The Notebook on the Billiards Table: Why Data Never Tells the Whole Story

Late on the twelfth of November, I sat alone in a small apartment in Hai Phong, replaying the footage of an international billiards final. The statistics panel on the right side of the screen reported that the leading player had an average scoring rate of 1.847 per inning — a figure in the highest bracket of the season. But when I fast-forwarded to the twenty-third frame, the numbers began to fall apart. That player ran three consecutive shots that I had personally marked in red ink as “impossible to complete at that angle” — all three went in. I sat still for a long while, then added one line to my notebook: the data is not wrong, the person reading it is. Data never lies, but I have misheard it before.

I began keeping systematic records on billiards at seventeen. Back then, I had just been burned by a model built on expected-goals that predicted a football match wrongly, and I assumed I could carry the same toolkit over to the billiards table. The result was a disaster. Billiards does not operate the way football operates. There is no expected value for a shot, no model for a safety exchange. You can count successful shots, points, and innings, but nobody counts the pressure of a single visit in a deciding frame.

So I built my own criteria. Every match, I record four things: the starting layout, the table speed, the humidity of the hall, and the number of times a player actively changes their solution mid-visit. Those four variables, after nearly a thousand matches, have taught me more than any automated statistics panel. Because they force me to sit still, to look, and to ask myself why this player chose that line instead of the safer option. The model knew in October. I only found the nerve to believe it in May.

Vietnamese billiards is a peculiar market. Anyone who has walked into a club in Hai Phong, Da Nang, or Ho Chi Minh City in the evening will notice one thing: the tables are never empty. But the paradox is that, despite an extremely strong grassroots scene, the amount of publicly available data on players is very thin. Domestic tournaments rarely publish shot-by-shot detail. International tournaments do, but the data is split across several organisations and scoring systems.

I spent two full years comparing data from three major tournament systems. The first thing I learned: the same player, in the same season, can show an average scoring rate that differs by as much as 0.3 points per inning depending on the measurement system. That gap does not come from form. It comes from how “an inning” is defined. Some systems count failed visits, others only count successful ones. This is why I never use a single figure on its own.

A metric without a published method is not a metric. It is just a number someone wants me to believe.

In my notebook, I keep a section called “conditions to verify”. Before I use any figure, I must answer four questions: who measured it, how it was measured, under what conditions, and what it is hiding. It may sound extreme, but I have paid the price. At twenty-two, I used average scoring data to predict a semi-final. The model gave me a razor-thin result. The match ended the opposite way, and the most striking indicator held by the winner never appeared in the data I had.

That player won not by scoring more. He won by controlling rhythm. Every third visit, he pushed the balls into a difficult layout, forcing his opponent into a defensive call, and that “rhythm” is counted by no system. Correlation is not causation. A player can have a lower average scoring rate and still win, because he wins in the moments the statistics treat as “ordinary visits”.

This is the part I have rewritten many times. I once believed that a player with a higher average scoring rate had a higher probability of winning a title. It sounded entirely reasonable. But when I went back and analysed seventy finals over three years, the opposite appeared more often than I expected. In nearly half of those matches, the player with the higher season-wide average scoring rate did not lift the trophy. The winner was usually the player with lower variance — fewer swings, fewer explosions, but also fewer collapses.

Stability beats variance. That is what the statistics say, but they do not tell me why.

I have no complete answer. But after rewatching dozens of matches, I have a hypothesis: in billiards, a missed visit does not only lose points — it hands the opponent control of the table. The loss in points may be small, but the loss in position is large. And position is what statistics do not count. That is why I always annotate each visit: “lost points” or “lost position”.

The crowd laughed. The data did not. Since that day, whenever someone quotes a number without stating its method, I pause for a second and ask myself what I am being shown, and what is being hidden behind it. Not out of suspicion, but because I once misheard a number and lost an entire weekend rewriting a report.

So what truly separates an analyst from someone who merely reads a scoreboard? I believe it is the acceptance that one can be wrong. If my data predicts one outcome and the match goes another way, I do not rush to blame the data. I ask what my data is missing. What is missing is usually the unmeasurable: psychology, the rhythm of breaks between frames, even the noise in the arena. Three thousand matches taught me that one match can teach more than all of them.

There was one season when I followed every match of an international event with limited spectators. The interesting part was not the results. The interesting part was this: when the stands were emptier, players tended to take more risks with long, difficult shots. Shots they would normally play safe, they now attacked directly. My data showed the attacking call rate rising, but the success rate falling slightly. When home is no longer a fortress, I learned to listen to the empty stands.

I am not saying spectators do not matter. The opposite. I am saying the playing environment is a real variable, and any model that ignores it is incomplete. A player who performs well in front of a crowd may perform worse in silence, and vice versa. That does not make data useless. It means data must be read alongside context.

I make a habit of writing dates next to every figure. A metric that is right today may be wrong next month. Form changes, table speed changes, even the rules on preparation time between visits change. Every coefficient has an expiry date. I once used a coefficient built on last season's data to predict this season, and was so wrong that I apologised to readers in the following article.

That lesson taught me two things. First, old data must be rechecked before reuse. Second, sometimes the most honest way to handle a broken model is to say publicly that it is broken. I have done so at least three times in my short writing career. Each time, I lost a little face in the short term but kept the respect in the long term.

Comparing billiards nations has also taught me a great deal. Billiards in Vietnam has a feature many places lack: a dense and widespread club system. A young player in Hai Phong can play against older regulars at a nearby club every evening. In many countries, that chance exists only in professional academies. This accessibility is both strength and weakness. The strength is a broad base. The weakness is the lack of a structured pathway.

In the last two years, I have seen a notable trend: more young players are approaching data. They do not merely practise shots — they review footage, they record their own success rates. That is a positive sign. But I worry too. There is a trap behind it: when players trust their own numbers too much, they begin to avoid risky shots that data labels “high risk”. Yet in billiards, the reward sometimes lies precisely in those shots.

The best data is the data that makes me ask the question again, not the data that makes me stop thinking.

I do not write to convince anyone. I write so that the data has a witness. Every analysis of mine, long or short, begins with the question “what needs verifying”, and ends with a list of what I still do not know. Some readers say I write drily. Some say I am too cautious. I accept it. Because I would rather be called dry than hand a reader a number whose origin I do not understand.

Looking back, in my early years I made the same mistake over and over: I tried to turn billiards into a maths problem, when billiards is a conversation. Each visit is a question, and each layout is an answer the player must find alone. Data can record the answer, but it cannot ask the question for me. That is why I still sit down after every match, the desk lamp still on, the notebook still open.

If there is one thing I want readers to take from this piece, it is this: next time someone hands you a billiards number — a win rate, a scoring rate, any metric — ask one more question. Not to nitpick, but to understand. Data never lies. But it only tells us the part of the story the person measuring chose to record. The rest, the most interesting part, is still on the table, waiting for someone willing to sit long enough to see it.