Empty Records in Football's Analytics Pipeline: An Unclosed Control Gap

A piece of analytical output was escalated to the deep-assessment layer. Title: N/A. Source: N/A. Article type: unclassified. One-sentence summary: blank. Author stance: N/A. Article purpose: N/A. Information points: empty list. Entities involved: cannot be identified, because the information-points list above is empty. Time sensitivity: not assessed. Source quality: not assessed, because the source fields themselves carry the value N/A.

Nine analytical dimensions were required to be output in full. None of them had data to assess. The record still travelled the whole pipeline, the skeleton stayed intact, and the line entered into every cell was "insufficient information, cannot assess".

The result is a dossier that looks full and is empty inside. Forty-four years of watching this industry taught me to tell two kinds of gap apart. The first kind is deliberate: a closed press room, a contract annex pulled from the archive, a server log overwritten on the very day someone asked a question. The second kind is accidental, produced by a broken valve somewhere in the system. This record belongs to the second kind, which is precisely why it deserves to be written about. The record says more about the information-production machine than about any particular club.

THE FOOTBALL WRITING TRADE HAS CHANGED TWICE IN MY LIFETIME

When I started out at local radio stations in 2026, the tools were videotape, a notebook and a pay-per-minute telephone. To prove a defender was positioned wrongly, I had to borrow a colleague's tape, rewind it thirty times in one night, and write out the minute and the coordinates by hand. The tools now are APIs, server logs and probabilistic models. In 2026 they closed the press room; three decades later I prised the file wide open. At the 2026 World Cup in the United States I was the only woman in the press conference room for a quarter-final, and there I learned that an answer without video evidence is not an answer.

The football analytics industry has boomed in two directions at once. Clubs have built their own data departments and hired from statistics and physics. Broadcasters have set up tactical graphics units. Agents use valuation models in negotiations. And in between, a new middle layer has appeared: organisations that collect, process and convert football text into structured data.

That process usually runs in two steps. Step one, extraction: read the article, pull out title, source, article type, list of information points, list of entities, time sensitivity, source quality. Step two, deep assessment: split into nine dimensions, from tactics and transfer finance through results, league landscape, rules and governance, dressing room, risk profile and media expectation, all the way to industry-wide transmission.

In other words, what Vietnamese newsrooms have imported in recent years is no longer an article but an entire assembly line. And every assembly line has a point of failure: the joint between step one and step two.

The vocabulary here is dense, so let me slow down. xG, expected goals, is the probability that a shot becomes a goal, based on location, angle, shot type and number of defenders. PPDA is the number of passes the opponent is allowed per defensive action; the lower the figure, the higher the press. FFP is UEFA's financial rulebook; PSR is the Premier League's counterpart. La Liga's salary cap limits total squad spending relative to revenue. Transfer amortisation spreads a purchase fee across the years of a contract. A sell-on clause gives a former club a percentage of the next sale. The final contract year is the season in which negotiating leverage changes hands. The "FIFA virus" is what happens when players return from international duty overloaded or injured.

Those concepts only carry weight when real data sits behind them. The record I am describing does not contain a single line.

NINE DIMENSIONS COLLAPSE, AND WHY EACH ONE COLLAPSES

The first dimension is tactics and technique. To judge how sophisticated a system is, you need at minimum a formation, a playing style, a pattern of personnel use, and at least one comparison anchor such as a similar opponent or the league norm. The record has no formation, no style, no possession or passing data. The conclusion entered in the cell is: cannot assess. I agree with that conclusion, and I want to stress the reason. A tactical analysis written without a shape and without process data is simply fiction.

The second dimension is club finance and the transfer market. The standard frame has four lines: broadcast revenue, commercial revenue, wage bill, net debt. No club is named, so all four lines are empty. No transaction is mentioned, so a contract cannot be decomposed into instalments, performance add-ons, sell-on clauses or release clauses. And with no wage data, the wages-to-revenue ratio cannot be computed - the single most important indicator of whether a squad is living within its own means.

The third dimension is results and the public-opinion cycle. It needs a table, a form sequence, a fixture density. The sample here is zero matches. No sample means no form curve, no sack pressure, no gap between expectation and reality.

The fourth dimension is the league landscape. With no league, no division and no club, you cannot draw the competitive hierarchy from title contenders down to the relegation group. You cannot compare squad value, financial power or academy output.

The fifth dimension is rules and governance. This is the dimension I care about most as an investigative writer, because it links data to accountability. The checklist has four cells: financial fair play, transfer registration rules, disciplinary sanctions, competition eligibility. When no conduct is described, no precedent can be mapped onto the file. The precedents are available and heavy: the Manchester City case, the points deductions handed to Everton and Nottingham Forest in the Premier League in 2026-24, the Juventus case in Serie A. But mapping precedent onto a void is the work of a fiction writer, not an investigator. The conclusion recorded is: cannot assess, and that is the correct conclusion.

The sixth dimension is management and the dressing room. To talk about a dressing room you need to know who the captain is, whether factions exist, how long the key players' contracts run, where each sits on the career arc. Not a single individual is named in the record - not an owner, not a sporting director, not a coach, not a player.

