The Data Gap in Vietnamese Table Tennis: When a Blank Scoresheet Is Itself a Signal

On the referees' table at a national table tennis championship final, the scoresheet has exactly seven columns: player name, shirt number, set, point, service fault, receive fault, and the referee's signature. No column records where the third ball of the left-hand player's serve landed. No column notes how far that player retreated from the table edge after losing the seventh rally. No column states how many seconds the decisive exchange lasted, or who changed direction first. When the applause dies, the sheet is signed, folded, and put away. The entire body of the match — the part that explains why the score was 4-3 and not 4-0 — vanishes from history at that exact moment.

I sit in the fourth row, with a ruled notebook and a pencil, and I redraw it. That has been my habit for ten years. But this season, while completing the last arrows of the sixth set, I realised something very different from everything I had assumed. The problem with analysis in Vietnamese table tennis is not that we lack data. The problem is that the lack itself has become a data point, and almost nobody bothers to read it.

A total void, not a partial one.

That means: no sample to compare against, no baseline to deviate from, no average to be measured against. In professional sports analysis there is a term for this — the null payload. A null payload is more dangerous than a bad payload, because a bad payload at least raises an alarm, while a null payload stays silent. It passes through the system without a sound, and the people at the far end — coaches, commentators, parents, sponsors — receive an empty table and read it as one sentence: there is no problem.

That is the most expensive mistake in Vietnamese sport, and it repeats across every discipline, not just table tennis.

Context: Where Vietnam's table tennis data infrastructure stands

Vietnam's table tennis system runs on three tiers: the national competition system run by the federation, the provincial and club training centres, and the international selection tier. All three share one trait: they generate enormous amounts of information, and almost none of it is stored in a queryable form.

A national round can contain hundreds of matches in a single week. Each match, if I sit and chart it, yields roughly forty basic variables: serve placement, spin type, third-ball win rate, return direction, number of exchanges, who changed direction first, rally length, number of retreats from the table, number of crossover steps to the left, number of forehand pivots, win rate when leading by two points, and a dozen more. Multiply that out. One season could produce a dataset rich enough to build form models for at least twenty players.

What do we actually get? A list of scores posted on the federation's portal. Sometimes photographs. Occasionally a shaky phone clip filmed from the stands, where you cannot see the feet.

The gap between what could be collected and what is collected is a factor of one hundred. That is my own estimate, not an official figure, and I state that clearly so readers know its confidence level.

Mechanism: Why the gap appears, and why it sustains itself

There are four causes, and three of them have nothing to do with money.

The first is budget. A national-level table tennis match cannot afford a dedicated statistician. Everyone cites this cause, and it is true, but it explains only a quarter of the problem.

The second is professional habit. Vietnamese table tennis coaches are mostly former players, trained by eye and by hand. Their expertise lives in motor memory, not in a computation file. This is not inherently bad — it is a different form of data — but it cannot be copied, cannot be transferred, and disappears when that coach retires.

The third is the culture of the scoresheet. In Vietnam, the accepted evidence of a match is the score. Whoever wins is remembered. Nobody asks how. When an entire system agrees that the score is sufficient, deep data collection becomes a strange, redundant, time-wasting activity.

The fourth cause, and the one worth discussing, is the self-sustaining loop. With no data, nobody dares to conclude. Without daring to conclude, nobody sees the value of data. Seeing no value, nobody invests the effort to collect it. The loop closes and runs so smoothly that insiders mistake it for the natural state of the sport.

I lived inside that loop for years, until the pandemic season, when every tournament stopped and I was forced to work with what I had long ignored: the fragments in my notebooks.

Core analysis (1): What a table tennis data model contains

To avoid the trap of pure storytelling, here is the structure I use when I chart by hand. It has five layers, and every layer can be collected with a pencil when no camera is available.

The first is the serve layer: stance position, spin direction, trajectory length, first bounce, second bounce, and speed. Six variables, enough to build a serve map for one player after roughly three matches.

