The Empty Spreadsheet: Notes from a Domestic Three-Cushion Tournament

At 2:47 in the morning on 14 March 2026, in a billiard hall on Lach Tray Street in Hai Phong, table three had been dark for six hours. On my laptop was a spreadsheet with fourteen rows and seven columns. Only one column held any data: the scoreline. The other six were empty. No innings. No moyenne. No high run. No shot-clock times. No note on cloth, temperature or ball set. I had watched most of that national three-cushion event from seven metres away. I remembered the decisive rail in the third game of the semi-final. I had nothing on paper to prove that my memory was right.

The temptation arrived quickly and in a very reasonable voice: you remember it, so just fill it in. I sat with that voice for forty minutes. Filling it in would have given me a complete analysis, a clean table for readers, and almost no way for anyone to check me. Not filling it in gave me an article about missing data, which sounds like a long apology.

I chose the second. But the choice was not generic caution. It was the product of nine years of note-taking, and of at least three occasions when I misheard my own figures.

Data never lies, but I have misheard it before. I first wrote that line in 2026, and I still repeat it whenever my hands reach the keyboard near three in the morning.

A sport with stars but no ledger

Vietnamese three-cushion billiards moved, within half a decade, from the edge of the world carom map to a country with a world champion and a player who has held the top of the UMB ranking. Tran Quyet Chien, Bao Phuong Vinh, Duong Quoc Hoang, Nguyen Quoc Nguyen, Chiem Hong Thai, Ma Minh Cam — that list is long enough for a national sport to consider itself a regional centre. It is also long enough for fans to ask why domestic events are still covered with photographs and feelings rather than with records.

At international level, every three-cushion match arrives with almost complete data: innings per player, points, moyenne generale, high run, sometimes a point-by-inning chart. The systems that serve the world federation and World Cup events have done this for years, and Vietnamese fans who followed World Cups staged in Ho Chi Minh City are used to seeing statistics running alongside the table.

Domestically the gap is far wider than most people assume. Tournament organisers record scores to settle results and schedule matches. The marker sits beside the table with a notebook and a pen. The job is the final number, not the quality of the number. The result is a professional paradox: more matches are streamed than ever before, more clips are cut than ever before, and the number of innings in a national semi-final may exist nowhere.

Based on my own experience of watching matches, I believe this is the single biggest bottleneck in Vietnamese billiards today. Not a shortage of players. Not a shortage of events. Not a shortage of spectators. A shortage of recording infrastructure.

Before using any billiards metric, I run a seven-point checklist: who recorded it, with what tool, on what unit, over how many innings, under what table conditions, with what cross-checking, and whether the data was edited after publication. Those seven points are not bureaucracy. They are the boundary between analysis and storytelling.

What a valid three-cushion dataset looks like

A usable match dataset needs innings per player, final score, moyenne generale, the moyenne over the last five innings, the high run, missed innings separated from forced-safe innings, conversion rate on scoring opportunities where possible, average shot time, and a qualitative notes field for everything a device cannot measure.

The most important figure in three-cushion analysis is not the average; it is the number of innings attached to that average. Two players who finish on 1.20 are telling two entirely different stories if one reached that number over forty innings and the other over twenty-five. In a match to 40 points, the difference between twenty-five and forty innings is the difference between an unusable confidence interval and a usable one. Much of the online argument about who is better after a short event is taking place inside the noise.

Three kinds of missing, and only one is the analyst's fault

Missing data means the data does not exist. Nobody recorded it. That was my case. It is a failure of the recording system, not of the analyst, and the only correct response is to state publicly that the data does not exist.

Dirty data means the data exists but is wrong. This is the most dangerous kind because it does not announce itself. A notebook that miscounts one inning produces a wrong moyenne, and that wrong moyenne travels into every downstream calculation as a fact. I once received an internal statistics sheet from an amateur event in which total innings exceeded reality by roughly ten per cent, because the marker had counted practice shots taken before the referee signalled the start. Nobody intended it. The consequence was identical.

Context-orphaned data means the data is correct but stripped of the conditions that produced it. A moyenne of 1.35 on fast cloth, in a cold room, with a fresh ball set does not mean the same as 1.35 on slow cloth, in humid air, with balls months old.

These three failures require three different responses: re-collection, cleaning, and context restoration. Lumping them under a single word is the fastest route to a confidently wrong conclusion.

What a goalkeeper taught me

In 2026 I applied expected goals to Vietnamese football for the first time. I took data for Hai Phong against Sanna Khanh Hoa in V.League round 18. Hai Phong created 2.8 xG; the opponent 1.0. I predicted a 3-1 win. The match ended 0-1, and goalkeeper Tran Buu Ngoc made seven saves, dismantling my model. The lesson was not that the model was wrong. It was that I had used a single metric to describe a match containing a variable the metric could not measure.

Three-cushion has an almost identical blind spot. Moyenne does not measure how hard the scoring opportunity was. It does not measure what the opponent left. It does not measure the table. A player at 1.40 given easy layouts on fast cloth is playing a different sport from a player at 1.40 who had to break open locked positions every visit. Since then I never conclude from a single index, and every article begins with a list of conditions to verify.