The seventh dimension is the risk profile, and this is the only place where the record dares to issue a rating. All six industry risk categories - sporting, financial, personnel, regulatory, reputational, systemic - are left blank. But one risk is graded High, and it sits in the sixth group: systemic risk. Specifically, the risk that an analysis was launched on top of an input containing no content. This is the sentence I want newsrooms to read slowly. The biggest risk in an assembly line is not that it writes something wrong, but that it runs the process perfectly while having nothing to write about.

The eighth dimension is media and expectation. With no headline and no source, no information can be graded for credibility. In my trade, source tier is everything: a named journalist who has covered one club for fifteen years is not the same as an aggregator page with no editor accountable for it. With no source tier, all information is equal, and when everything is equal, nothing is trustworthy.

The ninth dimension is industry-wide transmission. Transmission analysis only means something when there is a triggering event: a transfer, a rule change, a commercial deal. No event, nothing to transmit.

Nine dimensions, nine empty cells. But the point here does not stop at nine empty cells. The point is that an empty record made it through the entry gate of the deep-assessment layer, and was stopped only by the honesty of the writer at the far end of the line.

I have been in this trade long enough to recognise the pattern. I do not trust transfer fees; I trust the figures that have been struck through. Girona inflated a player's price with a bot network; the real value was in the server log. Betis hid doping in a contract annex; I read backwards page by page until I found it. Three times in a row, the answer was not in the published number but in the number that was missing. The same applies here: the problem is not the nine empty cells, it is that no valve closed when the file walked in.

Picture the concrete risk. An empty record crosses into layer two. Layer two is obliged to output all nine dimensions. A model not trained in null-handling discipline will fill the blanks with the most plausible content available: a mid-table club, a manager under pressure, a Brazilian winger, a round transfer fee. All of it sounds entirely real, and none of it exists. That is fabrication risk, and it sits at High.

Next comes downstream contamination. If unassessable cells are emitted as default values - "medium risk", "club unidentified" - then filters further down the chain will ingest false signals. A transfer-rumour filter may count a phantom club into its own statistics. A small error, but multiplied across records it becomes a false trend.

Third is traceability loss. With no source, no origin URL and no trace of the first fetch, this record can never be audited or re-linked to the original text.

And the most worrying element is that the failure is systematic. If extraction returned empty for technical reasons - a 403, a 404, a paywall, an anti-bot challenge, or a malformed document - then this is a silent defect. Silent defects never happen only once. They happen across a whole batch, and nobody knows until someone sits down and counts.

Counting is simple: track the share of records carrying at least one information point across each batch. If that share falls close to zero over several batches, it is an early warning of systemic failure rather than a content trend. In parallel, fetch errors should be taxonomised into separate groups: access errors, paywall errors, parsing errors, malformed-document errors. Whichever group dominates tells you whether the fault is on the access side or the processing side.

The technical fix is not expensive: a minimum-content validation gate before handover to the assessment layer, requiring at least one information point, a resolvable source, a non-empty entity list and a time-sensitivity assessment. Fail, and stop loudly.

AN EMPTY RECORD IS A DATA VALUE, NOT A FAILURE

There is a counter-reading worth taking seriously. Many people in the industry will say that all nine dimensions returning "cannot assess" means the model performed badly and a better model is needed. I think that diagnosis has the wrong address.

The record behaved correctly. It kept the skeleton, it did not invent a club, it did not assign a convenient medium risk rating, it refused to fill the blanks with plausible-sounding names. By the standards of information discipline, that is the behaviour of a good file. A machine willing to return an empty string is more trustworthy than a machine that always knows how to fill space.

The problem lies elsewhere: in sports media, the value of a record is measured by length and fluency, not by the accuracy of the places where it declines to speak. Algorithms reward complete content. Editors are under volume pressure. And audiences, quite naturally, want a story to argue about every day. That is the structural pressure pushing the whole line towards filling gaps.

So the solution is not in the model. The solution is to accept that an empty string, if flagged distinctly and impossible to confuse with a real value, is a valid and useful result. This record, in its current form, has methodological value: it is a template for handling data scarcity without generating content.

Its weakness lies elsewhere: it was not born in the right place. A file like this should stop at step one, with a full error log and an origin URL attached. The fact that it reached step two shows nobody answered the simplest question of all - where did the data come from.

CHAIN OF CUSTODY FOR INFORMATION

In a doping laboratory, a blood sample is not trusted because it exists, but because it has a chain of custody: who took it, when, how it was sealed, who opened it, where, and whether a camera was running. A sample that loses its chain of custody is thrown out, whatever the test result says.

Football has no chain of custody for information. An analysis is published and nobody knows where the original text sits, when it was retrieved, who processed it, or at which step the data was dropped. Until this trade accepts that an honestly recorded blank is worth more than a carefully completed conclusion, every analytical pipeline is still running with a valve open that nobody closes. From the press room in 2026 to the Girona bot network in 2026, power has only changed shirts - and the open valve is always the last detail anyone bothers to read.