Vietnamese Table Tennis: When Data Forces Us to Re-examine the Illusion on the Table

The Moment the Scoreboard Lied

I was sitting in the seventh row of a domestic arena, notebook open, and I saw something the electronic board never said. Third set, 11-9 to the favourite. The stands applauded. The commentator spoke of composure, of a moment of brilliance. But if you counted point by point, if you recorded every serve, every rally, every redirection, you would see an entirely different story. That winner of the third set won only 5 of 14 points in rallies lasting four shots or more. The loser had controlled the rhythm for nearly two-thirds of the time. The winner survived on six consecutive short points from serve, a form of advantage with an extremely low repeat probability.

That was the moment I realised something I keep telling my students: a table tennis match is not told by the set score. It is told by the structure of the points. And in Vietnam, we have almost never bothered to read that structure.

Numbers never lie; only those who read them deceive themselves.

This article emerged from a strange void. I was asked to analyse a table tennis data file, and when I opened it, I found that nearly every information field was empty. No tournament name, no athlete name, not a single quantitative data point. A complete scaffolding shell with a hollow interior. For a writer in the data tradition, that is a professional nightmare: you have every tool, but nothing to measure. And rather than invent a subject to fill the page, I chose to do something more honest — to write about the void itself. Because the data void in Vietnamese table tennis is, in the end, itself a finding.

Context: A Table Tennis Culture Running on Memory, Not Spreadsheets

Vietnamese table tennis has an interesting paradox. It is a sport where we have tradition, regional achievements, and names written into the collective memory of generations of fans. But it is also a sport where our data infrastructure is alarmingly thin. We remember who won, we remember the matches of a lifetime, we remember the shots passed down through storytelling. We have almost never preserved the point-win rate on serve, the distribution of rally lengths, the efficiency at decisive points.

Imagine this in football. What would it be like if the Premier League had no xG, no passing numbers, no heat maps, only scores and emotional commentary? That is roughly the state in which Vietnamese table tennis operates. A sport told through feeling, judged through impression, selected through the eye of a few trusted authorities.

I do not deny the value of that eye. A veteran coach watching an athlete train for five minutes can see what my model takes a week to quantify. But that eye has its limits. It is biased. It is anchored to impressive moments. It remembers the beautiful backhand loop more than the steady success rate of the short push. And in a sport where each point lasts only seconds, human memory is a particularly suspect tool.

There is a saying I like: data is the confession of those who once believed in feeling. Vietnamese table tennis, at this moment, is still in need of such a confession. Not to deny feeling, but to test it.

The structure of a modern table tennis match — played to 11 points, best of four sets at major events, with a 40+ mm plastic ball — has changed how points are distributed. Since the rule change from 21 points to 11, every point has become heavier, every serve a tactical weapon, and the margin of error between winning and losing has narrowed considerably. In an 11-point set, just two random points landing the right way can shift the entire picture. This means that the further we go, the more important it becomes to read the structure of points rather than the score. And that is where data must speak.

The problem of Vietnamese table tennis is not a lack of talent. The problem is a lack of language to describe that talent objectively. We say a player has "competitive composure", but we do not define composure by the win rate at decisive points. We say a player has "a good physical base", but we do not measure average footwork distance per set. We say a player is "unstable", but we do not separate technical variance from psychological variance. Every adjective we use conceals an assumption that has never been tested.

Core: Dissecting a Table Tennis Match Through Numbers

When I began building a set of indices for table tennis, I started from a single question: what actually distinguishes the winner from the loser, beyond the final score? The answer does not come from a single index, but from a cluster of indices interacting with each other. I will go through each layer.

Layer One: Serve and Receive Efficiency

In table tennis, the serve is the only phase a player fully controls. No opponent interference, no time pressure from the other side. Therefore, serve efficiency — the win rate when serving — is the index most directly reflecting a player's technical and tactical quality.

But here the first trap appears. Many assess a good server by how many points they win directly from serve. This is a flawed reading. A serve may not win directly, yet it may create a return that is favourable for the server to attack in the next phase. That is a form of indirect advantage, and it only appears when you track a sequence of points rather than individual points.

I once tracked a young player across about ten consecutive matches. Raw statistics showed he won only about 12% of points directly from serve — a seemingly unremarkable figure. But when I widened the observation window to the two phases following the serve, his point-win rate shot up to nearly 60%. In other words, his serve was not meant to finish, but to launch an attacking sequence. If you look only at the direct figure, you will draw an entirely wrong conclusion about its value.