Pressure metrics, and another football lesson

In 2026, I wrote about Mexico beating Germany 2-1 at the World Cup. Germany held 66 per cent possession and made 613 passes, but Mexico's PPDA of 8.4 meant Germany were allowed only 8.4 passes before being interrupted. I concluded Germany would exit early. The piece was mocked. Two weeks later Germany lost 0-2 to South Korea and went out.

What matters is the type of metric I used: one that measures pressure, what an opponent is forced to do, rather than what a team does. In three-cushion, the equivalent metric exists but has not been standardised in Vietnam. I call it the table-pressure index: the number of visits in which a player forces the opponent into a safe option instead of an attacking one. It explains matches won with unimpressive averages, and matches lost despite early scoring. It requires markers who can tell a miss from a forced attempt. Without training, the metric does not exist.

Empty stands and home advantage in billiards

In 2026, the Bundesliga returned with 81 matches behind closed doors across the final nine rounds of 2026/20. Home win rate fell from 44.7 per cent to 33.3 per cent, and away expected goals rose from 1.15 to 1.32. I proposed cutting the home coefficient in my betting model to 0.18 goals per match, ran a chi-square test with p = 0.045, published the result with caveats, and won 62 per cent of Asian handicap bets in that window.

Vietnamese three-cushion events staged at home raise a similar unmeasured variable. When home is no longer a fortress, I learn to listen to an empty stand. I have no billiards data to answer it. I have data from another sport showing that silence measurably changes results — and billiards is a sport of breathing rhythm.

Anatomy of a data pipeline

A billiards data pipeline has four stages: recording at the table, entry into a system, cross-checking, and publication. The first stage determines everything downstream. Digitising a blank sheet still produces a blank sheet.

There is a difference between a low-information match and a data-lost match. They look identical on a screen and are opposites in nature. The first is a fact about the match. The second is a system failure. Labelling the second as the first writes a false verdict into the event's history. I call this silent null propagation.

The null-carrying principle, and three levels of error

When information is missing, state that it is missing rather than guessing. The first level of filling a gap is outright fabrication. The second is inference: taking a real number from one context and applying it to another. The third is silence: saying nothing about missing data so readers assume verification occurred. The third creates false confidence, which in a sport building an analytical culture is more dangerous than a wrong number.

What an empty spreadsheet says about integrity

Data is a preventive fence. Snooker's match-fixing cases were detected partly through pattern analysis of unusual scoring and visit rhythms. A tournament that does not record in detail strips itself of the ability to defend itself. In Vietnam, three-cushion events now attract international players, prize money, media and betting. At that stage, recording infrastructure stops being an analyst's technical detail and becomes the infrastructure of trust.

Contrarian angle: an empty table beats a dirty one

Forced to choose between an empty spreadsheet and a wrong one, I take the empty one every time. Empty forces honesty; dirty gives me a reason to be confidently wrong. Most of my mistakes came not from missing figures but from figures that looked persuasive. The most common three-cushion version of correlation-as-causation is the cue change: a player switches cues, then wins more, and the cue is credited while an easier schedule, weaker opponents and friendlier table conditions go unmentioned. Volume of data cannot rescue poor recording. A large dataset built on a bad first stage is systematically wrong, which is worse than randomly wrong because it manufactures patterns.

The crowd laughed. The numbers did not. A year later, I rewrote that piece.

The human part is outside the spreadsheet

My quantitative background makes me over-measure what can be measured. Billiards is a sport where much of the outcome is untouched by data: the shaking hand on the third rail of a decisive visit, the loss of composure after an opponent's lucky shot, the twenty-minute wait for a first visit. That is why every sheet of mine carries a qualitative notes field recording stance rhythm, body language, rest intervals and recovery speed. One goalkeeper missing a catch is a mistake. Three goalkeepers missing the same catch is a signal. One marker omitting innings is an error; three markers at three events omitting innings is a national recording problem.

Data still needed

Innings and moyenne records for at least thirty domestic matches; table conditions by venue including cloth, temperature and humidity; shot-time data in front of full and empty stands; and a systematic qualitative archive of pressure situations. This article rests on insufficient data for technical conclusions. It is sufficient only for a methodological one.

What I will do next

I decided not to fill in the spreadsheet. I asked the organisers for the original scorebooks and received a scan with scores for every match and no innings — a result, not a failure, because now I know exactly which data exists. I drafted a two-page recording template with a referee friend, requiring no software, only a marker who understands the difference between a missed visit and a forced one. And I wrote this article so the gap would not disappear quietly.

I have no data to say where Vietnamese billiards goes in three years. I have enough to say the road is shorter if every match leaves behind a line of truth instead of a blank. Three thousand matches taught me that one match can teach more than all of them — but only if someone writes it down. I do not write to persuade anyone. I write so that the data has a witness. On 14 March, my memory was the only witness in the room. Memory is not data. Until our recording systems improve, Vietnamese billiards fans will keep having to trust the memory of those sitting seven metres from the table. The question for the next round is not who will win. It is whether, when the next event ends, we will know how many innings the winner needed to get there.