The Discipline of Zero Information Points: The Rule of Saying 'I Don't Know' in Football's Data Audit Chain

Twenty-seven years of watching screens, and some screens still stop me. Last night the first stage of the two-step analysis pipeline returned a table where every cell read "N/A" — no title, no source, zero information points. The junior colleague beside me said, "There's nothing here." He was right. That is precisely where the work begins. The most dangerous moment in football analysis is not when the data arrives wrong. It is when the data does not arrive at all, and the keyboard starts filling the blank cells by itself. In 2026 in Chattogram I built an xG and PPDA model for Abahani Limited Dhaka versus Sheikh Russel KC; it predicted a 1-1 draw and the match finished 1-1. The data arrived that day, so the courage was legitimate. Today it did not arrive. A two-stage pipeline is simple in structure. Stage one fractures the raw text or raw match feed into information points — the atoms of fact: which team, which player, which minute, which number, which source, which date. Stage two joins those atoms into nine dimensions of analysis: tactics, finance, results, league position, governance, management, risk, media narrative, industry transmission. One condition governs everything: every conclusion must stand on at least one information point. Zero information points means zero conclusions. Failure arrives in two forms that look identical from outside. Either the source document was empty — a title sent with nothing inside — or the source was fine and the extractor broke: parsing bugs, encoding errors, tokenisers choking on mixed Bengali and English text. The treatments are completely different. One needs a new source; the other needs a new extractor. Downstream, both surface as the same "N/A". Analysts make their biggest error here: they assume the source is missing, or they assume their own machinery is flawless. Neither assumption is auditable. The right question is small — did the raw input ever enter the pipeline, how many characters, which encoding, who sent it, when? Without those four answers, "there is no data" and "I did not look for data" are the same sentence. In Bangladeshi football data this happens almost daily. No club sends us an optimised event stream. A photocopied sheet arrives after the match, handwritten, sometimes Bengali, sometimes English, often both. Building information points from that sheet means re-verifying spellings, numbers, player identities every single time. During pre-season foreign tours the sheets become rarer still — clubs are busy on commercial trips, and the data disappears under the convenient umbrella called load management. Nobody records who ran how far in which minute; and if they do, they do not publish it. In 2026 at Port City Data I tried to close exactly this gap. After the Abahani–Sheikh Russel match I forced every reporter to file a mandatory post-match data sheet — shots, xG, PPDA, distance. Editors trusted the numbers, and that was the real achievement. There was a price: the copy locked into a single mould, every report opening with the same table. When a template becomes a rule, it stops being a tool and becomes a lock. My own rule requires three conditions before analysis begins: at least one named entity — club, player, coach or competition; at least three independent information points, at least one of them numerical; and at least one time anchor — match date or season. If those three fail, I do not write analysis. I write "insufficient information". Why three and not two or five? It is a tradeoff. One or two points produces a ten-storey building on a single brick. Demanding five or more pushes everyday small stories permanently outside analysis and loses readers. Three is the point where at least two facts can cross-check each other. The sensitivity range is three to five: hard prohibition below three, optional strictness above five. I state that range inside the piece, because a threshold kept secret stops being a threshold and becomes power. Nine dimensions of analysis are not nine essays. They are nine questions. Is there tactical data? Financial arithmetic? A results sample? A regulatory precedent? When every answer is "no data", the honest output is nine "no data" lines — not nine invented paragraphs. An analyst who produces nine pages from zero input has simply done the same work nine times: guessing. The zero-point report tests six gates. Does the tactical claim have data behind it; the financial claim an arithmetic; the results claim a sample; the governance claim a rule or precedent; the public-opinion claim a source tier; and above all, does the analysis have a named subject. A "no" at any gate turns the report from analysis into diagnostic. The fourth gate in practice: a claim that a club faces sanction survives only with the specific rule article, the calculation of the breach, and a precedent of similar treatment. Miss one and the piece is rumour, not judgement. Every football number lives at three levels. Level one is the raw event — who touched the ball, from where. Level two is the model output — that shot's xG. Level three is interpretation. A fracture at level one fakes levels two and three. That is why I keep a standing caveat about xG: it is modelled, not measured. Two providers can assign 0.08 and 0.12 to the same shot. In one match the gap is nothing; accumulated across a season, two hundredths can move a table. So when someone says "our xG is higher, we are the better team", the question must be: which model, which