On the opposite layer, receive efficiency — the win rate when receiving — is usually read alongside a secondary index: the safe-return rate. A good receiver is not the most aggressive attacker, but the one who minimises errors in the first return. At the elite level, most points lost on the receive come from technical errors, not from the opponent playing too well. This sounds obvious, but it changes how we assess: a player who loses many points on the receive may not be facing a great server, but may simply be losing the points themselves.

Layer Two: Rally Length and Match Rhythm

I classify every rally into three groups: short (1-3 touches after the serve), medium (4-6 touches), and long (7 or more). The distribution among these three groups creates a fingerprint of the match, and it often contradicts the audience's perception.

In the match I mentioned at the start, the winner of the third set won most of his points in short rallies. This means his victory depended on serve quality and first-phase execution. But when he moved into the fourth and fifth sets — where fitness and focus decline — his win rate in short rallies dropped markedly, while the opponent won in long rallies. The rhythm had shifted, and the third-set winner turned out to be the slower to adapt.

This is what the set score never reveals. A won set can conceal a trend turning against the winner. A player who wins 11-9 through six lucky short points is in a worse position than one who loses 9-11 but controls the rhythm. If you read only the score, you are reading half the story.

Rally-length distribution also reveals a player's style. A fast attacker will try to pull the match toward short rallies. A defender or a mid-distance player will try to extend rallies. The real tactical battle does not take place in the score, but in who controls the rally-length distribution. When a match's short-rally share rises, the one who wants to extend is losing the rhythm battle — even if the score may still favour them.

Layer Three: Performance at Decisive Points

This is the index I consider most misunderstood in Vietnamese table tennis. We call it "composure", and we treat it as an innate psychological quality. I do not believe that. I believe what we call composure is largely the result of structured practice, and it is fully measurable.

I define a decisive point as one where the leader can close the set if they win, or the trailing player can equalise if they win — that is, points from 9-9 onward, or set points. In my tracking sample, performance in this group shows enormous differentiation between players of comparable technical level. Some win 70% in the early part of a set but only 45% at decisive points. Others do the reverse.

Notably, this gap is not as stable as we assume. It depends heavily on whether the player has a "process" for the big points — a fixed sequence of actions from receiving, choosing tactics, to execution. Players with a clear process maintain more stable performance across decisive points. Players who rely on inspiration tend to fluctuate widely.

This is why I always tell coaches: do not tell your student they "lack composure". Tell them they "haven't yet built a process for the big points". The second phrasing points to a solution. The first only creates an accusation.

Layer Four: Age Structure and the Career Curve

A dimension almost entirely ignored in Vietnamese table tennis is the career curve by age. In many sports, analysts identify peak age and build development pathways on it. Table tennis is peculiar: it is a sport where tactical experience matters almost as much as fitness, while peak reflexes are reached quite early.

This creates an interesting golden window. In pure reflexes, young players have the edge. In reading the game, choosing placement, and managing rhythm, older players have the edge. World-class winners typically fall at the intersection — where both factors remain strong enough.

When I compared this curve with the Vietnamese reality, I found a systemic issue: we usually assess a player by recent form rather than their position on the age curve. A 19-year-old on the ascending slope is rated on par with a 27-year-old at peak, simply because their recent results are comparable. This means we invest in the wrong place: we devote resources to those at peak rather than those with room to grow.

I once staked my reputation on a bet, and football answered with data.

I use that line for football, but it holds for any sport with data. The issue is not the gamble, but accepting that some predictions only come true when you read the curve correctly, not the moment.

Layer Five: Stability Across a Match Series

One match is a noisy sample. Five matches are a trend. Ten matches are a model. This is a basic principle that Vietnamese table tennis rarely follows, because we lack the infrastructure to preserve series data systematically.

With series data, you can separate two types of variation. The first is healthy variation — the natural fluctuation of form around a stable ability level. The second is abnormal variation — a sign of a technical problem, an injury, or a psychological crisis. The two look alike if you see only one match. They differ sharply if you see ten.

In table tennis, the technical-variance index — the degree of fluctuation in the success of a specific shot (say a forehand loop) — is a powerful diagnostic tool. A player with an average forehand-loop success of 65% but ranging from 40% to 85% is an entirely different player from one with an average of 65% ranging from 60% to 70%. The second is more reliable in big matches, where stability matters more than peak.

