When the Data Isn't There: 'Null Input' and the Trap of Fabricated Facts in AI-Driven Football Analysis

An Empty File, and a Large Question A document that recently entered an automated sports-analysis pipeline had no headline, no source, no author stance, no publication date. It had no player name, no club name, no competition name, and not a single number. Every substantive field of the framework produced in the first stage was blank. When the analyst standing at the second stage picked up that framework, there was exactly one honest answer available: no football conclusion can be drawn from this. This report is about that incident. It is not a match report, not a club review, not an evaluation of any player's performance. It is the account of a procedural event in which the framework had to be completed despite the total absence of raw material. And that is precisely where the biggest risk in contemporary sports journalism and sports analysis now sits. The question is simple; the answer is uncomfortable. When an analytical system is forced to operate without information, what does it do? Two paths are open. One is to accept that there is no answer. The other is to fill the template's empty boxes with invented numbers and invented names written in confident prose. The first path is boring, but true. The second looks elegant, complete and publishable — and is dangerously deceptive. This piece explains why the second path is so easily taken, why it corrodes the basic ethics of sports journalism, and why a null result is not a failure but a successful quality control. What a Null Input Actually Is Modern sports-analysis frameworks usually split the work into two stages. In the first, the raw article is read and its content extracted: headline, source, one-sentence summary, author stance, article purpose, list of information points, named entities (clubs, players, coaches, competitions), time sensitivity and source quality. In the second, nine analytical dimensions are applied to that extracted material: tactics and technique; club finance and the transfer market; results and the public-opinion cycle; league landscape and team positioning; rules and governance; management and dressing room; risk profile; media narrative; and industry transmission. Now suppose the first stage comes back empty-handed. The information-point list contains not a single line. The entity list is absent. Time sensitivity has not been assessed. Source quality has not been verified. Under those conditions, the honest answer to every second-stage question is the same: insufficient information. But note that this answer is itself information. It tells us that something in the pipeline has broken. The likely causes can be named: the original text may have sat behind a paywall; a JavaScript-rendered page may have defeated extraction; the underlying asset may have been video or image rather than prose; or an empty body may simply have been passed downstream. Whatever the cause, the problem is infrastructural, not editorial. That yields the first real decision. Is the fault in the content or the infrastructure? If it is infrastructure, the fix is to repair the pipeline — re-fetch the source, verify outlet, author and timestamp, and re-run stage one. The fix is never to complete the framework with whatever happens to be available. Nine Dimensions, Nine Admissions Worth reviewing what each dimension returns when confronted with an empty input, because the pattern itself is instructive. Tactical analysis asks what the team's structure is, how it presses, how it builds up, what its set-piece design looks like. With no team or coach named, none of those questions can be answered. At the statistical level it asks for chance-quality metrics, pressing-intensity metrics, possession and pass-completion rates, conversion efficiency. With no numbers, none of it can be evaluated. Club finance asks about broadcasting revenue, commercial revenue, wage expenditure, net debt. On transfers it asks how far the fee exceeds fair value, how the contract is structured, what the wage curve looks like, whether sell-on or buy-back clauses exist. Without a single figure, a panic premium cannot be detected — measuring a premium requires at least one price and one verifiable comparison. Results analysis asks where the team stands against expectations, what recent form looks like, how difficult the fixture run has been, and how far process data diverges from results. With no league named and no sequence of results, all of it is impossible. League landscape analysis places a club in a tier — title contenders, European places, mid-table, relegation battle. That requires at least one club and one competition. Governance analysis checks financial rules, transfer registration, disciplinary sanctions and competition eligibility. With no charge, allegation or hearing cited, no worst-case, central or best-case sanction scenario can be modelled. Management analysis examines owner patience, recruitment quality, structural stability, leadership layers, manager-player relations, generational transition. Each of those needs names. Risk analysis separates sporting, financial, personnel, regulatory, reputational and systemic risk. Here, for once, one risk is genuinely identifiable: the systemic risk of acting on an unverified evidence base. Narrative analysis locates a story on its cycle — emergence, acceleration, climax, backlash. If even the headline and author stance cannot be extracted, the cycle cannot be positioned. Transmission analysis traces impact across the academy-to-broadcast chain. With no first-order event, second-order effects cannot be drawn. Nine dimensions. Nine zeros. And beside each zero, one honest admission. The Temptation of the Template, and False Precision Return to the central question. If someone is told to fill these nine boxes, and that someone is a language model, what happens? The difficulty is that a model's training material contains a great many football reports, transfer stories, match reviews and tactical breakdowns. From that material it knows what a good football article looks like — which passage carries which kind of number, which sentence uses which register of confidence. Faced with a blank box, it can produce prose that is highly credible, rhythmically polished and entirely publishable, in which every sentence sounds true and none has any relationship to reality. This phenomenon needs a name. Call it false precision: the condition in which the evidential basis behind a claim is zero while the language of the claim is so specific and so confident that readers take it for verified fact. A subtle but important distinction applies here. An