Vietnamese Billiards Is Missing a Data Layer: When the Table Leaves No Numbers to Cross-Check

Eleven at night at a billiards club on Tran Nguyen Han Street, Hai Phong. Table four, three-cushion carom, race to 40 points. My squared notebook is open in the middle, 33 rows ruled for 33 innings per player.

I record four things: who is at the table, how many points the inning produces, where the three balls sit when the inning ends, and how many points the opponent scores on the very next visit. The match closes 40-32. The two general averages are 1.21 and 0.97 points per inning. The winner stands, shakes hands, and the first thing a spectator says to him is: “Good shooting.” Nobody asks about the average. Nobody asks how he left the table on inning 21.

What decided that match did not live in the shots people remember. The loser had nine innings that produced nothing, and six of them ended with a leave that opened the table for the opponent to score four points or more. On the cloth, memory keeps the beautiful shot. In the notebook, what gets kept is the miss.

The next evening I sat down with an analytical report I had received from a sports data source. It carried nine sections: technique and playing style, player data, tournament system, power map, rules and governance, career ecosystem, risk, media narrative, industry chain. All nine sections carried the same phrase: insufficient information. The report was not sloppy. Public, verifiable data on Vietnamese billiards barely exists.

At the international level the story is different. The world billiards federation issues an official match sheet for every three-cushion carom match, recording three figures: innings, general average and best run. The American Accu-Stats system breaks the overall performance average into components: shot-making, position play, safety, break, kicking. Professional pool events under the Matchroom banner run a shot clock and publish break and run statistics. A top-level match there leaves a trail you can look up years later.

Here, we have players who have touched the summit: Tran Quyet Chien has risen to number one in the world three-cushion ranking, Bao Phuong Vinh took the world title in 2026 on home tables in Binh Thuan, Nguyen Tran Thanh Tu sits permanently in the leading group, Duong Quoc Hoang carries Vietnamese pool into international events. The achievements are abundant. The data layer beneath them is empty.

Eight years ago I paid the price for reading a metric without asking under what conditions it was measured. Round 18 of the 2026 V.League, I pulled Understat numbers for Hai Phong against Sanna Khanh Hoa: Hai Phong created 2.8 xG, the opponent 1.0. I concluded a 3-1 win. The match ended 0-1, and goalkeeper Tran Buu Ngoc made seven saves. Data never lies, but I have misheard it before. What I misheard was not the number. It was the condition that produced it.

Applying that same question to billiards, four gaps become obvious.

The largest gap is that the general average hides the distribution. Two players both averaging 1.2 points per inning can be two unrelated stories. One runs five or six points and then misses; the other collects one point across many visits. In a race to 40, the first wins most head-to-heads, while the second survives on the opponent's errors. Telling them apart requires a distribution of run lengths, not another average. The general average is the only figure a billiards match leaves behind, and it is the most misleading one.

The metric I consider the most important has no name on any score sheet yet: the dangerous-leave rate. The calculation is simple — count the visits that end with the opponent able to score four or more, then divide by the player's total visits. In that match, the winner left the table dangerous nine times out of 33, or 27 percent. The loser left it dangerous 17 times out of 33, or 52 percent. That 25-point gap explains the result far better than the 0.24-point gap in general averages. In billiards, defence is not a shot; it is the position of three balls at the exact moment you leave the table.

Another variable no current score sheet captures: the quality of the opening visit. In three-cushion carom, the break decides whether a player scores immediately or shifts into a safety exchange. A good break keeps the waiting ball in an easy half of the table; a bad break pushes all three balls tight to the cushion, effectively wasting the visit. The scoring rate on opening visits is a direct measure of break quality, and it swings wildly between players who share the same general average. In pool, the equivalent metrics already exist abroad but are unused in Vietnam: break-and-run rate, safety success rate, legal break with a ball potted. Domestic organisers do not publish them, and because they do not, nobody cross-checks.

The next gap is sample size. A race to 40 yields roughly 30 to 35 innings per player. A five-match tournament yields around 165 innings — enough to spot a broad trend, not enough to conclude anything about a player. The shorter the format, the smaller the sample, and the more technical advantage is compressed, giving way to randomness. Put differently, short formats do not produce underserving winners; they produce a smaller sample in which skill barely has time to show. I once watched a player declared finished after a race-to-25 event, then reach the semi-finals of a race-to-50 event three months later with a general average 0.3 points higher. A conclusion drawn from three matches is measurement error, not truth.

Table conditions are the most neglected variable of all. Hai Phong humidity is not Ho Chi Minh City air-conditioned humidity; new cloth is not worn cloth; room temperature changes how the cushions rebound. In billiards, home advantage is a physical variable, not a feeling. If a player switches clubs and his average climbs from 1.0 to 1.3, the popular explanation will be a leap in form. The explanation that gets less attention: faster tables, drier cloth, more consistent bounce. Correlation between changing clubs and rising averages says nothing about cause, and in billiards the two variables are almost inseparable when table conditions are never recorded.

I ran into exactly this type of variable in another sport. In 2026, the Bundesliga returned with 81 matches played without spectators over the final nine rounds of the 2026/20 season. I collected all of it and found home win rate fell from 44.7 percent to 33.3 percent, while away-team average xG rose from 1.15 to 1.32. I proposed cutting the home coefficient in my model to 0.18 goals per match, ran a chi-square test that returned p = 0.045, and published it with a warning about the small sample. When home stopped being a fortress, I learned to listen to empty stands. Billiards has no equivalent test, simply because nobody records table conditions.

The human element still sits outside every spreadsheet. The pressure of a deciding break, the changed rhythm of the cue arm after a missed visit, the psychology of watching an opponent run eight points in front of you — none of that converts into a general average. Handwritten notes let me see it. They do not let me quantify it.

People like to say more data means better predictions. With billiards I am not so sure. The decisive variable in a visit is a single contact point between cue ball and cushion, measured in millimetres and shaped by things nobody controls. The share of variance that technique explains in billiards is far smaller than in football. More data will raise explanatory power; it will not necessarily raise predictive power by the same amount. Anyone who promises you otherwise deserves one question back: over how many innings was your data collected?

The bigger trap lies in how numbers get used. A seven-match tournament at an average of 1.45 is a media miracle, but not enough to call someone a 1.45 player. I once sat through a forum argument about the strongest player in Vietnam based on the two most recent events, and the only thing I learned was that more people joined the argument than there were matches with statistics. Three thousand matches taught me that one match can teach more than all of them. In a sport where three thousand matches pass without leaving a single line of data, one fully recorded match teaches more than the whole of memory combined.

I also have to warn myself. The habit of waiting for more data has delayed my publishing many times. Once I kept a conclusion in a drawer for four months simply because I wanted 50 more observations, when it had been solid enough to say out loud in month two. In billiards, where the baseline is empty, waiting means never publishing at all.

The largest risk sits on the governance side. Match-fixing monitoring in professional sport works by comparing the behaviour of a single match against a player's own statistical baseline. No baseline, no anomaly. An unexpected average of 1.8 at a small event cannot be cross-checked against 1.2 the previous week, because the 1.2 does not exist in any public record.

This season I keep my own notebook, with a target of 100 matches logged across four fields: points per inning, run length, dangerous-leave rate, and table conditions. No model yet, no coefficients, only steady hand-recorded notes. When someone hands me a number about Vietnamese billiards, my first question will always be: who measured it, how, and on what table.

I do not write to convince anyone. I write so that data has a witness. The next Vietnamese billiards title will arrive, and within 72 hours almost nobody will remember the champion's general average. If we do not record it, ten years from now we will still be arguing from memory. So who opens the first notebook?