This leads to a conclusion unfamiliar to many in Vietnamese coaching: at the elite level, the value of stability often exceeds the value of individual peak. A player with a lower ceiling but higher floor can deliver more points across a long tournament.

Contrarian Angle: When the Model Deceives Itself

At this point I must tell a story I often use to remind myself. At twenty, I built a prediction model based on serve efficiency and rally-length distribution. The model performed well in most matches. Then it failed catastrophically in a match I had confidently predicted on social media. I spent the whole evening figuring out where I went wrong.

The lesson I drew — and the one I want to give anyone building an analytics system for Vietnamese table tennis — is this: a model has no bias, but the person building it does. And that bias seeps in through the most innocuous decisions: which variables to measure, how much weight to assign, which outliers to discard.

There is a phenomenon I call the "pretty-number trap". When you have a clear index, you tend to prioritise it over harder-to-measure ones. In table tennis, the point-win rate is an easy number to measure. But what decides long-term outcomes is not the point-win rate of one match, but the structure of repeated decisions. A player can win through wrong decisions that got lucky. They will lose the next match if they repeat them. If your model reads only outcomes, you will overrate someone you should underrate.

This is where video analysis combined with data becomes indispensable. Numbers tell you what happened. Video tells you how it happened. Only by placing the two together can you distinguish the result of a healthy process from the result of a statistical accident.

There is another counter-intuitive angle I want to put on the table: the assumption that a stronger serve is always better. In a small study I ran for myself, I found that very difficult serves — those that force a weak return — correlated with more points won in the short term, but did not correlate with long-term match-win rates in some cases. The possible reason: when a serve is too difficult, the server also loses the ability to predict the return trajectory, and falls into a passive state in the second phase.

In other words, there is an optimum for serve difficulty, not a case of the harder the better. And that optimum differs by opponent, by table surface, by climate conditions. This is the kind of understanding that only long-series observation can provide.

Form is an illusion; only the sequence of numbers is the true current.

What the Data Does Not Say

I want to devote this section to the limits of the very method I apply, because an analyst who does not admit their limits is a dangerous analyst.

First, data cannot measure meaning. When a player reaches a final after ten years of waiting, every point they win carries a weight that cannot be quantified in percentages. Table tennis is a sport of compressed moments, where a single point can shape an entire career. No model captures that, and I think none should try.

Second, data cannot measure randomness. In a sport with a light plastic ball, complex spin, and reaction distances measured in fractions of a second, every point contains a large amount of randomness. That is why I never claim a model predicts the exact result of a specific match. I only claim it correctly describes the structure of a large sample. Between describing structure and predicting results lies a professional ethical distance I try not to cross.

Third, data cannot measure a person's development. A player today may be at their ceiling, but in three months may surpass it through a small change in training. Models based on the past tend to miss leaps, because they assume continuity. In reality, table tennis develops in leaps, not straight lines.

And finally, the most important: data cannot measure what a player wants to become. I have met athletes with unremarkable indices but an iron will, and over years that will transformed their indices. My model never predicted that. People often overtake the numbers toward themselves. And that is why I always end my data analyses with a compulsory humility.

Signals for the Next Cycle

If Vietnamese table tennis wants to enter an era of data-driven operation, I think the starting point is not complex models. The starting point should be a simple notebook: record every point, classify it by rally length, by winner, by stroke type. Do this persistently for one season, and you will hold an asset money cannot buy. You will have a rhythm model of each player, a map of each opponent's strengths and blind spots.

What I most look forward to in the next cycle is not a perfect model. It is an environment where hard-to-hear numbers are welcomed rather than defended against. Where a coach can tell a student: "Your serve index is declining, let's review it together," without being seen as betraying their talent. Where data serves people, rather than forcing people to serve data.

When the world cries out for emotion, I choose to listen to the chart.

I wrote this article from an empty file. No match to analyse, no player to assess, no tournament to position. Only a shell and a void. But perhaps that void is the most honest message: Vietnamese table tennis has many stories, many moments, many talents — but we still lack the tools to record them as knowledge. There is a rhythm on the table we have never recorded. There is a structure we have never drawn.

And when table tennis stops rolling between competitive cycles — what I still call the summer of truth — perhaps this is the time to start recording. Not to prove that numbers tell the truth, but to prove that we are ready to listen, even when they say what we do not want to hear. The football-free summer is when truth emerges, without the media smoke. And perhaps it is also the time for Vietnamese table tennis to ask itself one question: if every point were recorded, what would be the first thing we discovered about ourselves?