opinion may be unproven and that is fine — the author owns the claim. But when numbers, names, quotations and events are unproven, the result is not opinion but fabrication. A columnist may write that a given club lacks balance in midfield; that is a view, and readers know it. But if someone writes that a given player's expected-goals contribution per ninety minutes is a specific figure, readers treat that as verified data — and may act on it in fantasy leagues, in betting markets, in squad selection. That is why the gravest risk in sports analysis is not bad analysis but an empty framework that looks like analysis. Bad analysis invites argument and can be corrected. Unfounded certainty cannot be corrected, because the reader does not know there is anything to correct. The Data Revolution: From Chance Quality to Pressing Intensity Some background is necessary. Match statistics were once narrow: who scored, who passed, who shot. The problem is that shot counts do not express attacking quality. Five shots from thirty yards and one shot from six yards cannot meaningfully be counted in the same column. Expected-goals models filled that gap, weighting each shot by position, angle, pressure and body shape to estimate the probability it becomes a goal. That makes it possible to see how many quality chances a team actually creates, whatever the scoreline. Alongside it came expected goals conceded, and with it the debate over whether a goalkeeper's performance reflects skill or variance. Pressing intensity is measured with an elegantly simple metric: how many passes an opponent is allowed before it performs a defensive action. A low figure means intense pressure; a high figure means passive pressure. It reveals where a team actually loses the ball — in the opponent's half or its own. The value of these metrics lies not in the numbers themselves but in their comparability. A single figure means little without a league average, a seasonal trend, an opponent adjustment. Which is the whole point: however sophisticated the framework, without input numbers the metrics do not exist. A model cannot forecast on empty data; it can only infer from a prior distribution — and calling that inference a forecast is not science, it is gambling. Finance and Transfers: Rules, Accounting and Premiums Football's financial layer is now analysis-driven too. European financial fair play rules and the English profit-and-sustainability regime both ask the same underlying question: how much is a club spending relative to its revenue, and is that spending sustainable? One accounting concept matters: a transfer fee is spread across the length of the contract rather than booked at once. The consequence is that selling a player books the whole fee as profit immediately, while buying spreads the cost over years. That asymmetry drives a good deal of end-of-season manoeuvring. Another concept is the panic premium — the excess a club pays over fair value under deadline pressure or public pressure. Detecting it requires two data points: fair value and price paid. Without either, the premium can be guessed at but not measured, and a guess cannot be presented as analysis. Wage health is measured the same way: the ratio of top wage to average wage, total wages to total revenue, the number of star contracts in the structure. All of it needs numbers. Without numbers there is only one word left — 'unusual'. And 'unusual' is not analysis. Agents, Third-Party Ownership and Training Compensation The regulatory layer needs care, because this is where the most misinformation circulates. One basic rule: a contracted player may not be approached directly without his club's permission. Proving such an approach is hard, because contact happens informally; here, documents rather than inference are required. Another rule: no third party — an investment fund, for instance — may hold a share of a player's economic rights. The arrangement is banned because it creates influence over a player's professional future from outside the club. The third rule is the least discussed and the most humane: training compensation and the solidarity mechanism. Clubs that trained a player during specified age bands are entitled to a share of a later transfer fee, sometimes many years afterwards. Money from a transfer therefore does not stay between two clubs; it flows down the academy chain. None of this can be analysed without names and documents. In an empty input, the only correct action is to acknowledge that there is no subject to discuss. Governance and Sanction Modelling: Three Scenarios, Zero Inputs Governance analysis is not always about finding allegations. It is about identifying which rule system an event engages, what outcomes would follow if a breach were alleged, and how likely those outcomes are. The conventional method builds three scenarios. Worst case: a heavy fine, a transfer ban, a points deduction, even exclusion from European competition. Central case: financial penalty and monitored conditions. Best case: the complaint is dismissed or never brought for lack of evidence. But building those three scenarios requires an allegation, a charge, or at minimum a reference to a hearing. With no event at all, the modelling is an empty grid. And if someone fills that grid by writing that a named club faces a high risk of a points deduction, that is not governance analysis — it is defamation. An important distinction applies. Governance risk comes in two kinds: club risk and pipeline risk. What this report identifies is the second — a failure in the information supply chain. No club faces any allegation here, because no club is named. The Dressing Room, the Coach, and the New-Manager Bounce At management level the discussion covers owner patience, recruitment quality, structural stability. Dressing-room health is read through leadership structure, manager-player relations and the pace of generational transition. One concept recurs: the new-manager bounce, the short-term improvement in results that often follows a coaching change. Three causes are usually proposed — opponents' unfamiliarity, psychological release among players, and the unpredictability of a new coach's approach. The hard analytical task is separating genuine structural improvement from short-term noise. That requires match samples, opponent strength and process data. Without numbers, only a narrative remains: 'the new coach came in and changed the team'. That narrative may be true, may be false, and is most likely a mixture. A subtler point is the coaching power model. At some clubs the coach is head coach, managing the team only. At others he is manager, with authority over transfers. The two roles are