The Blank Ledger: Auditing a Zero in Cricket's Data Supply Chain

At 11:40 pm in a Mumbai flat, the laptop showed twelve fields. All twelve came back empty. Title — not applicable. Source — not applicable. Article type — unclassified. One-sentence summary — blank. Information points — an empty list. I counted them on my fingers: one, two, three, four. In 2026, at Hyderabad, Mumbai Indians defended 129/8 to beat Rising Pune Supergiant by one run, and that night I hand-coded all 120 death-over deliveries of the match into a spreadsheet, with two columns for yorker length and boundary probability. That night the counting meant something. This night the counting was the result.

My first reaction was not professional. My first reaction was: shall I fill them in? The mind immediately offered three or four sentences — a match, a score, an incident that would fit the platform's traffic calendar. That is the exact moment on which cricket's entire data economy rests: the empty cell, and the hurry to fill it. I shut the screen and left the blanks blank, because what sat in front of me was not a cricket story. It was a specimen of how cricket stories get made. In nine years of writing I have traced every borrowed number back to its root. For the first time, there was nothing at the root.

The supply chain: ball-by-ball to fantasy point

No cricket number travels alone. It moves in a chain. A scorer writes ball-by-ball at the ground; a broadcast desk turns it into graphics; an agency puts it into an API; a fantasy app turns it into points; a newspaper turns it into a headline; and finally the reader turns it into an argument. At every hop, some information is dropped and some is added. What gets added is almost always narrative; what gets dropped is almost always condition — which format, which ground, which toss, which stage of the tournament.

So when I see a two-stage pipeline, where stage one decomposes an article into atomic facts and stage two builds analysis on those facts, I read it like a scorer's sheet. Stage one is the scorer. Stage two is the commentator. What happened here is that the scorer handed over a blank sheet and the commentator was expected to analyse the match from eight different angles.

The failure did not feel unfamiliar. In February 2026, when the Bundesliga returned to empty stadiums, I hand-coded 306 matches to measure the home-win rate before and after the crowds left. It fell from roughly 43% to 33%. The essay ran three weeks late because I kept rebuilding the sample. That experience gave me a rule I still keep: write the methodology before the first sentence, and set a data-freeze date on day one. Otherwise the habit of cleaning a dataset will kill the story.

That rule held that night. The data was frozen, and it was zero.

A map of where a cricket number actually goes

Ask the person on a Delhi news desk where the strike rate in the morning headline came from and the answer will arrive in three steps: an agency, then an API, then 'someone sent it'. That is the real problem. In Indian cricket the transaction record is clean: IPL purses, retentions, bids, all public. The process record is nowhere. For any number you cannot say on what date, from what source, who extracted it and who altered it.

When the process is invisible, error is invisible. And when error is invisible, the only remaining option is estimation. Estimation written with confidence reads as fact. That happened to me in 2026, when Spain drew 1-1 with Russia at the World Cup: 25 shots, around 74% possession, roughly 2.3 xG, not one open-play goal, elimination on penalties. I wrote a fourteen-part thread. A Mumbai editor lifted three of my numbers without credit. I kept the spreadsheet and the grudge. Since then I timestamp and archive my own posts so the next theft is traceable.

I write all this because the blank sheet that night was not a technical accident. It was a link in the same chain, one in which verification is nobody's job.

'Insufficient information' is an active sentence

That report had eight analytical dimensions: format, player technique, team landscape, league commercial structure, rules and governance, risk, public narrative, and industry transmission. All eight carried the same line: insufficient information. On first reading that looks like eight failures. In fact it was eight acts of discipline.

The distinction matters for the whole industry: 'no risk found' and 'no subject found' are entirely different sentences. A report saying a team carries no injury risk is making a claim. A report saying the team cannot be identified is not making a claim; it is declining to. Indian cricket argument erases this difference almost every time. When someone says 'the numbers say otherwise' in a selection debate, nine times out of nine those are not numbers. They are someone's opinion wearing a number's clothes.

Take an example. A fast bowler's economy is 8.9. Good or bad? The answer depends on format. In T20, 8.9 is useful — in the powerplay it is good. In ODI cricket today it is on the expensive side. In Test cricket the question is absurd, because economy is not the measure there. Until the format is fixed, a number is a number, not an analysis. This has been the most common error of my career, and I have watched others make it constantly, especially when a short-form league performance is presented as proof of international readiness.

The same applies to a team. India's home-away differential is larger than almost any other cricket nation's. A team assessment without a venue axis is as incomplete as a forecast without a date. That report had no venue, no date, no toss, no dew. Any analysis written on that base would not be cricket. It would be literature of guesswork.