The second is the third-ball layer. After serving, does the player attack, push, or block? Selection rate, and win rate for each selection. This is the most important layer in modern table tennis, and the most neglected in Vietnam.

The third is the receive layer. What does the opponent return with, where to, and who takes the initiative in the fourth ball.

The fourth is the tempo layer: rally length in touches, time between touches, and the number of defensive-to-offensive switches within a rally.

The fifth is the movement layer: number of steps, direction of steps, and lateral court coverage.

Stacked together, these five layers produce something the scoreline can never express: the structure of a match.

Every hand-drawn diagram is a story the numbers cannot tell.

Core analysis (2): Serve and third ball — where the match is actually written

At elite level, roughly seventy percent of points are decided within the first three touches. If that holds, most analytical effort should go into those three touches, not into the long exchanges that dominate television but account for a smaller share of points.

When I chart matches from the later rounds of the national championship, a fairly stable pattern emerges. The group with the highest third-ball win rates is usually not the group with the strongest serve. It is the group with the least predictable serve — meaning their placement distribution spreads across all four zones, and the ratio of short to long serves does not deviate far from one-third to two-thirds.

Conversely, the group with the strongest serves — heavy spin, high speed — tends to win more points directly in the opening exchanges, but falls behind on the fourth and fifth balls once an opponent handles the serve. The reason is specific: a heavy serve requires a more tilted torso position, and that position makes the recovery step after serving about a quarter-beat slower.

A quarter-beat, in table tennis, is the distance between a winning loop and a loop that gets blocked back.

The point is not that heavy serves are bad. The point is that serve selection is always a trade-off, and a trade-off can only be measured if data exists. Without data, a coach can only rely on feel — and feel in table tennis is easily deceived by two or three beautiful rallies early in a match.

Core analysis (3): The receive layer — what the scoresheet hides

If the serve is the question, the receive is the answer, and the quality of the answer determines who gets to ask next.

At Vietnamese training centres, I observe a fairly common tendency: players train the receive by training the ball to land on the table. The goal is safety. But in modern table tennis, a safe receive against a good attacker is equivalent to inviting that attacker to strike. A good receive is not one that is hard to miss; it is one that prevents the opponent from attacking on the correct beat.

There are three types of receive that meet that standard, and I classify them by purpose rather than technique.

The first is the tempo-stripping receive — slow, short, forcing the opponent to step into the table before hitting. It costs the opponent about one foot-step and a third of a second.

The second is the angle-stripping receive — fast, deep into the open corner, forcing the opponent to expand their range. It costs about two foot-steps.

The third is the option-stripping receive — unusual spin, forcing the opponent to push back instead of attacking. It costs the opponent the freedom to choose.

These three have different effectiveness depending on each opponent's height and reach. A tall player with long reach will be far less affected by angle-stripping receives than a short, quick-footed player. Without data on reach and movement speed, a coach can only offer a general instruction: receive tight.

Core analysis (4): Tempo and rally length — the most forgotten variable

In a table tennis match, tempo is the least discussed variable, yet the easiest to observe if you sit close enough to the table.

The measurement is simple. Count the touches in each rally. Classify into short rallies (one to three touches), medium (four to six), and long (seven or more). Then calculate each player's win rate by rally type.

Across several seasons of hand-recorded data, the pattern I obtained is notable: most players in Vietnam's leading group have their highest win rate in medium rallies, not short ones. This contradicts conventional intuition, which holds that playing fast and finishing early is the shortest path to points.

My explanation is this. At domestic level, serves are not yet dangerous enough to generate direct points in short rallies at high frequency. But precisely because of that, players spend more time training medium rallies, and medium rallies become their comfort zone. This is a form of structural compensation, and it is only visible if you bother to count.

The pattern can reverse in international play. Against opponents with better serves, short rallies become more dangerous and the medium-rally comfort zone narrows. A national team stepping onto the regional stage without knowing its own win rate by rally length steps out with half a map.

Core analysis (5): Footwork — the part of the body the camera never shows

Of all the data layers described, movement is the hardest to collect with the naked eye, because it sits below the frame. Television cameras track the ball. They do not track the feet.