not the same, and assessing a club's stability requires knowing which model is in force. Without a name, the question cannot even be entered. The Media Emotion Cycle: From Emergence to Backlash Sports narratives have a life cycle. First emergence: a new idea or a new face appears. Then acceleration: the story spreads, numbers accumulate, comparisons are drawn. Then climax: everyone repeats the same sentence. Finally backlash: one failure arrives, the narrative inverts, and those who praised loudest criticise loudest. The core danger is that a narrative can move faster than the truth. A strong narrative needs only a handful of examples to support it, and the smaller the sample, the greater the chance it breaks. The rule that follows is simple: the bigger the narrative, the more evidence it requires. In practice the opposite happens — the bigger the story, the less verification. A large narrative generates its own illusion of evidence; readers assume that if enough people are saying it, it must be true. A useful test emerges. Any sporting claim can be interrogated with three questions. First: what sample is this built on? Second: what contrary evidence lies outside that sample? Third: where did the numbers come from, and who verified them? Without answers, the claim is an opinion — and presenting an opinion as analysis is deception. The Flood of AI-Generated Sports Content Sports content is now produced faster than at any point in history. What is required is not only time but a tool that offers its greatest advantage precisely where the danger is greatest: the ability to generate rapid copy filled with numbers, names and confident prose. That tool has a limitation many users do not appreciate: it does not know what it does not know. Faced with a blank, it does not fall silent — it fills. And as it fills, it never uses the one word most needed: 'I don't know'. The most honest solution is procedural. The pipeline needs a layer that checks the number of information points, verifies the presence of entities, confirms whether time sensitivity has been assessed — and halts the analysis if the conditions are unmet. Call it a null-safe framework: one that stops with 'insufficient information' rather than inventing. The incident at the centre of this article is, in fact, a successful test of that framework. All nine dimensions returned zero, and not one was fabricated. It is less satisfying, but it is true. And in the long run the true is what holds. Had the piece gone the other way — had someone inserted player names into the blank spaces, invented metric values, attached a headline warning of collapse at a named club — it would have been read far more widely that day. But once exposed, the damage cannot be undone, because readers stop believing anything. Why a Null Result Has Value Sports analysis inherits a basic principle from journalism: the value of a claim depends on its verifiability. A whole football article can be built on verifiable claims. An unverifiable claim supports only a deception. A null result is not a failed analysis but a successful check, and it has three specific values. First, it diagnoses the pipeline. It identifies where the failure sits — source, extraction, or transmission. Without that information the same failure recurs, each time in new clothing. Second, it reveals the shape of the fabrication risk. The risk is not bad analysis but a complete framework that resembles analysis. Knowing that changes the countermeasure: verification of content is not enough; verification of the framework is required. Third, it creates a standard. A simple threshold can be set for future stage-one inputs: at least three information points and at least one named entity. Below that, stage two does not begin. Together those three values produce a decision: a null result may be published, but only on one condition — as a diagnostic, not as a finding. The disclaimer that the piece is not an assessment of any club, player or match must travel with it, because without the disclaimer readers will take it for analysis. What a Proper Source Should Contain The remedy for a null result is editorial, not technical. A minimum list can be set out. First, headline, outlet, author, and date and time of publication. Without these, source quality cannot be verified, and without verification the confidence ceiling for any analysis is low. Second, at least three concrete information points containing specific facts or figures. Third, named entities: clubs, players, coaches, competitions. Losing proper nouns is the single most damaging extraction defect in football analysis, because without entities nothing can be executed — tactics, personnel, finance, governance. Fourth, every quantitative anchor present in the source: fees, contract length, wages, standings, expected-goals figures, pressing metrics, dates. Fifth, the author's stance and purpose, or permission to infer them from the body text. Five conditions, easily stated and most often violated — because the strongest pressure is always to publish today, and that is where the largest errors are made. Conclusion: The Necessity of Saying 'I Don't Know' The most important sentence in this report is probably the least attractive: no conclusion is offered about any club, any player, any coach or any match, because there was no information with which to reach one. That is not a weakness. The weakness is reaching a conclusion anyway. Football is among the most inference-dependent industries in the world — thousands of predictions are generated before every match and exactly one is realised. The enemy of good inference is not uncertainty; it is the pretence of certainty. 'I don't know' is the most expensive and the most necessary sentence in sports analysis. An organisation that can say it may produce less. But its readers' trust endures. An organisation that cannot say it produces reports that all look credible — until one is caught. Then they all are. The incident at the centre of this piece is small but emblematic: an empty file, an empty framework, and a decision that incompleteness is better than lying. Sports journalism's entire value rests on one plain principle — report what is there, and do not invent what is not. In the age of generative tools that can produce fast, fluent and entirely confident falsehood, that principle has become both an ethic and a competitive strategy. The answer to it is one thing only: a verification framework, disciplined naming, and the courage to stop. If the reader takes one thing from this piece, it should be this: looking complete and being true are unrelated properties; and the thing that looks incomplete is often the most honest of all.