The pressure to fill: a lesson from the Kochi auction room

In December 2026 I sat in the fourth chair at a startup's table at the Kochi mini-auction, pricing players against the incoming Impact Player rule. My board had an uncapped leg-spinner at ₹5.1 crore. The room took him at ₹9.2 crore. The gap was not technical. The gap was narrative. Buyers do not purchase value; they purchase a story — 'this boy wins us a match next summer' — and the story is paid for in the price.

The same rule governs numbers. Nobody buys a blank cell. Not the editor, not the platform, not the reader. So the system fills blanks with narrative, and because the reader cannot verify, the filled cell survives. What does not survive is 'I don't know'.

I felt that pressure on my own skin in Ahmedabad in November 2026. On 19 November 2026 India lost the World Cup final to Australia by six wickets, and my editor wanted an elegy. I did not send one. I sent a structural post-mortem — the final was lost to a design, not to an evening. It cost me that column and won me a franchise retainer. Since then I publish ranges with stated failure conditions instead of point predictions.

The blank sheet is the last step of that discipline. There, even a range is unavailable, because a range needs at least one observation.

Error that spreads through a batch: recounting 306 matches

One warning in that record stung most: if the failure is in the parser, sibling records in the same batch may be quietly degraded too. This happens daily in cricket analytics and nobody notices.

In 2026, when I coded all 60 IPL matches in the UAE for toss and dew effects, the first pass looked clean. On the second pass, one column of dates was shifted by a day. That single shift put three matches into the wrong night and changed the entire dew conclusion. I found it because the spreadsheet said chasing was easier, while my memory said that in at least two matches it was not. Memory won. The dataset lost. I spent two nights on that one column.

Which is why I never treat an empty record as an isolated event. An empty record signals that something upstream broke — encoding, a paywall, a JavaScript-rendered page, a wrong URL. In cricket that translates directly: if one column of a season's data is empty, every other column in that season is suspect until it has been recounted. Trust is not a unit. It is a quality check repeated on every row.

Three things that should be public

If I could demand three changes to Indian cricket's data architecture, they would be these.

First, a source and a date attached to every published number. This sounds trivial and is not. The day headlines carry 'source: agency, 14 March', half the borrowed numbers will disappear.

Second, count what the women's game was promised. After the 2026 home World Cup, the WPL's broadcast commitments are a measurable contract — minutes, matches free-to-air, scheduling slots. The post-2026 analogy only holds if it holds in the numbers rather than in the press release. I keep a separate ledger on this and it is still incomplete, because slot data is not cleanly public. That is the irony: the thing most in need of measurement has the thinnest record.

Third, treat selection as a data problem, not a talent problem. The hardest team in world sport is India's second XI, and the shortage is not of talent. Put Ranji returns, India A exposure and IPL workload in one frame, then ask how often domestic performers actually get capped. The ratio is small and it is stable year after year. When a ratio stays flat across years, the constraint is decision structure, not depth. When I built that ratio into my own board sheet, several seasons showed almost no relationship between domestic average and the probability of selection. That is the zero that deserves to be public — far more than the blank sheet that night.

The contrarian angle: the fault is not the parser, it is our habit

The convenient answer is that the pipeline broke and must be fixed. I had already written that answer — pre-register the boring answer, publish the inversion only if the boring answer fails the data. Then I asked the data, and the boring answer failed.

Fixing the pipeline does not fix the problem. The problem is that our editorial habit treats a blank cell as an emergency. A system that cannot tolerate a blank will manufacture a number. The vulnerability is cultural, not technical: Indian cricket's data culture rewards the confidently delivered figure and punishes 'I don't know'.

The second point is even less heard. It is true that we lack facts. It is also true that we have a great deal of archive — every highlight, every spell, every press conference. The archive grows and verification does not. That is why an empty record feels so strange: it points at the one place where the chain refused to work. Everywhere else it worked, and nobody checked.

And a disclosure, because nine years of proximity to franchises and boards softens audits. I keep a private ledger of claims I cannot make while holding that access. The ledger is longer than this essay. That too is a form of the same failure — not missing data, but an excess of interests attached to the data we have.

What to watch next

The next time a selection debate, an auction review or a 'what the numbers say' graphic lands in your feed, ask the question nobody asks: what was its stage one? On what date, from what source, were three facts extracted, and who verified them? If there is no answer, what you are reading is not analysis. It is a hurry to fill a blank.

That record will not reopen until the source is repaired and returns at least three information points and one named entity. The day it does, all eight dimensions become writable. Until then only one honest sentence remains, and it is the one I wrote that night: insufficient information. The fields I did not fill in taught me more than the ones I did.