On 12 December 2026 I reopened my billiards record file. It has fourteen columns, built up since 2026, one row per match: player, opponent, format, innings, points, longest series, average, table conditions, tournament, date. In 2026, eleven of those fourteen columns were empty across most rows. Not because I stopped recording. Because there were no matches to record.

The billiards hall on Lach Tray Street in Hai Phong, where I have sat every Saturday afternoon for years, had a closed sign up from July. The owner, Mr. Hung, draped cloth over four tables and asked me: "What is there left to record?" I had no answer. An analyst without data is like a cueist without a table.

Context: a year removed from the observation chain

Vietnam's fourth COVID-19 wave began in late April 2026, and it treated billiards no differently from football or basketball. Domestic tournaments were postponed or cancelled in bulk; grassroots competition clusters — where most young cueists accumulate matches — shut down almost simultaneously. SEA Games 31, originally scheduled to open on 21 November 2026 in Hanoi, was pushed to May 2026. For a sport whose international calendar was already thin, losing another year means losing nearly an entire observation cycle.

The data structure of carom three-cushion needs spelling out, because it differs from snooker and from pool. A three-cushion match is not measured in frames but in innings — the number of turns each player is permitted. The primary metric is the average: total points divided by total innings. A world-class cueist typically sits between 1.8 and 2.1 points per turn. The second metric is the series, the longest consecutive run within a single turn, which often decides a short match. The third is win rate, which only means something when paired with opponent quality.

Those three layers stack on top of each other. Losing one layer can be patched. Losing all three leaves nothing to analyse.

When home court stops being a fortress, I learned to listen to an empty stand. In 2026 the empty stand was not a metaphor. It was the normal operating state of an entire industry.

Core: three layers collapse together

In 2026 all three layers vanished for most of the year.

I ran a simple count to test the severity. Across the four seasons from 2026 to 2026, my file recorded more than three hundred matches with complete innings and points, spread across multiple tournaments and multiple opponent tiers. In 2026 that number was eleven matches, most of them internal friendlies arranged during brief windows when distancing rules eased, with no official referee and no opponents from outside the province. Statistically, eleven matches do not form a smaller sample of the old series; they form a different kind of sample. Merging them into the 2026-2026 series produces a hybrid metric that describes nothing real.

The average is a fraction, and 2026 removed the denominator. As innings approach zero, the metric stops being low or high — it becomes undefined. This is where many year-end summaries go wrong. I read at least four pieces asserting that a cueist "held his form" on the basis of two or three matches. None published innings. None named the opponent, the venue, or whether the table was fast or slow. A conclusion built on a fraction with an empty denominator cannot be verified, and what cannot be verified should not be published.

I set myself three conditions before using any metric. 2026 was the year all three failed at once.

First, who measured? In billiards, the measurer is usually the tournament organiser, and each organiser records innings under a slightly different convention — some count foul strokes, some do not. Two scorecards of the same match can differ by 0.1 points per turn purely because of convention.

Second, under what conditions? Table bed, humidity, room temperature and cloth speed all shift the average in ways people rarely anticipate. A cueist on a fast table in a cold room can run 0.15 points per turn higher than the same man on a slow table in a hot room. In 2026 most remaining matches were played on hall tables, not calibrated competition tables. The conditions changed, so the ruler changed.

Third, who was the opponent? Against a strong defensive cueist, innings rise and the average falls mechanically even when form has not moved an inch. This is the condition that breaks the most models, and I learned it through an old shock.

In 2026, while still applying football data to the V.League, I predicted Hai Phong FC would win 3-1 against Sanna Khanh Hoa on the back of an xG gap of 2.8 to 1.0. The match ended 0-1, and goalkeeper Tran Buu Ngoc made seven saves. Data never lies, but I have misheard it before. The lesson I took was that my model carried no variable for opponent quality. In billiards, the equivalent lesson is this: a cueist who averages 2.0 against weak opponents and 1.4 against strong defenders is still the same cueist. Record only one of those numbers and you are describing the opponent, not the man.

In 2026 most remaining Vietnamese billiards matches were internal friendlies. The sample was not merely small — it was systematically skewed toward familiar opponents. A cueist who plays the same three people in the same hall will produce an average so stable it becomes suspicious. That stability is a product of the environment, not of technique. And a metric that is stable for the wrong reason is more dangerous than one that fluctuates for the right reason, because it manufactures a false sense of safety.

Contrarian: an unfalsifiable story, and an untracked base layer

One narrative circulated widely in 2026: the lockdown is an opportunity; cueists at home have time to train properly, so standards will rise. I do not dispute that practice helps. I dispute how the story gets used — as a finished conclusion rather than an untested hypothesis.

The problem is that it cannot be wrong. If a cueist returns and plays well, people say: see, the training worked. If he plays badly, they say: he lacked match sharpness. Both branches lead to the same conclusion, so it carries no information. A hypothesis that cannot be falsified is not a hypothesis; it is a belief rewritten as analysis.

The second contrarian point matters more, and it points the opposite way from where the crowd is looking. When the whole of Vietnamese billiards reduces to a handful of names — Tran Quyet Chien, Nguyen Duc Anh Chien, Ma Minh Cam, Duong Anh Vu — there is an invisible pressure to write about them, because they are the only living data. But the layer damaged in 2026 was not the top. The top has contracts, sponsors, personal training schedules, people looking after it. The damaged layer was the base: grassroots halls closed, youth tournaments stopped, provincial competition clusters never took place. That is the talent pipeline. It generates no news, so it goes unmeasured.

The biggest blind spot in every 2026 retrospective sits here: we measure what is easy to measure and ignore what determines the next ten years. A billiards nation can lose one season at the top and recover within six months. Losing a selection generation takes nearly a decade to refill, and throughout that decade nobody sees the outline of the gap, because a gap has no results to record.

I have to warn myself here too. The temptation of a data person is to turn missing data into a finding. "No bad news" becomes "no risk" after one very small step. In an empty file, the absence of recorded violations does not mean there were no violations. Absence of evidence is not evidence of absence — and in a year when tournaments were cancelled, that absence is almost certainly a consequence of nobody observing, not of nothing existing to observe.

What to watch in the next cycle

The signal I am waiting for is not a title. It is a series.

SEA Games 31 now sits in May 2026. When it takes place it will supply what the whole of 2026 did not: a group of cueists playing each other under identical conditions, with referees, with recorded innings, with opponents from outside the province. That is the starting point of a new sample, not a conclusion. One tournament is still one tournament, and I have learned enough not to rebuild an entire model from a single week of play.

Three things go into the file before anyone is allowed to conclude: innings per match, opponent name and tier, and the number of consecutive matches within the same month. Miss any one column and that row is marked unusable — even if it carries a handsome scoreline.

Three thousand matches taught me that one match can teach more than all of them. But three thousand matches also taught me that a match without a denominator teaches nothing. I do not write to persuade anyone. I write so that the data has a witness. In 2026, Vietnamese billiards did not lack stories. It lacked denominators.

Method note: Average = total points divided by total innings within a match or tournament. Series = the maximum consecutive points within a single turn. Win rate is used only when the sample contains at least ten matches against opponents from at least three different tiers.