The N/A Report: The Piece a Cricket Data Desk Refused to Write

The memo from that night is still in my file. Nine sections, fourteen tables, and one word in every cell — N/A. No match name, no player name, no team, no league, no governance dispute. The document that landed on my desk five minutes earlier had not brought any cricket with it. It had brought a skeleton, and all the air inside the skeleton had already leaked out.

The template carried nine sections: format and match analysis, player technique and data, team landscape and rankings, league and commercial ecosystem, rules and governance, risk-side analysis, public narrative, and cricket industry transmission. Each with its own table, its own evidence row, its own confidence level. I scrolled. Under Information Points, zero entries. Article Title blank. Article Source N/A. Core Viewpoints N/A. That is the moment a data desk decides whether it is a newsroom or a sales counter.

I have been writing notes while watching matches for twenty-one years, and one habit has never changed: the first line of any claim states how many balls or how many matches sit behind it. Editors used to trim that line. Later they started asking for it. The note I sent from the Bengaluru auction hall in February 2026 did not open with a name. It opened with every Rashid Khan delivery pulled from the 2026 Under-19 World Cup and Afghanistan's legs at the 2026 World T20, weighted by the quality of the batter facing him, converted into a wicket probability. Base price INR 50 lakh. Sunrisers Hyderabad paid INR 4 crore. He finished with 17 wickets at an economy of 6.62 and took Emerging Player. The memo was four pages; the market only ever read the fee.

The strength of that note was never its conclusion. It was that there were deliveries to count — line, length, batter, match state, over number, all tagged. A model could be built on that, argued against, and the argument could turn into a cheque. This week's template contained no deliveries at all. Nothing gets modelled on zero balls. What gets written on zero balls is intent dressed as inference.

There is a statistical distinction the industry keeps collapsing into one disease. Weak evidence is one thing — three matches, forty deliveries, a wide confidence band, and still a claim you can defend with the word 'likely'. No evidence is another — zero matches, zero deliveries, zero names. In the first case you write 'probably'. In the second, anything you write is invention. Reading an empty cell as a zero is the most expensive error in cricket analysis, because missing data is not zero — it is absent. A bowler's economy of 8.4 is a finding. A bowler's economy that was never recorded is not a finding; it is a failure of the record. Put both in the same cell and the model starts lying from that day forward.

This emptiness did not arrive suddenly. India's domestic data architecture has looked like this for years. Thousands of deliveries are bowled every Ranji Trophy and Vijay Hazare season, and how many of them are tagged is not something a public database will tell you. The tracking systems bolted onto international and franchise cricket do not reach the places where recruitment actually happens. The consequence shows up at every mega auction: names surface that nobody has scouted publicly, and a franchise buys them because it runs a private desk that already has the file. Nobody had tagged him, which is not the same as nobody had seen him. Talent identification has quietly moved off the zonal selector's table and into a spreadsheet whose door stays shut.

That is where this stops being a story about one memo. A blank sheet is not one desk's failure; it is a measurement of the system. When nine sections cannot produce a single verified fact, the meaning is not that cricket did not happen. The meaning is that the cricket that happened was never recorded, or was recorded and never released. Either way the loss is identical: the market will still price that performance, and it will price the story rather than the fact.

Franchise economics runs on exactly this. What is tagged is what sells, and what sells is what gets tagged next year — the loop runs backwards. The 2026 note worked because the delivery count came before the name. Had I walked in with two highlight clips and a scout's chai-time opinion, I would never have written a INR 4 crore recommendation. Recommendations do get written on empty data. They simply stop being recommendations and become preferences, and preferences are never audited.

The second lesson is model age. Since the Impact Player rule arrived in 2026, what a bowling average or a strike rate measures has quietly shifted. A model built on a pre-2026 sample and never re-tested is still running with full confidence, and it is wrong. I date-stamp every model, run the same query on both sides of 2026, and report the discontinuity instead of smoothing it away. This report on empty input comes from the same discipline: where the model does not apply, running it anyway is a victory of habit over data.

Now the part no data desk writes openly. Emptiness has a price in this industry. An N/A memo does not get filed, an unfiled memo earns nothing, and earning nothing means the next call does not come. Against that stands the memo with all nine sections filled — even if the filling rests on one phone call and one tweet — which files instantly, publishes instantly, and gets cited onward. Five explainers went out that summer. Four came back filed. The fifth is why I write.

The inverted conclusion, as I read it, is this: the most valuable output a data desk produces is not the thing it proved but the thing it refused to write. The day a desk fills an empty cell is the day every green light it issues becomes suspect. A reader is entitled to know which claim stands on three hundred deliveries and which stands on three tweets. A desk that will not draw that line is not a desk.

I keep a file of things the data said first and the pundits said later. This week it gained a new entry, and the entry is blank. That sounds strange until you notice that a blank entry is also evidence — it tells you which joint in the pipeline snapped. I know that when this template finally fills, the first three things I will check are how many deliveries are logged, how many overs are tagged, and how many names nobody has publicly touched. The day those answers arrive, this piece becomes a real one.

Until then, the question I append to every model, appended here as well: will this report be assumed full if nothing changes? If the answer is yes, the problem was never the input.