When the Spreadsheet Returns N/A: Data Discipline and the Trap of Belief

One morning in March 2026, I reopened my tracking file and found the most important data column completely empty. Not a few missing rows — empty from the first row to the last. The first reflex of anyone who has done this work is to fill it in: with an estimate, with memory, with a number that sounds reasonable. I moved my hand toward the keyboard. Then I stopped.

Six years of working with spreadsheets taught me that the hardest moment is never when a model is wrong. Wrong can be fixed; fix it and run it again. The hardest moment is when the data is missing and the writer is tempted to tell a story to cover the gap. Every time that happens, analysis quietly turns into inference — and inference dressed as data is the most dangerous kind of information in sport.

That morning I left the cell empty. I typed N/A, closed the file, and went back to step one: verify the source.

Refusing to guess is a skill, not a weakness

Sports data in Vietnam has travelled a long way in a decade. Around 2026 there was almost no detailed data source for the V-League. To analyse a match I had to click through every phase from video, define every variable myself, and carry the error margin alone. Today the major regional tournaments produce fairly complete event data, federations publish technical reports, and aggregator platforms allow cross-checking. But more data does not mean better data.

I divide sources into four tiers and I never mix them.

Tier one is raw event data: shot coordinates, timestamps, the player involved, the outcome. This is the only tier I allow as calculation input.

Tier two is technical reports published by tournament organisers or federations. This tier is for cross-checking, not for causal reasoning.

Tier three is data aggregation platforms, where I check whether my number sits far from the general baseline. A small gap is normal; a large gap means I am wrong, not that the world is wrong.

Tier four is media interpretation. It is useful for understanding public expectation, but it is never allowed to become model input. Confusing tier four with tier one is the most common error I see among newcomers, and it always costs a wrong prediction.

The evidence chain: measured, inferred, decided

Every conclusion of mine passes through three layers, and I do not let myself skip one.

The measured layer answers what happened. The inferred layer answers what it means. The decision layer answers what to do. The fatal mistake is blending the second into the first: taking interpretation as evidence and then building the third layer on that foundation.

My first xG spreadsheet taught me: every goal has a hidden story. Since then I keep the habit of noting beside each figure which layer it belongs to, and where it came from.

Based on my experience tracking matches in Southeast Asia, the most recent example I followed closely was the ASEAN Championship 2026. On 2 January 2026 Vietnam beat Thailand 2-1 in the first leg at Viet Tri. On 5 January 2026 Vietnam won 3-2 in Bangkok and took the title 5-3 on aggregate. Nguyen Xuan Son scored seven goals, won the Golden Boot and the Player of the Tournament award, then suffered a serious injury in the second leg.

That is the measured layer. It says nothing on its own about the future.

The inferred layer is where the argument actually lives. A striker scoring seven goals in a short tournament does not automatically mean his team has solved its attacking problem. How many shots produced those seven goals? What was the quality of the chances? What share was skill and what share was position? If I cannot answer, I have no right to talk about the future.

This is where most football content separates from analysis. Media jumps straight from layer one to layer three: there is a goal, therefore a conclusion about strength. I refuse that step.

In the summer 2026 transfer window I assessed a target striker for a mid-table club. My model showed his actual goals were 4.5 below expectation across a season. That is not a sign of decline; it is a sign of a small sample and a little bad luck. The club signed him and he scored on the opening matchday.

But I have to be honest about my limits. The same model cannot measure whether he would fit the dressing room. That gap I left empty, and I told the club plainly that it sat outside my scope. A player's value is just a number — until you read the error in how it is calculated.

At the defensive layer, data speaks first

There is a paradox I meet often: the most reliable signals sit in the part audiences remember least. A goal is replayed ten times; a well-positioned block is replayed never.

That is why I built the habit of opening every report with defensive data. Pressures per opponent pass, average distance between lines, ball recoveries in the opponent's half — these indicators are far less noisy than goals. A team can win on a single moment; a defensive system only improves after thousands of phases.

Morocco 2026: when defensive data spoke first, the world listened later. I was eighteen that year, still at school, but I had enough method to recognise an aggressive defensive block while most viewers looked only at their low possession share and called it luck.

When home is no longer home, I am forced to rewrite every assumption. The pandemic season of 2026 taught me that unforgettably: football returned in empty stadiums and my entire home-advantage model became meaningless overnight. I had to rebuild from scratch, and for the first time a prediction from my raw data came true within three rounds.

Data gaps are always filled with prejudice

This is the part I want readers to carry away.

When a dataset is incomplete, it does not stay silent. It gets filled. And what fills it is almost always a pre-existing prejudice: big clubs must be stronger, expensive players must be better, big wins must mean better football.

I have followed the Vietnamese League of Legends scene for years, and 2026 was an expensive lesson. A large group of individuals was suspended after an investigation into match-fixing, running into the dozens. Before the official conclusion, forums already knew who was guilty and who was innocent — based on form, based on facial expressions, based on defeats that looked strange. Almost all of those guesses were presented as though verified.

When the real list was published, it did not match the list the community believed it already knew. That is the price of filling gaps with belief.

Football works the same way. A defender who errs in a loss is called weak. A goalkeeper beaten by a hard shot is called poor. A coach whose substitutions fail is called conservative. All of these are layer-three conclusions built on layer-one evidence that is close to zero.

I do not predict the future with intuition; I only read the traces numbers leave behind. When there are no traces, the only honest answer is to leave the cell empty.

One thing I learned from my own mistake. On 13 August 2026 I filed a report late because I was trying to make the model hit a perfect standard I had set myself. A colleague told me something I have never forgotten: a model that is eighty percent accurate and filed on time is worth more than a perfect model filed after the match has ended. Some people treat an empty cell as failure. I treat it as a conclusion — provided it arrives on time.

On referees and VAR

The same logic applies to VAR. Controversy does not disappear when technology arrives; it moves. It used to happen on the pitch, between referee and player. Now it happens in the review room, between people reading the law and a grey area that has never been defined clearly enough.

What the data shows is that the number of disputed decisions does not fall at the rate the public expects. It changes type. From the referee did not see it to the measurement is wrong. Both are problems of definition, and definition cannot be solved by bolting on more cameras.

For anyone working in data this is a familiar lesson: adding data does not automatically reduce disagreement. It only relocates disagreement somewhere nobody can see it.

Signals for the next round

I will track three things in the coming period.

First, the quality of public data in domestic competitions, especially positional and pressing data. As tier one thickens, the value of tier four falls automatically, and that is good news for readers.

Second, how clubs value young players. Current models are very good at measuring technical potential and very poor at measuring dressing-room chemistry. A player with beautiful numbers in the wrong environment looks like a bad investment, even when he himself has not changed.

Third, and most importantly, the frequency with which sports writers dare to publish N/A. Every time someone leaves a blank instead of guessing, the information baseline of the whole game rises a little. Football and esports differ on the surface, but the same layer of data sits underneath — and the same temptation sits underneath it too.

For anyone patient enough to wait a season to prove a number — leave the cell empty. It will fill itself in, with data, in a way no commentary can.