N/A Is Not a Finding: The Report That Looks Complete and Contains Nothing

A report landed on my Delhi desk this week. Eight dimensions, eleven tables, a risk matrix, three scenario projections and an information-value rating — one star out of five, awarded for diagnostic usefulness. Every cell was populated. The problem was that every cell said "N/A". The information-point count was zero. Title: N/A. Source: N/A. Time sensitivity: "not assessed in Stage 1." And in the field meant for entities, the extractor's own instruction was still sitting there — "identify from the information points above."

This was not the scorecard where a batsman was dismissed for a duck. This was the scorecard where the innings never began, and somebody recorded it as zero. Cricket punishes that confusion more than any other sport, because zero and absence look identical on paper and mean opposite things.

I have watched matches for close to two decades — from matching scorebooks to video tape, to ball-tracking dashboards and death-over economy. The most useful thing I learned in that time was not a model. It was a question: where exactly did the data in this cell come from?

Here is how the pipeline is meant to work. Stage 1 extracts the title, source, information points, entities, time sensitivity and source quality from an article. Stage 2 builds eight dimensions on top of those points: format and match, player technique, team landscape, league and commercial ecosystem, rules and governance, risk, public narrative, industry transmission. The atom of the entire system is the information point. The rule is unambiguous — every conclusion must trace back to a numbered point, or it is not a conclusion, it is a guess.

This time the atom never arrived. Zero. And there is a detail worth recording, the one that unsettles me most: the fields were not simply blank. They contained the instructions for filling them. That is not ordinary emptiness. Those are the fingerprints of the man who brought the form and left again — the template stamped itself in place, the values never came. Data lines call this a silent failure, and it is far more dangerous than a noisy one.

To see why, look at the shape of the document. Its risk rating is "High" — but it states plainly that this is not an on-field or commercial risk, it is an information-supply risk. Alongside sits the information-value table: sporting value zero, industry value zero, timeliness zero, and one star under reference value, awarded solely because the document reliably evidences its own defect. For a technical document, that is extraordinary honesty. For its reader, it is a trap: a fully rendered, table-formatted output that the eye cannot, at a glance, distinguish between "nothing was found" and "something was found and it was negative."

I have a lived example of that confusion. In August 2026, when my previous employer cut the live-betting desk by two-thirds, I spent six weeks rebuilding home-advantage coefficients from scratch. Closed-door stadiums and an IPL season entirely relocated to the UAE — first match on 19 September 2026 in Abu Dhabi — meant that home venues barely existed. The biggest danger then was not that my coefficient would be wrong. It was that it would quietly return "home advantage = zero" when the truth was "home advantage = undefined", because the variable that carried the information was no longer present. Every ghost game taught me the same thing: the crowd was a variable. When it left, the coefficient did not become zero — it became missing.

That distinction is what broke in this report.

Economy rate on zero balls is not 0.00 — it is undefined. That one line is the calibration for the whole pipeline. When the sample is zero, every statistic computed on it is not a number but a lie with a valid unit attached. The report filled eight dimensions of cells and put N/A wherever a figure belonged and "cannot be assessed" wherever a judgement belonged. The difference is that in a cricket selection meeting, the same act happens with the opposite sign: N/A gets read as a negative.

Take the auction table. A 22-year-old batsman, eleven IPL innings, thirty-eight balls in the death overs. His death-overs strike rate is blank because the sample is too thin to record. Now watch what the blank does: he acquires the label "not a death-overs finisher", and his price never rises for the role he should have been tested in. Absence of information has silently become deficiency of ability. The Impact Player rule has made that lie cheaper — with a substitute available, teams use fewer bowlers, part-timers' ball counts shrink, and every year we select from less depth than we possess.

Reading absence as a negative is the biggest, least-caught theft in sports analysis, and every N/A row in that report is the polite version of it.

