When the Table Tennis Tracking Sheet Goes Blank: A Data Analyst's Discipline Against Fabrication

On August 9, 2026, in the round of 16 at a WTT Contender event held in a packed indoor arena, a match stretched to seven games with eighty-four points scored. I sat in the seventh row, notebook open, tablet loaded with a spreadsheet I had built six years earlier. When the referee called the end, I looked down and saw an entirely empty column: the column recording the win rate on short serves. Not an empty cell. The whole column. Eighty-four points, seven games, two players, and not a single usable line of data.

What made it notable was that I could easily have filled that column with numbers that sounded perfectly reasonable. The winning player almost certainly held a serving advantage. Readers want a percentage. Editors want a figure to put in a headline. And I told myself I had enough experience to estimate. But this time I left the column blank and typed four words into it: insufficient information. This article is about those four words.

Context: a sport that is tracked less than people assume

Table tennis is the fastest-paced of the popular combat sports. An average rally at international level lasts under five seconds, and a game can end in four minutes. That speed makes recording expensive. Football offers thirty easily captured variables from a televised match, whereas in table tennis every point must be coded by hand if an analyst wants to know whether it was won by a serve, a third-ball forehand loop, or an opponent's error.

The WTT points system runs on a rolling 52-week mechanism. A tournament's points expire after exactly one year, and players must constantly defend old points while hunting new ones. That mechanism creates a paradox for data people: the published ranking is highly transparent, but the technical data beneath it is almost never released. We know how many points a player holds, yet we rarely know the technical structure that produced them.

In Vietnam, this gap is wider still. National championships, youth events, and internal ranking matches usually preserve only match results, sometimes game scores. Information about serving tactics, court positioning, or a player's decision to change spin direction in the fifth or sixth game mostly lives in coaches' memories, or on handwritten pages that get lost.

I entered the profession through exactly such pages. My first V.League dataset contained hundreds of errors, yet it taught me more cleanliness than any course could. I mistyped dates, names, even columns. One day I summed a whole season's goals incorrectly and published an article based on the wrong figure. A reader in Hai Phong messaged to point out the error. I had to correct it, apologize, and from then on set myself a rule: if a data cell has no verifiable source, it stays empty rather than being inferred.

That rule holds even more firmly in table tennis. Every point is the product of three consecutive decisions: choosing the serve type, choosing the placement, and choosing how to handle the return. Skip one and the analyst misreads the entire chain. A serve win-rate without context on the opponent, surface, rubber, and physical condition is just a pretty number on paper, not a competitive truth.

The evidence chain: where data weakens

When a table tennis tracking sheet has a blank column, the cause sits in one of three layers. The first is collection: the recorder could not click fast enough as the rally unfolded. The second is coding: the recorder clicked, but definitions differed across games, producing inconsistent data. The third is storage: the data was recorded but cut off midway because of a dead battery, a dropped connection, or simply a forgotten sync.

Those three layers map onto three types of missing data in statistics. Missing completely at random occurs when the loss is unrelated to the match itself. Missing systematically occurs when the loss is tied to a specific condition, for example a recorder who only misses points in the deciding game because of tension. Missing with meaning occurs when the very inability to record reflects something about the match, such as both players moving too fast for anyone to code.

What readers rarely see is that the third type usually holds the most valuable information. My blank column on August 9 was not junk data. It was a signal. Seven games and eighty-four points without a single line recorded on short serves means the match's pace exceeded the threshold for manual capture. And when pace exceeds threshold, tactics usually do too, meaning the two players were operating on a different speed tier from the rest of the field.

To verify, I did something simple: I re-counted the games that contained at least one spin-direction change from the seventh point onward. The result was six out of seven games. Against reference data I had collected from WTT events of the same tier over the previous two years, the average was roughly two out of seven. The gap between six and two needs no elaborate model to notice. It shows this match was governed by serve variation far more than usual.

From there I rebuilt an analytical frame with four layers. Layer one is the win rate on serve, separating short and long serves. Layer two is the win rate on the third ball, meaning the stroke immediately after the opponent's return. Layer three is the rally-length distribution, split into under four exchanges, four to seven, and above seven. Layer four is the win rate at decisive points, meaning when the margin is down to one point.

