The Spreadsheet Came Back Empty: Cricket Analytics' Costliest Output Is 'No Risk Found'

It was two in the morning and the file open on my screen had eight columns. Every column had a table, every table had cells, and every cell carried the same sentence: "insufficient information, assessment not possible." The structure was complete — headings, risk flags, confidence levels, a recovery note. Everything was there except the thing that was supposed to be assessed.

My first reaction was not the correct one. My hand moved toward the blank space on its own. The mind was already improvising — a familiar name, a recent series, a scoreline that could be made to fit the frame, and then all eight columns would be full. The report would look finished. Nobody would suspect a thing.

The real lesson of that night had nothing to do with cricket. It was about the moment an analytics system comes back empty-handed and the man sitting downstream decides to fill it in. Cricket serves up that temptation every week — the reporter with an empty notebook whose only available entry is "the mood in the camp is understood to be tense."

And the story doesn't stop there. The filled-in report reaches a decision desk, and the person at that desk reads it as: no risk. An empty dataset quietly converts itself into "nothing to worry about." That is the most expensive failure in cricket analytics, and it is born not in a model but in a human habit.

The cricket analytics apparatus now standing in India — scouting, auction valuation, workload monitoring, broadcast graphics, fantasy scoring — rests on a two-storey foundation. The ground floor does the pulling: scorecards, ball-by-ball logs, injury registers, auction ledgers, contract dates, venue reports. What comes out of that floor is what the terminology calls an "information point" — a discrete, checkable fact. The first floor starts from there, and its contract is explicit: every conclusion must be bound to the facts from the ground floor. No facts, no conclusions.

This time the ground floor came back empty. No title, no source, article type "Unclassified," domain label cricket_world where the standard schema requires Cricket. The information-point list was empty. No core viewpoint, no thesis, no entities, time sensitivity unassessed, source quality unassessed. Eight dimensions out of eight had their route closed.

There were two roads available upstairs. The first: fill it top to bottom from imagination so the template looks complete. The second: open all eight dimensions and state plainly, in each, that assessment is impossible. The second road was taken, and a recovery protocol was appended — title, source, date, at minimum one information point, entity list, format, and a raw-HTML snapshot.

One thing needs to be made explicit, because the whole argument turns on it. The eight dimensions were not arbitrary. One: format and match context. Two: player technique and data. Three: team landscape and rankings. Four: league and commercial ecosystem. Five: rules and governance. Six: risk. Seven: public narrative and expectation. Eight: industry transmission.

There is not one dimension among those eight that can be touched without a format. Test batting is judged on average and endurance; T20 finishing on strike rate, where the death-overs benchmark is generally placed near 180; ODI sits between the two. With the format cell blank, every benchmark knocks on the wrong door. That is precisely why the report refused any match-level conclusion. Cross-format contamination is the most common error in cricket analysis, and on an empty record that error is close to guaranteed.

A scorebook holds two different kinds of empty, and the distinction is the foundation of all analytics. One: the batter scored zero. Two: the batter's name was never entered. The first is a result. The second is an absence. Put both in the same column and your average starts lying to you.

That distinction sits at the centre of this report. "No risks found" and "no data with which to assess risk" are different sentences, and an entire decision changes in the space between them.

I learned that lesson first in 2026. I was seventeen, watching the Champions Trophy final at The Oval, India losing to Pakistan by 180 runs. What followed on every channel that night was a single sentence: a bad day. The sentence bothered me. Over the next month I wrote a script to parse the tournament's ball-by-ball data. India's top three had a powerplay strike rate of 78 against Pakistan's 95, and their dot-ball percentage through the middle overs was 48. The blog got fifty thousand views.

The lesson was not that data performs magic. It was that "a bad day" is a narrative until you ask what that night actually was. In exactly the same way, "no risk found" is a narrative until you ask where the data came from — and, if it didn't come, why not.

cricket_world versus Cricket is not cosmetic. Article type is the router that determines which dimensions get opened in depth. A match report opens dimensions one and two. An auction story opens four. A governance story opens five. With the type sitting as "Unclassified," the router is blind, and the system opens all eight at once — which means it opens none of them properly.

The cricket version of this is straightforward. Blank out the format field on a scorecard and an economy of 8 is excellent in a Test and close to useless in a T20. Same number, inverted meaning. One labelling error can silently drop every field beneath it — and a dropped field never throws an error; it merely looks empty.

The most useful part of that report is not what it assessed. It is what it declined to assess. Those eight blank columns are a map — they mark the places in cricket where public data either does not exist, or exists in quantities too small for any system to recognise.

Take the governance dimension. The report notes that governance stories routinely survive a parser as "empty," because they carry few numbers and few names. That is not a technical coincidence. It is the symptom of what I think of as cricket's institutional silence.