But the feet determine whether the hand arrives in the right place.

There are three basic movement patterns to distinguish. The first is the side step, used when the ball arrives close, with a small deviation. The second is the crossover step, used when the ball arrives far, with a large deviation, and it is the most energy-expensive. The third is the torso-rotation step, used in the backhand corner when a player must rotate the whole body to play a forehand from a position that belongs to the backhand.

In my records, the usage ratios of these three patterns differ markedly between player groups. Players with consistent results use the side step at a higher frequency — meaning they are already in the right place before the ball arrives, rather than running after it once it has passed. This is worth reflecting on, because it suggests that most of the advantage in table tennis is not created by foot speed, but by the quality of judgement before the ball leaves the opponent's racket.

In other words, fast feet cannot rescue slow judgement.

The smallest detail on the table always has a reason.

Core analysis (6): Equipment — a variable outside the table but inside the result

In table tennis, equipment is a genuine competitive variable, unlike most team sports. The blade determines feel and vibration. The rubber determines spin and ball speed. A change of rubber can alter a player's scoring structure for weeks.

The equipment landscape in Vietnamese table tennis has one feature I consider notable: two distinct schools exist. The first uses pimpled or anti-spin rubber on one side, leaning toward defence, blocking, and tempo disruption. The second uses sponge rubber on both sides, leaning toward spin-based attack from both wings.

The pimpled school once held firm ground in domestic youth tiers, because it allows a player with an underdeveloped physical base to achieve results quickly. But it has a clear ceiling. Once opponents are familiar with the unusual trajectory, the advantage disappears, and the pimpled player must switch styles at an age when attacking technique is already hard to restructure.

This is a structural decision, not merely a technical one. And it is exactly the kind of decision that long-term tracking data could support well — if such data existed.

Core analysis (7): The youth pipeline and the age curve

In table tennis, the peak age for an Asian player typically falls between twenty-two and twenty-eight. Before twenty-two, players often lack psychological stability at decisive points. After twenty-eight, reaction speed and between-rally recovery decline, though experience compensates in part.

This means the youth pipeline is not an abstract concept. It is a time window of specific width. A cohort trained from age twelve enters its peak zone about ten years later. If in the seventh year of that cycle you hold no data on which categories the player is improving and which are stalling, the final three years will pass without adjustment.

Three years, at the scale of one cohort, is a generation.

At national youth tournaments, there are more matches than at senior events, fewer spectators, and essentially zero record-keeping. This is the system's greatest paradox: the place that generates the most data is the place where the fewest people record it.

Core analysis (8): The event system and ranking points — data from outside

Another paradox: most of the data Vietnamese table tennis has about itself comes from the international system, not the domestic one.

The international federation's ranking system operates on accumulation and defence of points. Each event carries a different point value depending on tier and result. Points expire, and when they expire, they are deducted. This creates what is called points-defence pressure: a player must not only win to climb, but win at the right events to avoid slipping.

For Vietnamese players, most point-earning opportunities sit in the lower and middle tiers of the system, alongside regional events. The number of accessible events per year is limited, and travel cost is a real barrier. This produces a consequence worth stating plainly: a Vietnamese player's competition calendar is not decided purely by sporting merit, but by a set of budget and logistics constraints.

Understanding that changes how a reader views the ranking table. A ranking position does not only reflect ability. It also reflects how many opportunities that player was permitted to access.

Core analysis (9): The regional landscape and Vietnam's position

In Asia, table tennis is clearly stratified. The leading tier is China, with a depth no other nation approaches. The second tier includes Japan, South Korea and the Chinese Taipei region, all with complete youth development systems and large numbers of athletes on the international circuit. The third tier is the group of nations and regions with potential but limited depth, including Singapore, Thailand and India — where in recent years some players have reached world class through systematic investment in a few exceptional individuals.

Vietnam sits in the transition zone between the third tier and the tier below. We have players capable of competing in Southeast Asia and capable of troubling third-tier opponents in a single match, but not yet enough depth to sustain results across multiple events.

