectively empty. When a Vietnamese player loses 2-3 in the qualifying rounds of an international event, I know the score, but I do not know what percentage of points he won on his own serve in the fifth game. In another sport, people would call that a small gap. In table tennis, it is the loss of the entire story.

Based on my own experience following matches, from domestic arenas to international events on live streams, one observation keeps repeating: Vietnamese table tennis viewers remember a great many fine rallies, but they retain none of them as data. The collective memory of this sport is rich; its record is poor.

I say this from a very specific place. Back when I hand-recorded a table for a domestic football season, I got hundreds of cells wrong. My first V.League dataset had hundreds of errors, but it taught me cleanliness better than any course. That lesson bites harder in table tennis, because the density of decisions is far higher here: an 11-point game, an average rally under four seconds, and serve metrics governing nearly half of all points.

Core

Start with the simplest and most neglected metric: points won on serve. At the elite level, this rate usually hovers around 55% for men and slightly lower for women, depending on playing style. Two-winged attackers live on short serves combined with the backhand flick, while pips styles produce flat trajectories and broken rhythm, pulling that rate down but raising the share of points won on receive in return.

Without that metric, every judgement about a Vietnamese player is merely formal description. People praise a player for a 'spinny serve'. But spin is not a measurable quantity. The measurable quantities are: how many points that serve won, against which opponent, in which game, at what score.

The second metric is rally length. This is what I use to classify playing styles, faster than any label. An average rally under 3.5 shots means the player wants to finish inside the first three shots. Above 5.5 shots means they accept length, pushing the opponent into a far-from-table defensive state. At a national event with no rally-length data, the argument between 'attack' and 'counter-attack' becomes an argument about taste, not about outcomes.

The third metric is performance in the decisive stretch. In table tennis, the gap between 9-9 and 10-9 is one point, but the psychological weight is entirely different. A player who wins 48% of points across the match but 70% of points from 9-9 onward is a fundamentally different player from one whose two rates are reversed. Without this data, we unconsciously reward the beautiful player in the middle of the match and punish the solid player at the end of it.

I once did exactly that, unconsciously. Before the 2026 World Cup, I ran a regression on 500 international matches and produced a 78% probability that Germany would reach the semi-finals. The reality: Germany lost 0-2 to South Korea and finished bottom of Group F with three points. I went back through all the footage and counted 12 counter-attacks leading to goals conceded by that team, the highest figure among the eliminated sides. The 2026 World Cup taught me one thing: the model did not collapse, I was the one who had believed it absolutely. Historical data could not measure the midfield's unwillingness to run. In table tennis, the equivalent of 'not running' is footwork speed in the two shots after the serve, and nobody records it.

The fourth metric is the head-to-head coefficient. In table tennis, head-to-head is not simply win-loss. The same player against the same opponent produces different results depending on the rubber. A long-pips player can completely disrupt an opponent ranked above them, while losing easily to a flat-looping one. This is why the concept of 'ky ro' — a bad stylistic matchup — exists in Vietnamese, and also why it is useless when it is not quantified.

The key point: a data shortage does not make judgement more humble; it makes judgement more confident, and dangerously so. Fans do not say 'I do not know'. They say 'I can see it clearly'. And because nobody can cross-check, that statement is never rebutted with numbers.

The fifth metric, and the one I track privately: ranking-points defence pressure. The ITTF and WTT ranking system runs on a rolling 52-week mechanism — old points expire, new points replace them. For a player holding a good result at a major event, the pressure is not winning the next title but avoiding the loss of points already banked. That is a scheduling-management variable, and it barely appears in Vietnamese table tennis commentary.

I also track the injury-and-return variable. A player's return schedule is usually controlled by the team's communications department, and the phrase 'wait until the weekend' in a release means something very specific: the injury has not healed. In table tennis, shoulder and wrist injuries do not need strapping to be visible, but they completely change the quality of the backhand. A player with wrist pain reduces spin amplitude and increases the rate of balls hit long. Without trajectory data, viewers only see 'a bad day'.

Then there is the youth tier. In any country with a strong table tennis base, U15 selection rests on a measurable index set: reaction speed, backhand stability, the ability to hold rhythm in long rallies. Vietnam selects through direct observation and paper notes. That method is not wrong in principle, but it produces no record comparable across years. A 13-year-old noted as 'promising' in 2026 still has only that note in 2026. Nobody knows whether they improved or regressed, because there is no benchmark to compare against.

Finally, the relationship between player and personal coach. I have long held that talent-evaluation models exaggerate young potential and underweight dressing-room chemistry. In table tennis, the equivalent body of that chemistry is the stroke structure handed down by a coach. A young player changing coaches can take six to twelve months to rebuild that structure. Without longitudinal data, that period gets read as 'stalling', and talent gets written up as failure.

I read a player through thirty variables before I listen to a commentator. But if those thirty variables do not exist, what am I reading? I am reading my own memory, and memory cannot be cross-checked.

Contrarian angle

The easiest conclusion from all of the above is: Vietnam needs more cameras, more software, more note-takers. I disagree with that conclusion.

Missing data is not the root problem. The root problem is missing provenance. A tournament can generate hundreds of thousands of data points while producing not one that can be traced back: recorded by whom, at what time, under which definition. Data without provenance is worse than data that does not exist. The first creates a feeling of certainty; the second forces us to admit we do not yet know.

Data does not need my belief. Data needs my checking.

There is one more counter-intuitive point. I once believed that the more digitised a sport becomes, the better it is for the athletes. Looking at esports, where every decision leaves a trace, I see the downside: players are professionalised into assembly-line products, and individual style is sanded smooth inside digitised training. Vietnamese table tennis escapes that risk because it is not yet digitised. That is only an accidental delay, and it will vanish the moment a platform strong enough appears. The task is to build the data tier before building the optimisation tier, so that the right to read the data belongs to the players, not only to the buyers.

Takeaway

That empty CSV file is still on my machine. I keep it because it is the cleanest evidence of one thing: in Vietnamese table tennis, what is missing is not talent, nor is it audience. What is missing is a record thick enough to argue with facts instead of with a louder voice. If a Vietnamese player walks into an international event next year and I still have only the score to tell, then every analysis will remain memory — and memory protects no one.