Applying these four layers to the August 9 match revealed a much fuller picture than a single blank column. The winning player had an unusually high win rate on long serves but a lower win rate on short serves than his opponent. That sounds contradictory until you look at the rally-length layer: this player deliberately pushed the match into the above-seven category, where he held a fitness advantage in the final two games. His short serve was ineffective, but it was bait to drag the opponent into long rallies.

This structure explains why a single metric always misleads in table tennis. People often cite the serve win-rate as a measure of strength. But that rate depends on whether a player wants to win quickly or to extend. Two different goals produce two different rates, and both can lead to victory. Data does not need me to believe in it. Data needs me to check it. And checking begins with the question: what is this player trying to do with each serve?

I read a team through thirty variables before listening to a commentator. That habit formed over years of realizing that most media judgments are written before a match ends, based on feeling rather than a causal chain. In table tennis the gap between feeling and data is even larger, because the speed of the sport makes spin direction hard for the human eye to follow, while a properly kept record cannot lie.

The contrarian angle: correlation is not causation, and blankness is not failure

The 2026 World Cup taught me one thing: the model did not collapse, I was the one who believed it absolutely. I had run a regression across five hundred international matches and felt certain a strong team would go deep. Reality was far harsher. That lesson applies to table tennis in an unexpected way: an analyst's greatest temptation is not making a wrong prediction, but filling blank cells with plausible-sounding guesses.

There is a very real professional pressure. Readers want numbers. Editors want numbers. Sponsors want numbers. When a blank column appears, the natural reflex is to interpolate, to estimate, to assign a tournament average to this match. That act sounds harmless, but it creates a chain of consequences. An estimate from match A becomes reference data for match B. By match C, no one remembers whether the original value was a guess or an observation. After a few seasons, an entire data system can be built on sand.

In table tennis the subtlest trap is the correlation between serve win-rate and match outcome. The two nearly always move together, leading people to conclude that good serving causes winning. But the causal order can be reversed: a player who is winning serves with more confidence, more boldness, and therefore wins more service points. A high rate is a symptom of leading, not necessarily the cause of leading.

I once wrote a wrong article by mistaking this causal direction. I praised a young player for an outstanding serve win-rate, then months later reviewed the footage and found most of those points came against opponents a tier weaker. Against peers, the rate fell to average. I had to rewrite the old article with a correction. Since then, every rate I publish carries an opponent classification column, even if only three levels: stronger, peer, weaker.

Blankness in a dataset is not an analyst's failure. It is a reminder that this sport is broader than our recording capacity. When the Bundesliga played in empty stadiums, I realized home advantage is just a variable waiting to be erased. When a table tennis sheet has a blank column, I realize some dimensions of a match are also just variables waiting to be erased from the analysis, at least for this cycle. Preserving that blankness is an honest choice, not a lazy one.

There is another counterargument worth weighing. If an analyst always says there is insufficient information, will readers still read? My experience says yes, and more carefully. When I publicly disclosed a blank data column instead of filling it, many readers in Hai Phong and Ho Chi Minh City wrote back that they trusted the remaining columns more. Honesty about data limits increases the value of the data that survives verification. That is a trade-off against media instinct, but true to the nature of data work.

Signals for the next cycle

I will be brief about what I will track next cycle. First is the collection quality of serve data at domestic events. If we can capture serve type and placement for at least two hundred points per event, we can start comparing the tactical structures of Vietnamese players against regional benchmarks, instead of comparing only results.

Second is rally-length distribution by game. This metric reveals fitness and tactical intent more clearly than any aggregate figure. A player whose distribution skews toward the above-seven group in the final two games is usually the one controlling the match's rhythm, regardless of how the individual game scores look.

Third is the win rate at decisive points, which I separate entirely from the overall win rate. In table tennis, the ninth and tenth points of a game are a different sport psychologically from the first six. Whoever scores most in that zone decides the fate of major tournaments.

And fourth, perhaps most important, is the discipline of keeping columns blank. Any table tennis data system that wants to last must enforce a hard rule: a cell without a verifiable source must be marked as insufficient information rather than filled with an average. A dataset with a few blank cells is worth more than a packed one where nobody knows which cells are real.

The eighty-four points from the August 9 match remain, and my short-serve column remains blank. I do not consider it a failure. I consider it the starting point of a more rigorous data-collection cycle. Vietnamese table tennis does not lack good players. What it lacks is records clean enough to show why they are good. And to build such clean records, the first step is accepting that some cells we do not yet know, rather than pretending we have known all along.