The Longest N/A Column of My Career: A Night of Billiards Analysis in Hai Phong and the Limits of Data

The clock read 1:47 a.m. in Hai Phong. I reopened the data folder for a domestic billiards tournament I had followed for three full days of play. The spreadsheet had been built in advance: a date column, a player column, an innings column, a points-per-innings column, a high-run column, a win-rate-after-trailing column. Structurally tidy, exactly how a decent set of notes should look. The content held a single value repeated on every row: N/A.

I stared at that column for a while. Seven years earlier I had sat at this same desk with a spreadsheet that was full. Matchday 18 of the 2026 V.League season, Hai Phong FC against Sanna Khanh Hoa. I pulled the Understat numbers, saw Hai Phong generating 2.8 expected goals against 1.0 for the visitors, and predicted a 3-1 win. The match ended 0-1. Goalkeeper Tran Buu Ngoc made seven saves and demolished my entire model. I was wrong that night, but I was wrong on top of real numbers.

Tonight was different. There was no bad data to misread, only an absence. And an absence cannot be analysed.

Billiards does not use the match as its unit of analysis. Its unit is the inning. A three-cushion game to 40 points usually runs 40 to 50 innings, and each inning is a decision: attack, play safe, or push the balls into a neutral position. To say anything with weight about a player, I need inning-level data, not match-level data.

What is publicly available for Vietnamese billiards is thin. The world federation ranking, qualifying results, final scores, and occasionally the tournament organiser publishing the champion's average and high run. What is missing: a per-inning sheet, successful safety rate, deliberation time per shot, table condition, hall humidity, and above all a long enough series to compare anything against.

The domestic three-cushion scene is rising fast. Tran Quyet Chien, Bao Phuong Vinh, Duong Quoc Hoang and Ma Minh Cam are the names that have made Vietnamese audiences familiar with world-class finals; I still copy their results into a separate notebook for later cross-checking, because the habit of the trade forces me to find a second source before I believe anything. National championships, open cups, pool halls across Hai Phong, Da Nang and Ho Chi Minh City keep appearing. The data infrastructure, however, lags far behind the competitive infrastructure.

I picture my whole analytical process as a two-stage pipeline. The first stage collects: who played, who won, how they won, under what conditions. The second stage interprets: compares against context, looks for patterns, builds a model. When the first stage returns empty, the second can do nothing but write N/A into every cell. My report that night was structurally flawless and informationally hollow. That is an honest output, but it helps nobody understand a single player better.

Data never lies, but I have misheard it before. Seven years ago I misheard it because I believed expected goals covered the whole story of a match. It does not cover goalkeeper form, especially in games where the weaker side deliberately sits in a low block. After the Buu Ngoc shock I dropped the habit of concluding from one metric and switched to hand-notating twenty consecutive matches so I could check myself. Hand-notating is painfully slow, but it taught me something no software can: you have to know the conditions under which your data was born.

Before I use any figure, I run a checklist. Who measured it? How was it measured? Under what conditions? What is this index hiding? What is the sample size? Against my billiards spreadsheet, all five questions returned the same answer. Nobody measured. There was no method. No conditions were recorded. And the sample size was zero.

What is telling is that I can still picture the metrics that would change how a billiards match is read. Points per inning is the most basic, but it blends three different things: attacking ability, safety ability, and the quality of the opponent. The high run is even more seductive. A run of twelve in a single inning can lift a player's match average into attractive territory while the other thirty innings of that same player were a struggle. You need the standard deviation between innings, not the mean, before you can see consistency.

One goalkeeper missing a catch is a mistake. Three goalkeepers missing catches is a signal. That principle applies to billiards differently. A single missed shot is personal. But three different players missing from the same position, in the same hall, in the same afternoon slot, means the thing that is broken may be the table rather than the person. To tell those two possibilities apart, I need recorded playing conditions. Without them, every technical conclusion is a guess dressed in the clothing of numbers.

In pool, the metrics are far more standardised: break-and-run rate, safety success rate, win rate after the opponent breaks. Those are variables you can compare across players, tournaments and years. Three-cushion is harder, because every inning is a geometry problem with countless solutions, and the quality of a solution cannot be captured by a win-loss value alone.

This asymmetry has an economic cause. Football has detailed data providers because there is a betting market and a pay-TV market large enough to fund them. Professional billiards has a far thinner market, and in Vietnam a complete value chain has barely formed: halls, equipment, tournaments, media, sponsorship and youth development are still being pieced together. When nobody pays for data, data does not get produced. That is a structural problem, beyond the reach of any individual.

The world ranking is the clearest example of how a single index can mislead. The ranking is built from points accumulated across events, but it says nothing about current form, about a player who just changed his cue, about a player returning from a long layoff. Using it to predict a final is the convenient approach, and it is usually wrong.

The hardest part of this job is not the model. It is accepting that there are things you cannot yet say. An analysis made entirely of N/A sounds like a failure, but it is an honest description of the state of the data infrastructure. Had I invented a coefficient to make the spreadsheet look fuller, I would have betrayed the very principle that kept me at my desk at nearly two in the morning.

The biggest risk in this trade sits in a confident conclusion built on an empty foundation, not in a wrong number. Fans still argue about form and nerve on forums every evening, and that argument needs no line of notation at all. When data goes quiet, narrative fills the gap. That is a rule, not anyone's fault.

I have also fooled myself in the opposite direction. There was a period when I leaned on the phrase insufficient data to postpone every conclusion, and caution became a hiding place. Transparency about limits is a virtue; using limits as a shield so you never have to be accountable is a vice disguised in technical language.

Another blind spot sits where models increasingly push into the dressing room. I once read a quantitative analysis of a player that was right in its numbers but off in its rhythm. Nobody measures the pause between two innings, the look in the eyes before the decisive shot, or the way a player breathes when trailing by three points in the fortieth inning. Billiards is a sport where pressure converts directly into geometric error, and a player's feel is data, simply data nobody has written down.

When home stops being a fortress, I learn to listen to an empty stand. Vietnamese billiards has never had a season without spectators the way football did, but it has another version of the same problem: events held in near-empty pool halls, the air compressed, the click of the balls so clear a player can hear his own breathing. Every home-advantage coefficient I had ever calculated for other sports becomes meaningless there. That experience taught me that context is not the decoration of data. It is part of the data.

My plan for the coming months is simple and old-fashioned. Hand-notate twenty national-level three-cushion matches, inning by inning, with notes on table conditions and humidity. Ask organisers for inning sheets, even if nobody has ever asked them before. Publish my own errors openly, because data has value only when someone can check it again.

I do not write to convince anyone. I write so that the data has a witness. If in eighteen months the Vietnamese billiards data column is still blank, that too will be worth writing down, and it will be more trustworthy than any dazzling prediction.