This gap is not a gap in talent. It is a gap in repeatability.

And repeatability, ultimately, is a measurable index. It simply is not measured in Vietnam.

Core analysis (10): A simulated example — one set read through data

To illustrate, I take a hypothetical set reconstructed from hand records I have made across many matches. The aim is not to describe a specific match, but to show how differently a set can be read.

Suppose the set ends 11-8. Player A wins. By the scoresheet, A won because A scored more points. That is all.

Read through a hand-drawn diagram, the story is entirely different. A served eleven times: five short to the left zone, three long to the right zone, two short to the middle, one fault. B won the point on the third ball four times — meaning on four of A's serves, A attacked first and lost. On the remaining seven, A chose to push on the third ball, and won five of seven.

The scoring structure of this set says A did not win by attacking. A won because he switched to pushing after realising the first attacking ball was ineffective, and B had no attacking answer to that push.

The scoresheet records that A won 11-8. The hand diagram records that A won because he corrected himself from the fifth rally onwards.

This is the kind of information a coach needs, and the kind a scoresheet never provides.

Contrarian angle: The trap of believing that no data means no analysis

At this point I must refute myself once.

The conventional argument runs: without data, analysis is impossible, so we must wait for data. It sounds reasonable, and it is the argument I myself used for years to postpone writing.

But it fails in one place. It assumes data must come from outside, supplied by someone, after someone has invested. In reality, table tennis data at national level can be generated by one person in the fourth row with a notebook. No camera, no software, no analysis department required.

The issue with that approach is quality, not quantity. And this is the genuine blind spot of Vietnamese table tennis: we do not lack the ability to collect data, we lack a cross-check mechanism to know whether the data we collected is correct.

A single person charting alone can err at three levels. At the observation level: misreading placement. At the classification level: filing a spinny push as a block. At the recording level: mislabelling one rally and carrying the error through the set. With no second independent recorder, none of these three errors can be detected.

The lesson, then, is neither to wait for data nor to assume that collection alone is good. The lesson is that data collected by one person must be cross-checked against at least one independent source — the referee's sheet, another person's record, or one's own record taken at a different time.

Without that cross-check, every model built on self-collected data is merely a prettier way of presenting a feeling.

At a deeper level, there is another trap worth naming: the small-sample trap. With three matches, one can draw any conclusion one wants. With three hundred matches, one can draw only the conclusions the data permits. Vietnamese table tennis is currently at the three-match stage, and the greatest risk is not the absence of conclusions, but the abundance of confidently drawn ones.

When the arena falls silent, I learn to hear the data speak.

Self-audit framework: Three past predictions and where they missed

I hold to one rule: every analysis must include an audit section. Without it, a writer gradually remembers only the times he was right.

Three recent judgements of mine, and the actual outcomes.

First, I argued the shift from pimpled rubber to sponge rubber in youth tiers would happen fast, within about two seasons. It was slower. The cause I identified after auditing: competitive pressure at youth events makes coaches prioritise short-term results, and pimpled rubber delivers short-term results faster. I underweighted institutional pressure and overweighted purely technical factors. This is an error at the analysis level, not the observation level.

Second, I argued that short-rally win rates among domestic front-runners would rise after more international exposure. It did not, clearly. On re-charting, I found the cause: the number of international matches per player per year is too small to change behaviour. A player competing in three international events a year, totalling fewer than ten matches, cannot reshape reflexes formed over a decade.

Third, I argued that taller players with long reach would dominate medium-length exchanges. In practice, that advantage was neutralised by a factor I did not anticipate: taller players tend to stand further from the table, and that distance degrades their ball speed more than it does for shorter players standing close in.

Three predictions, three misses. But each miss added a variable to the model, which means the model has widened rather than narrowed.

Why this matters to the whole sport, not just to one analyst

A fair question: what does one man charting in a notebook have to do with the national game?

The answer is that data does not only serve analysis. It serves four different groups, each missing something different.

