The Empty Sheet and the Full Scoreboard: An Audit of Silence in Cricket's Data Pipelines

At 1:47 a.m. in Ahmedabad, a ten-field form sat open on a screen. Seven fields read N/A — article title N/A, source N/A, author stance N/A, time sensitivity unassessed. And at the bottom, the field the entire report was supposed to grow out of, held two brackets and nothing between them: information_points: []. One character in the whole scaffold was alive — the domain label cricket_asia. The pipeline downstream opened the deep-analysis door anyway. No warning, no red light, no error message. Only a clean, well-proportioned, empty template. That was the fault: emptiness does not announce itself as emptiness.

The scene was not new to me. In January 2026, in a room in Ahmedabad with a brace strapped around my lumbar spine, I was watching a buffering stream of the Ranji Trophy final from Indore. A stress fracture of the L4 vertebra had ended my Gujarat Under-16 season in October 2026, and that winter the laptop was my ground. Across that season I hand-charted all ten of Gujarat's Ranji matches — roughly 4,180 balls. Dot-ball percentage by phase, scoring zones, and the bowlers Priyank Panchal actually attacked rather than merely survived. Panchal finished the season on 1,310 runs, and Parthiv Patel's Gujarat beat Mumbai for the state's first title. The notebook started as a rehab log. It ended as a title.

Context: standardise before prose

Stage-1 exists to pull facts out of raw copy, attach a source to each fact, and hand the next stage a ledger worth arguing with. The problem is its architecture. The entities field extracts nothing of its own; it is inferred from the information points above it. Source quality measures nothing of its own; it is built from each fact's source field. One field empties, three fields empty with it — and the file still reports as successful.

Cricket runs this architecture every day. A scorecard is a template: runs, balls, fours, sixes, overs, maidens. Nowhere does it record which phase a batter scored in, which bowler he scored against, how old the pitch was, or what state the match was in. Without those columns, a scorecard is a rumour wearing numbers. So I fix definitions before I write — one template, one column set, applied identically across formats. I do not treat a strike rate as a fact. I treat it as a condition. Not an average, an average with phase. Not a workload, a workload with rest days.

That winter a physio, tightening the brace straps, told me the scan looked fine but the body keeps its own timetable. That sentence fits no column in any spreadsheet. I keep it in the margin of every report, protected from the model.

Core: the four fields that lie quietly

Field one is absence. The information points list is empty, which means the thing to be analysed does not exist. In that situation deep analysis can only mean one thing: the writer fills the gap with memory and inference, inventing facts the source never carried. In cricket this is the same sin as printing a number with no match state attached. A spinner's economy is 6.20 — good or bad? Without over-phase and fielding restrictions, that figure says nothing about anybody.

Field two is provenance. Source quality was meant to be built from each fact's source field, and the source field is simply missing. Which means a report saying sources say had its credibility checked by nobody. In my own copy those sentences get cut before an editor has to ask.

Field three is time. Time sensitivity was never assessed, so nobody knows whether what is being written belongs to today or to three seasons ago. Printing last year's numbers under today's dateline is the most common and least caught error in Indian cricket coverage.

Field four is context. The domain label cricket_asia tells you the subject is cricket; it does not tell you which match, which auction, which contract. A label is not an answer to a question.

Now run the same scaffold against real cricket. In November 2026 Gujarat Titans drafted Hardik Pandya, Rashid Khan and Shubman Gill, then ran phase models into the February 2026 mega auction in Bengaluru. David Miller was flagged as mispriced at a ₹3 crore base, because his strike rate against spin had been buried by batting position, not decline. We bought him. He made 481 runs in IPL 2026, and on 29 May 2026 at the Narendra Modi Stadium, Gujarat Titans beat Rajasthan Royals by seven wickets. What survived in that sheet was not a column but a question: was Miller's dip a price problem or a position problem?

My 2026 World Cup xG sheet recorded England scoring 12 goals, 9 of them from set pieces. The sheet said so before the semifinal, because it had a set-piece column. In a template without that column the fact is never recorded at all — and the report says England batted fluently.

March 2026, Rajkot. The last live match I attended before the shutdown was the Ranji final, where Jaydev Unadkat's Saurashtra beat Bengal. The most useful column in my sheet that season was not bowling but rest — the days between two spells. Match reports never print that column, and it decides what the next spell looks like. Two weeks later my Under-23 season was cancelled, a second surgery was scheduled, and at 19 I stopped being a cricketer. From April to August I built a ball-by-ball database of every IPL match from 2026 to 2026, 756 games, and when the 2026 edition moved to empty UAE grounds I used it to test home advantage. The answer was mostly conditions, not crowds. When the season shut down, the database kept playing.

Contrarian: more data does not reduce error

The intuitive conclusion is that empty fields need more data poured into them. My experience runs the other way. The more columns multiply, the faster one empty cell breeds others, and the file looks more complete than it is. Seven fields were blank in the audit above and no alarm sounded, because the template had no alarm column.

Second, noise and silence cost the same. In the transfer market an agent's noise distorts price — that is now accepted. A data desk's silence is more expensive, because nobody suspects it. The commentary box arguing tactics without a ledger is the old complaint; the new one is an analytics desk holding a ledger and no audit.

Third, and sharpest: a model that back-tests well is not a story. A model is a story only when it changes a decision — a retention, a promotion, a spell cap. Nothing in that empty template changes a decision, which makes it not a report but a vacancy.

Limits

This audit is of scaffolding, not of any model's accuracy. The four cricket examples are drawn from my own work — Miller's price, Panchal's season, England's set-piece goals, Unadkat's workload — and all four carry a survivorship bias, because the sheets that turned out wrong are not in this piece. My 2026 World Cup sheet was publicly wrong for six weeks; the point was that the error carried a dateline. And it must be conceded that one empty template does not prove a pipeline always returns empty — here it returned empty once.

Forward signal

The next time a data file comes back with zero facts in it, the correct report is not deep analysis. The correct report is one line with a date on it: parse failed. The day cricket desks learn to call an empty sheet empty, every other number they publish becomes easier to trust. The question is not how much data we hold. The question is where, inside our pipeline, silence is allowed to live.