version, which sample. Sample size is equally unforgiving. Judging pressing metrics on three matches is reading climate from weather. My own rule requires six to ten matches, including at least two opponents of different quality, before making a PPDA or high-intensity claim. Otherwise the number is a snapshot, not a trend. The most useful application of blockchain in football data is not fan tokens or club coins. It is an immutable provenance ledger recording the birth of every information point: who sent the source, when, in which version, processed by which model, approved by whom. Each entry carries the hash of the previous one, so nobody can quietly edit a number later. Imagine the 2026 Abahani–Sheikh Russel sheet living in such a ledger: fourteen shots, Abahani 2.3 xG, Sheikh Russel 1.7, PPDA 8.7 versus 11.2 — with the tracker, the minute and the video noted beside each. If someone claimed two years later that xG was 3.1, the hash would settle it. Today the only rebuttal is my memory, and memory is not an audit standard. The ledger's second benefit is proof of absence. If "zero information points" is written into the ledger, nobody can later claim the analyst was lazy. The ledger says: at this time, from this source, through this pipeline, zero points arrived. Recording failure means admitting failure, and an admitted failure is the only kind that stops repeating. My biggest lesson on the 2026 World Cup live dashboard was latency. In the Croatia–England semi-final the dashboard showed Croatia 1.4 xG, England 0.8. Luka Modrić had covered 12.8 kilometres, completed 67 passes, and his late pressing dragged England's PPDA down to 12.9. Croatia won 2-1. The number nobody wrote that night was time: twenty to forty seconds between dashboard updates. In that gap a goal can happen, a card can come out, a substitution can be made. The dashboard shows the match a few seconds ago, not the match now. Reading live xG means holding that gap in mind. That experience produced my fifteen-minute post-match template: xG refreshed every fifteen minutes, timestamp alongside, source below, and one line at the end stating what the number cannot say. Start with the xG, but end with the cold Tuesday — the day the table holds no numbers, only a photocopied sheet, and a decision is still required. One confusion needs clearing. A 0-0 with 0.4 and 0.3 xG is not zero information; it is strong information. It says neither side created, the game jammed in midfield, perhaps both coaches were content. Zero information is the empty cell — the number itself absent. Small number versus no number: the first speaks about the match, the second about the pipeline. Confuse them and you repair the wrong thing — you rewrite the match interpretation when the extractor was what needed fixing. The dashboard is not the match; and without the dashboard there is no match either. The dashboard is a translation, and every translation carries the translator's fingerprint. The null report follows a fixed frame: one line of decision, then what was found, why it was not found, and what happens next, closing with an explicit sentence that this is a pipeline diagnostic, not a football conclusion. I bind that frame into a template because in a crisis people forget structure. A template's job is not politeness; it is memory. The re-run checklist stays short or nobody follows it: locate the source document and check its length, identify encoding and language, run the extractor separately on the same document, compare the information-point count against the previous run, log the result with a date. Until those five steps are done, stage two must not run — and that prohibition has to be written down, because spoken rules are forgotten. Now the strongest argument against this whole method. "Insufficient information, analysis not possible" is true and also comfortable, and comfortable things become habits. Eventually the analyst stops hunting sources and only writes null reports; the method turns from a knowledge engine into an excuse engine. Two defences: attach a re-run deadline to every null report, and track the null-report count as a metric. If one source keeps returning zero points, the problem is the source, not the analyst — change the source, not the template. The second danger is subtler. Fabricated analysis often passes editorial review because it reads well; fluent copy with no visible weakness slips through, while the rough-looking null report attracts suspicion. The system rewards the smooth guess and interrogates the rough honesty. The only fix is the provenance ledger, with a source hash beside every claim. One more thing data believers rarely say. Distance and high-intensity sprints get sold as proof of effort. Modrić's 12.8 kilometres is a lovely number, but it does not say which direction he ran — chasing the ball, chasing opponents, or merely performing chase. Pointless running also produces pretty numbers. Measuring effort means pairing distance with pressing triggers, recoveries and opponent progression; otherwise the number is a poem of praise, not analysis. The pipeline gate cannot be installed mid-tournament. A gate not fitted before the match will not be fitted after it, especially when everyone is drunk on results and social pressure demands every blank cell be filled. Today's zero-point report is really a test, and the test is not about football but about method. The question to leave open: can your source prove that what it sent is what you read? If it cannot, then however elegant next match's xG looks, it is not yours.