Coaches lack data for in-match decisions. Without it, substitution choices, timeout calls and serve-tactic switches rest on impression. Impression is not wrong, but it cannot be compared across situations.

Players lack data to know where they are improving. A player can markedly improve the backhand over six months without knowing it, because there is no way to compare against his own past self.

Parents lack data for investment decisions. This is the group that loses most and has the least voice, because they must decide based on someone else's account.

Administrators and sponsors lack data for allocating resources. How is an investment in youth tiers to be evaluated if there is no index to compare before and after?

Four groups, four gaps, all from one cause: information about matches is not retained.

A minimal, feasible proposal

I am not proposing a large analytics platform. Such proposals usually die at the budget stage.

I am proposing something much smaller, and in my view sufficient to change the picture within three years.

At each national round, all that is needed is one person, one notebook, and a form with six columns. Those six columns are: serve placement, third-ball selection, third-ball outcome, receive selection, touches per rally, and who changed direction first.

Six columns. No more.

One person recording these six columns for three matches a day across a seven-day final round generates roughly one hundred and twenty matches of data. Cross-check against the referee's sheet to validate the recording layer. After three seasons we would have a historical dataset on Vietnamese players deeper than any in Southeast Asia.

The cost is close to zero. The barrier is not money. The barrier is accepting that record-keeping is part of professional work, not an eccentric side task.

What would change if this were done

With such a dataset, three things become immediately feasible.

First, defining each player's tactical identity numerically rather than descriptively. Instead of saying a player is an attacker, we could say he chooses to attack on the third ball at a certain frequency and wins a certain percentage of those.

Second, early detection of stagnation. When a young player's improvement rate slows across three consecutive rounds in one specific category, that is an intervention signal — and it appears before the scoreline reflects it.

Third, and perhaps most importantly, building a baseline. A baseline means the average of the entire field. With a baseline, every result becomes a deviation from the mean, and deviation is the only thing comparable across eras.

Without a baseline, every comparison is a comparison between two memories.

Second contrarian angle: Data cannot replace the eye

I must state this plainly before closing, because it is the reverse side of the entire argument.

In many sports, once data became widespread, an unintended consequence appeared: people began to believe indices could replace observation. Coaches looked at dashboards instead of athletes. Commentators read charts instead of watching matches. The result was judgements that were numerically precise and meaningless in sporting terms.

Data tells you what happened. It does not tell you what could have happened. It does not tell you at which moment a player hesitated, where he looked before serving, or at which beat he changed his mind.

Those things live in the eye of the person sitting close to the table. Which is why I still carry the ruled notebook alongside everything else.

I look at the diagram first, and at the reputation second.

Points to track in the coming round

I set three tracking points for the period ahead, and I will audit them when the season closes.

First, the third-ball point-win rate of the young player group. If it does not rise, the service quality of the youth pipeline is standing still — a coaching problem, not a talent problem.

Second, the rally-length distribution in semi-finals and finals. If the distribution shifts toward short rallies, the level is rising in an early-attack direction. If it shifts toward long rallies, players are choosing a safer style, and that carries very different implications for international strategy.

Third, the ratio of side steps to crossover steps among the leading group. If side steps rise, pre-arrival judgement is improving. If not, fitness is still compensating for judgement.

Three points, three indices, and I will record them by hand, as I always have.

An open thought

Someone will read this and say: if there is no data yet, then the patterns above are just personal observation — why trust them?

Correct. And that is exactly the point where I want to stop.

This entire article rests on a paradox: I am using a thin dataset to argue for the necessity of a thick one. I have no way out of that paradox, and I do not try. The only thing I can do is state the confidence level of each judgement, marking what is observation, what is inference, and what is conjecture.

The data gap in Vietnamese table tennis will not be filled by one article. It will be filled on the first day somebody in a provincial arena accepts that recording a rally is part of the match, not a redundant chore.

Until that day, I will still be in the fourth row, with a ruled notebook and a pencil, recording what those seven columns will never hold.