Vietnamese Table Tennis and the Empty Data Table: The Discipline of Saying 'Not Enough Evidence'

2:14 a.m., Beijing time. I reopen my spreadsheet after the analysis pipeline returns its result. Nine columns. One column has text. The other eight are blank, and inside each blank cell is a line repeated over and over: not enough information. The only filled column is the domain label — table tennis.

I once sat in a gymnasium in Hanoi on a spectator-free afternoon, listening to the ball bounce off the table like a wooden drum. The electronic scoreboard above showed two things only: set score and point score. No spin. No placement. No rally length. A national-level table tennis match unfolded in silence and vanished from the record after forty minutes, leaving behind a single string: 3-2, 11-9, 9-11.

That night, when the system returned eight blank cells, I realised I was standing in exactly that scene — just one layer higher, in the analysis stack.

I work as a data consultant for teams. In football I have xG, PPDA, pass maps, transfer valuations expressed as probabilities. In table tennis the toolkit is far thinner, and in Vietnam it is thin enough that some days I work with a single column.

What is publicly available includes the world ranking published weekly by the International Table Tennis Federation (ITTF), results from the WTT circuit launched in 2026, SEA Games result sheets, and the national champion lists issued by the Vietnam Table Tennis Federation. Table tennis joined the Olympic programme at Seoul 2026, according to International Olympic Committee records, so continental head-to-head history is in fact more than three decades deep.

What is not public is the real problem. There is no measured spin. No placement map. No rally-length distribution. No win rate across the first three shots — serve, receive, and the third ball. For a sport in which the level gap is usually decided inside the first seven contacts of a point, missing precisely those three contacts is like analysing football without shot data.

The pipeline I ran that night received a source article, executed a deconstruction step, and returned an empty result. The domain label was still written correctly: table tennis. Every other field — title, source, article type, viewpoints, list of facts, entities involved, time sensitivity, source quality — was blank.

That gives the output a property I have to state plainly: it is a process finding, not a sporting finding. Nine analytical dimensions — technique and tactics, player data and head-to-head, event system and points, competitive landscape, rules and governance, coaching and talent pipeline, risk surface, public narrative, industry transmission — all returned "insufficient information". Vietnamese table tennis is not short of stories. That night it was short of raw material.

A blank cell and a zero are two different species, and the bad analyst treats them as the same animal. In a table tennis dataset, a blank in the spin column means nobody measured. A zero in that column means somebody measured and recorded no spin. They look identical on a quick scan, but one is a knowledge gap and the other is a conclusion. Every serious mistake in Vietnamese table tennis analysis starts here: taking a knowledge gap and assigning it a value.

When I have to work with thin sources, I build an expectation index for table tennis in the same spirit as xG. I call it xP — expected points. The calculation rests on scoreline context rather than scoreline outcome. A short sidespin serve to the middle of the table at 9-9 carries a far higher expected value than the same serve at 3-1, because the opponent is forced into a safer return that opens the third ball. When I sum a player's xP across a match, I learn nothing about who won. I learn something about who generated more chances to win.

Based on my experience tracking matches at domestic events and through open federation data, three pillars decide most national-level table tennis results: short-serve quality, wrist-driven receive quality, and the ability to convert from defence to a decisive shot on the fifth contact. None of those three appears in any public domestic result sheet. Result sheets say who won. They do not say who came closer to winning.

That is why I treat xP as a mirror of confession. An expectation index does not judge the loop drive. It only illuminates the table tennis you refuse to look at. A player can win 3-1 against a weaker opponent and still play a low-value game: long serves with no variation, pushed returns, a lucky third ball. Another player can lose 2-3 to a stronger opponent having generated a higher xP chain across four sets. The media records only the first case.

With Vietnamese players, I refuse to write sentences like "Nguyen Anh Tu is finding form" or "Dinh Quang Linh has stalled" without three independent metrics behind the claim. Nguyen Khoa Dieu Khanh, Mai Hoang My Trang and Tran Mai Ngoc all deserve to be analysed with a dataset, not with adjectives. My rule is simple: a claim about a player may only be published when at least three metrics from two independent sources support it, and one of those two sources must be qualitative — a live scorebook, a coach's note, or an on-site observer.

When I applied that rule to my own empty table, the result was eight blank cells. The deconstruction step failed at content extraction while domain classification still ran correctly. That is a valuable diagnostic detail: the fault sits in reading the words of the article, not in recognising the sport. Had all nine columns been blank, I would suspect the whole pipeline. When exactly one column survives, I know precisely where to open the logs.

At the same time I added a warning to my journal: the silent-fabrication risk. A nine-dimension framework exerts enormous structural pressure — it constantly pushes the writer to fill a gap with a plausible name, a plausible tournament, a plausible number. A wholly blank table is safe. A ninety-percent blank table with a ready skeleton is the most fertile ground for fabricated data, because every cell already has a heading waiting for content.

I drew up my own observation list. First, re-check every field after re-running deconstruction against this blank baseline. Second, confirm the source article is archived intact so it can be rebuilt. Third, inspect pipeline logs to see whether the article body actually reached the deconstruction step or was cut at ingestion. Fourth, track whether the "domain label only" failure recurs across other articles, because recurrence would make it a systemic fault rather than a single incident.

An empty gym does not breed ghosts. It produces the cleanest data a practitioner could dream of. But an empty gym does not generate data by itself either. Someone has to walk in, time the rallies, count the contacts, record spin by feel and cross-check against slow-motion footage. Without that work, the empty gym is just a quiet room, and every conclusion drawn from it is drawn from imagination.

The counter-intuitive angle sits here: an empty dataset is a more honest artefact than a full dataset assembled from guesswork. In sport, people fear blank space and love a full table. But a full table comes in two kinds. The first is filled with documented observation. The second is filled with stories that sound reasonable. The second is far more dangerous, because it does not present itself as fake. It presents itself as analysis.

The biggest risk to Vietnamese table tennis in the media is not a shortage of numbers. The risk is confident numbers with no source. A percentage with no provenance, passed hand to hand through three articles, becomes a fact within a season, and by then nobody remembers where it started. I have seen it in football, and I do not want to see it repeat in table tennis, where the public data foundation is many times thinner.

Another trap is correlation read as causation. A player winning four of the last five against a specific opponent does not make that opponent a nemesis. A five-match sample at national level, with changing opponents, changing tables and changing balls, carries near-zero reliability. Whenever I find a beautiful head-to-head streak in the data, I ask two more questions: how many of those matches fell in one player's peak phase and the other's injury phase, and how many ended in a narrow margin. If most were narrow, I do not call it a nemesis. I call it an hourglass that has not run out.

I sit in front of the screen to attack, but what I defend is the arrogance of numbers. And the best defence is to say plainly, "I do not have enough evidence" before saying anything else.

I do not write about table tennis. I write about the dents players leave on the chart. In Vietnamese table tennis, most of those dents have never been recorded. The signal to watch in the next cycle is not in the ranking. It sits in whether somebody starts logging serve contact, placement and rally length, or whether everything still stops at the score and the adjectives. A sport truly enters the data era only when people agree to record the points nobody won.