The report flagged its own second defect, and we should pick it up: the domain label was "cricket_asia", which does not match the specified taxonomy of "Cricket". Cricket has an exact twin of that error — losing the format anchor. Test, ODI, T20: without identifying the format, analysis cannot begin. Death overs are the last five of twenty, a fifth of an innings, a manufactured pressure inside a boundary; the Test new ball is different biomechanics, a different field, a different-coloured ball and a different mind. Quoting one format's economy rate in another format's selection room is the same category of mistake as treating a "cricket_asia" tag as match information. What is missing in both cases is the thing that mattered.

The third point stings most, and it is Ahmedabad. 19 November 2026, Narendra Modi Stadium: India bowled out for 240, Australia 241 for 4, Travis Head 137. The ICC scorecard records a six-wicket defeat. I still consider that match defensible on process — the toss, the conditions, the pitch's behaviour, India's middle-overs plateau, none of it pre-determined the result. But the country took a lesson out of that one night, and the lesson was "India chokes under pressure." That is precisely the error sitting inside the report: in place of an empty cell, we inserted a story, and then treated the story as evidence.

I also hold the counter-evidence, and I hold it only because I did not throw it away. In June 2026, three days before the Champions Trophy final, I built a model for The Oval — Pakistan's new-ball economy against India's top three (Rohit Sharma, Shikhar Dhawan, Virat Kohli), and India's middle-overs dependence on wrist spin. The model gave Pakistan 38 per cent; the market had them at 16. Two senior editors at the desk called it contrarian noise. On 18 June 2026 Pakistan made 338 for 4, Fakhar Zaman 114, and after Mohammad Amir's new-ball spell India were bowled out for 158. Pakistan won by 180 runs.

I built the Oval model to win an argument. It won. That was the problem. Being right was never the job. Keeping the email thread with timestamps was the job. I have never called that one night an edge, because a single match does not measure a model's quality. It measures the sample, and the sample was one.

The same discipline taught me that the 4% looked like edge. It was noise wearing a suit. A bowler's death-overs economy improves four per cent across forty balls; the number is real, the direction is real, and the conclusion is entirely airborne. When I went back and matched it against the quality of the batsmen he bowled to, nineteen of those forty balls had been against the tail. The sample was small, the selection was biased, and I was one sentence away from calling it an edge.

Now the sharpest line in the report, which it wrote against itself and which I want to extend. It says it cannot distinguish between "the article contained nothing material" and "the article was never retrieved." I would add a third state: the article contained material, and our classifier was looking for the wrong thing. That third state never gets recorded in data pipelines, and in cricket it is the most common of the three. The information existed, just not in the format we query. This is why a photograph of an old scorebook sometimes tells more truth than a months-old metrics dump.

Which is where I half-disagree with the report's advice. It says: do not file this as a null result, halt publication. Halting is correct. But the relief that follows is false, because the real failure was not in Stage 1 — it was at the gate that let Stage 1's hollow scaffold through. Our quality checks count whether cells are filled. They do not count whether any cell has a sample behind it. Exactly as our selection sheets have every column and not enough balls behind any of them. The newsroom had a narrative. I had a residual. The residual aged better. What the room calls a story, I call the leftover that did not fit the story — and the leftover is usually right by the following season.

The next time a report like this arrives, I will run one test, and it is not technical. I will count how many names and how many numbers the document contains. If either is zero, its status is not "no material change" but "void". The test is cheap, it leaves no stain, and it stops a thousand false conclusions per file.

My bet for the auction season: at least a dozen players will be written off as "not a fit for that role" on fewer than sixty balls. The franchise that publishes its sample size this year will get more trust from me on its bids — not because its model is better, but because it did the thing nobody else is doing: it told us how much it does not know.

The question is yours too. The next time a sheet shows an N/A — a selection sheet, an auction sheet, an analyst's report — ask whether it is an N/A or something that never arrived. The answer will usually be that we mistook an empty space for an opinion, and then bid on the opinion.