Data Discipline: When the Table Tennis Analysis Sheet Returns to Zero

There was an evening when I sat in front of a screen with a spreadsheet open and every cell empty. The data feed I use to build table tennis models returned nothing: no event name, no player, no score, no timestamp. The dashboard displayed exactly one line that any analyst fears more than any bad number — "insufficient information."

In nearly five decades in this trade, I have learned that a bad number is still better than a gap. A bad number tells me the direction is wrong. A gap only tells me I have nothing yet. The most dangerous thing is not the gap itself, but the reflex to fill it with guesswork.

Table tennis analysis in Vietnam is abundant in quantity but thin in quality. At every WTT event, every ITTF World Championships, social media floods with takes. Most of them begin with a feeling, then go looking for data to justify that feeling. The correct process runs the other way: data first, conclusion after.

I came to table tennis from football, where I once calculated xG by hand in Excel and discovered that possession does not decide victory. That lesson followed me into the celluloid ball: a player who holds the rhythm longer, serves with more spin, moves faster — all of it matters only when tied to the score. Table tennis is a sport of point sequences, and every sequence leaves a trace that can be counted.

What I want to say is very concrete. When a source returns empty data, the analyst has two choices: stop and say plainly "not enough information," or fill the gap with plausible-sounding inference. The second choice is always more tempting, because it yields a complete article — with an opening, a table, a conclusion. But it turns analysis into fiction.

A model without source data is not a model — it is a belief presented in the form of a table.

I split every claim into three layers. The first is raw fact: the score of each game, points won on serve, the average duration of a rally. The second is derived metrics: the share of points won on serve, the share won on receive, the point differential in the last five points of a game. The third is the conclusion. A conclusion stands only when it connects directly down to both layers beneath it. When the first layer is empty, the second cannot exist, and the third is merely an opinion dressed in terminology.

In table tennis, I pay particular attention to three metrics. The first is the share of points won in the decisive service sequence — where pressure turns technique into psychology. The second is the stability of the backhand under pressure, measured by the rate of unforced errors in the last five points of a game. The third is the ability to change direction from the third rally onward, once the opponent has read the rhythm. These three metrics do not appear on the scoreboard. They must be counted, and they can only be counted with match footage and precise timestamps.

Based on my experience watching matches, most mistakes do not come from misreading data, but from reading data that does not exist. I have seen analyses predict a table tennis match without even a list of participating players. They still read smoothly. They still had tables. They still concluded. Every fact in them was born from the wish to have a fact.

A goal is only the verdict. xG is the testimony.

I wrote that line for football, but it holds for any sport with a score. In table tennis, the game score is only the verdict. The point sequence, the service rhythm, and how a player reacts at 9-9 are the testimony. Testimony needs an honest recorder, not a decorator.

In table tennis analysis, I split risk into two kinds: measurable risk and unmeasurable risk. Measurable risk includes serve form, unforced-error rate, and stamina in the fifth game. Unmeasurable risk includes the psychological pressure of facing an unfamiliar opponent and the weight of public expectation. When the data source is empty, neither kind can be assessed, and any risk ranking is mere decoration.

People often think a good analyst is one who always has an answer. I believe the opposite. A good analyst is one who knows exactly when they cannot yet answer, and says so. When the data sheet is empty, the conclusion "cannot be assessed" is treated as failure. But set beside a fabricated conclusion that sounds reasonable, the fabricated one is the real failure — it just fails quietly.

I fear a wrong model more than a wrong judgment, because it is wrong systematically.

A wrong judgment affects only one match. A wrong model spreads through the entire chain of decisions behind it, and every application repeats that error under the appearance of precision. This is the biggest blind spot in sports analysis generally and table tennis specifically. We reward confidence, not accuracy. An article that dares to say "I don't know" is rarely shared; an article that dares to assert is shared at once.

The public story about a player usually runs ahead of the data about that player. A win at a youth event can create greater expectation than a whole stable season. Expectation is not wrong, but it must be balanced by baseline data: actual matches, actual opponents, and a sufficiently long time sample. A sample of three matches is not enough to conclude anything about a player, however impressive those three matches were.

Vietnamese table tennis has a major opportunity to build a proper data foundation. Domestic events, selection rounds, and international friendlies can all become source data if recorded correctly. A foundation holds only when every fact has a source, every source has a date, and every date is verifiable. Without those three things, every model stands on sand.

Data does not forgive emotion. And that is why I converted.

An empty data gap is not a full stop. It is a signal for the next round of analysis: find the source, verify the source, timestamp it, and only then build the model. I still keep the habit of recording the date of every fact. Without a date, a fact cannot be verified. Unverifiable, it is only a rumor with numbers.

When your data source is empty, what will you write — or will you wait?