Where is the player injury register? Which file holds the Ranji workload data? In which meeting are selection inputs decided, and why are the minutes never published? On what criteria is an NOC granted? How far did an ACU inquiry travel, and where did it stop? On what basis is the ICC revenue model split? How wide is the subjective space inside umpire's call, and who determined the slow over-rate threshold?

All of it exists. In a file, in a room, on an official's desktop. In public form it has no schema. So when a pipeline attempts to lift a story from that territory, it finds no numbers. Finding no numbers, the system records it as empty.

The consequence is that the most important stories look the least data-rich, and the least data-rich stories are the first to fall out of the system. That is not an analytics error. It is a design error.

There is another layer to this, which I call institutional survivorship bias. The 2026 Cronje affair, the 2026 Pakistan spot-fixing case, the 2026 IPL matter — all three sit in the public record because they were investigated and documents emerged. Matters never investigated have no pipeline record at all. Integrity data carries the same disease as auction data: you only ever see what survived the filter.

One of the report's risk flags says an empty payload is often the symptom of a feed-level or scraper-level defect. In other words the fault may belong not to that item but to the entire fetch window. The report honestly keeps confidence at "Medium," because the sample is one.

I have an older example of this. During the 2026 lockdown I was building a database of every death over. I pulled every death over from the pandemic years and the spreadsheet started confessing. Then one match's ball-by-ball log came back short. My first reaction was that something odd must have happened in that match — rain, perhaps, or a Duckworth-Lewis-Stern intervention. The check showed the log wasn't short. The source's render was incomplete. And four other matches from the same source that week had broken in exactly the same way.

That is when I made a rule that still sits on the first page of every dataset I build: investigate an empty cell through its neighbours, not through the cell.

Now the part least written about. The report recorded, plainly, that it would not invent an article. That can be read as technical honesty. It is larger than that. It is a refusal of the temptation that is most common in data journalism.

Because filling is easy. Drop in a name, drop in a scoreline, drop in a benchmark, and the report starts to look finished. The reader won't notice. The editor won't notice. The man who notices is the one who prices a player on the strength of that report, or decides to rest a fast bowler.

Fabricated data and empty data are equally expensive — the only difference is that fabricated data never announces its own mistake.

The remedy is already in the report, and it is simple. Put a minimum-evidence gate on the first floor: non-null title, non-null source, at least one information point. If one of the three is missing, the payload does not travel; it logs and returns. Keep a raw-HTML snapshot alongside it so the fault can be localised to fetch or to parse. These are two different diseases with two different cures — one fights a network or a paywall, the other fights a DOM structure.

This is not a heavy investment. One validator, one snapshot, three conditions. That is what I consider genuine information gain — not a new fact, a new gate.

Now the part where I argue with my own thesis.

The tidy story is this: the pipeline broke, fix the scraper, move on. It is comfortable, and it is probably incomplete.

First alternative: the item was never a cricket article. An index page, a redirect stub, an ad slot. The report records that possibility at low confidence. If that is the case, fixing the scraper yields nothing, because there was nothing to extract.

Second alternative: extraction did not fail. The source itself published an item containing no numbers and no entities. Cricket does this constantly. Opinion, rumour and "sources say" journalism are metric-free by construction. If that is the case, the problem is not in the pipeline but in a journalism that goes to press without a fact underneath it.

Third alternative, and the most uncomfortable: the report rated its overall risk as High. That risk belongs to the supply chain, not to cricket. The report itself warns that a decision-maker could mistake it for sporting risk. I would push further. A risk label remains incomplete until it states what the risk is about — the sport, or the system measuring the sport.

Fourth alternative, and the least comfortable of all: perhaps the empty payload was the honest payload. A large share of the cricket "analysis" published daily is a filled-in template — a headline, one statistic, one conclusion, and no chain connecting them. An empty file at least admits its emptiness. A filled one never does.

I should also mark my own limit. This entire reading rests on a single observed item. The sample is one, and the report itself calls it medium confidence. I will not go further. Claiming a systemic defect requires the other items in the same fetch window. The line between correlation and causation has to be drawn right here, or I commit the exact error I am criticising.

Three signals will settle what happens next.

The other items in that fetch window. If more than one comes back empty, the matter escalates from item-level to scraper-level and the priority changes.

The "Unclassified" rate. If it is climbing, routing is breaking — and broken routing means every story gets the same eight-dimension mould, whether it is a match report or a revenue model.

Entity-extraction coverage. If names are present in the body text and absent from the ground floor, the three dimensions beneath are running systematically weak — and that weakness will show up not in one story but in a hundred.

The question is not whether a pipeline broke. Pipelines break. The question is how many reports are sitting in cricket's decision rooms right now with "no risks found" printed on the last page — and a spreadsheet behind them that was never opened.