The Null Result: When a Sports Analysis Pipeline Has Nothing to Say

Manchester, a late-November morning. On my screen are nine tables stacked vertically, each with four to six rows, and nearly every cell carries the same short line: insufficient information. The tactical structure table is empty. The club finance table is empty. The league landscape table is empty. The risk profile table is empty. Down to the last table, on transmission effects across the football industry, still empty.

I am used to tables that are full. In nineteen years in this trade I have written about nights when every column danced, when every pass became a chart and every phase of play a data point. Looking at those nine blank pages, I think of something else entirely: the empty stands at Old Trafford during the months in 2026 when stadiums were closed, and Paul, the fifty-eight-year-old cleaner who had swept that place for twenty years, telling me that at night he could still hear the shouting echoing back from seats where nobody sat.

An empty seat still has someone sitting in it — we simply no longer hear their applause.

What sits in front of me is the output of a two-stage analysis pipeline. Stage one reads a source text and breaks it into discrete information points: an event, an entity, a timestamp, a verifiable figure. Stage two takes those points and applies them across nine analytical dimensions — tactical and technical structure; club finance and the transfer market; results and the opinion cycle; league landscape and team positioning; rules and governance compliance; management and the dressing room; risk profile; media narrative and expectation; and finally the transmission effects across the football industry.

Stage one returned an empty list. No title. No source. No article type. No season, no transfer window, no date. Not a single club, player or coach identified. When the upstream is empty, all nine dimensions below it close, and the analysis writes its own status: void, re-run required.

The interesting part lies elsewhere. A null result is not a wrong result. It is evidence that the fault sits upstream, not in the analytical layer. In this trade the two are constantly confused. "No signal found" is entirely different from "negative signal". A team that keeps a clean sheet because the opponent missed the target is one story. A team that keeps a clean sheet because the match was never recorded is another story, and it does not belong in a statistics table.

And this is the part that stopped me longest. The nine-dimension framework is highly generative. Put it in front of an empty input, and a language model under pressure to fill the template will produce content that is fluent, plausible, structured, and entirely fabricated. It will write about a back-three system that never existed. It will construct a wage-to-revenue ratio never published, a net debt figure with three decimal places, a salary cap nobody ever imposed. It will cite expected goals, which estimates the probability a shot becomes a goal, and passes allowed per defensive action, which measures pressing intensity. Both are real metrics, properly defined, and utterly meaningless when attached to a match that never happened.

Readers struggle to catch it. A wrong percentage does not incriminate itself. If anything, it looks far more professional than an empty cell. That is why the largest risk in this pipeline sits in presentation, not in computation. The report ranks that danger at the highest level and attaches an unhedged recommendation: any output that fills these tables without citing specific information points must be treated as invalid and discarded.

The failure signature here is equally concrete. The template scaffolding is emitted intact — section headings, bullet stubs — while the body is empty. That pattern usually indicates the source text never reached the extractor, or reached it in an unreadable state. In other words, the problem is in ingestion. To fix it you inspect where the text comes in, not where the conclusions come out.

There is a third trap, subtler and more dangerous for the reader. In an empty input, no allegation and a clean record are indistinguishable. The input contains no compliance content, no financial content, no injury information, and that silence grants nobody a certificate. Conversely, a "low risk" verdict in the absence of information is an actively misleading conclusion, because risk is a property of a subject-plus-exposure pair. Absence of information is not absence of risk.

Even the information-value scoring refuses to grade, and does so with discipline. One or two stars would imply a judgement about the original article, when the original article's content was never captured. The correct rating is unrateable, not poor.

In Moscow, on the night of 11 July 2026, at the Luzhniki, Croatia beat England 2-1 after extra time, Mario Mandžukić's decisive goal arriving in the 109th minute. I was there. I wrote nothing for six days afterwards. Had I forced myself to file that night, I would have produced a full, fluent, statistic-laden account that was wrong in substance, because I did not yet understand what had just broken. In Moscow that night, I learned that the final whistle is only a rest.

The same pressure repeated itself in 2026, when I interviewed Phil Foden, then a seventeen-year-old at Manchester City, after the FA Youth Cup final. The conversation lasted 34 minutes and he spoke twelve sentences, mostly about the team bus on the way home. The desk asked me to rewrite it as a rising-star piece and cut all the awkwardness. I did not file that version. The keyboard always has a better version than the truth ready to hand, and the writer's job is to refuse it.

Based on my experience covering matches, what does not happen is always harder to read than what does. A missed chance leaves a trace. A goal not scored leaves a gap, and gaps keep no minutes.

The sports industry pays for volume. It pays for the number of pieces, tables, metrics, conclusions. A system that answers "I don't know" is treated as broken, and whoever runs it as not having done the work. The paradox sits in the middle: the only honest thing in an empty input is the only thing judged a failure.

In Vietnam that paradox has its own dialect. We import analytical templates from Western finance and governance — margins, net debt, salary caps, financial fair play — without importing the data infrastructure that comes with them. The result is beautiful tables resting on empty cells. Every transfer window a source close to the deal appears, a fee circulates, and nobody can check where it came from, over how many years it is paid, or which add-ons it contains. What is verified stays small. What is presented stays large.

That analysis does the opposite, which is why it deserves reading as a document about the craft. It lists what a re-run requires: at least five discrete information points, at least one named club or competition, a source that can be quality-graded, and a time reference. Those four conditions are not dry technical requirements. They are the minimum definition of an article that can be checked.

I have covered eight Olympic Games, eight World Cups, and many grand tours across Europe, and what I brought home was not a faster ability to read a match. It was the ability to recognise when I had nothing to say yet. On the track, records are measured in hundredths of a second; outside it, lives are measured in breaths. In a data room, the distance between those two things is a single empty cell.

Before anyone becomes a name, they are only a running figure. That is true of a debutant, and equally true of a dataset just loaded into a machine. A good match is never fully told; it only waits for someone quiet enough to hear. If our analysis systems cannot say "I don't know", what exactly are they saying?