The most expensive mistake in an analytics pipeline is not the one that produces a wrong conclusion. The most expensive mistake is the one that mistakes an empty result for a result. In every cricket data report that has crossed my desk — from IPL auction valuation sheets to ball-by-ball Ranji logs — I have kept one rule: a sheet without numbers is evidence of something else, not evidence of cricket.

This piece is a fresh example of that rule. A stage-one data extraction ran, and the output was empty across the board — no headline, no source, article type unclassified, core viewpoints blank, information points an empty list. Only one thing survived: the domain tag cricket_asia. Four words. And no serious analyst can write a news story on four words. Anyone who does is writing fiction, not news.

Context: The pipeline link that breaks quietly

There is a truth in professional sports data that media houses rarely say out loud: the largest share of content production cost goes into the quality of the raw material, not the analysis. However fast an analyst may be, if the original text of the article does not reach him, whatever he writes is not a commentary on that article — it is a commentary on that article's absence.

I have had one habit since 2026. That year, at the M. Chinnaswamy for the Ranji Trophy final, I kept my own ball-by-ball sheet in parallel with the official scorebook. Across four days I found 23 deliveries that never reached the official card — wides, no-balls and a leg-bye. I wrote 900 words on the discrepancy rather than the result. Three weeks later the board reissued the card.

Since that day I have had one rule: when the record shows a discrepancy, the first question is not 'who won' but 'which step quietly disappeared.' What we have here is that same discrepancy — except a software pipeline has replaced the scorebook.

Core: An empty result is not a result, it is a signal

I read the data here in three layers.

First layer — the behaviour of the fields. Title N/A, source N/A, type 'unclassified', all three sub-fields of core viewpoints blank, information points an empty list, and entities/time/source-quality all annotated 'inferred from the information points above' — when there are no information points above. That pattern is itself a diagnostic. If the article had been read but not understood, the title and source would at least be filled and only the viewpoints blank. Everything being blank means the original text never reached the system.

Second layer — the likely cause. Three technical paths produce this kind of empty result. First, failure at the HTTP level — a 403, a 429 or an empty payload. Second, a JavaScript-rendered page, where the scraper receives HTML but the actual content is built later in the browser. Third, a paywall or geo-block, where the server returns 200 but the body contains only a subscription wall. All three share one thing: the system reported success while the job failed. This is the most dangerous state, because the downstream model gets no error code — it gets only empty space, and it often reads empty space not as 'absence of information' but as 'absence of subject matter.'

Third layer — the false reassurance of the domain tag. cricket_asia is a valid tag, but a tag names a subject, not a claim. Within Asian cricket, at least half a dozen entirely different stories are running at any one time: board power politics, domestic structure, television rights bidding, selection committee reasoning, and on-field strategy. None of these can be chosen from the tag. If you select a story from a tag, you are not selecting a story — you are inventing the story you prefer and sticking a domain label on it. The lesson from my archive of 340 one-day matches is exactly this: even with a large sample, a conclusion emerges only when you have done the counting; and here the basis for counting is missing.

Fourth layer — the cost calculation. Suppose an editorial team trusted this empty result and built further work on it. What happens? A writer sits down, sees cricket_asia, writes something, an editor publishes it, and if the original article said something else, readers get wrong information — and worst of all, no one will ever detect the error, because the original source was never entered into the system. This is the same problem cricket journalism had during the match-fixing era with 'sources say' reporting: confidence in place of verification. The only difference is that the source is now a scraper.

Contrarian angle: Sometimes treating an empty result as a result is correct

Here I must distrust my own habit. My entire identity is built on 'anomaly first, explanation later' — and here the anomaly is clear: an entirely empty data sheet. But like every anomaly, it has two readings.

The first reading says: the pipeline failed, fix it. The second reading says: sometimes an empty result is genuinely the correct result — as in a match where no wicket falls, an empty wicket field does not mean the scoring failed; it means no wicket fell. But that logic does not apply here, because the article's existence was never denied — only its text failed to arrive. 'Text not found' and 'text does not exist' are two different things, and conflating them is the most common self-deception in data analysis.

The second thing I want to say clearly: if all someone has is a domain tag, the most honest thing they can do is declare that they do not have the material. Writing 'general background on Asian cricket politics' is possible, but then it must be stated explicitly that it is not based on that article. The reader has a right to know whether the analysis in front of them is an analysis of the original material or of the analyst's memory. That distinction is what separates journalism from propaganda.

The third, subtlest angle: is leaving an empty result empty also a kind of failure? No — if it is named correctly. A clean 'data unavailable' report is always better than a filled but false analysis. The greatest crime in cricket analytics is not a wrong conclusion but false confidence. In the scorebook that was missing 23 deliveries, the most dangerous thing was not the missing deliveries — the most dangerous thing was that the card looked complete.

Takeaway: What to look for next time this happens

If this pipeline runs again, three things must become mandatory. First, every stage-one result must record the HTTP status code and the length of HTML received — those two numbers will immediately show where the fault occurred. Second, 'information points empty' must become a separate flag, telling downstream that analysis is not possible. Third, every analytical report must carry one mandatory line: 'This conclusion is based on which material, and on which it is not.'

My experience says the most reliable signal in cricket is not the one that arrives most loudly, but the one that speaks clearly about its own absence. The most useful line in my 340-match sheet was the one where I wrote: 'data incomplete for this over.' Next time you are handed an empty sheet, do not first ask 'what is in it' — ask 'what is missing, and who allowed it to go missing.'

And until that question is answered, do not write an article on that sheet.

---

The author has kept ball-by-ball logs for 25 years and writes on the institutional structure of Asian cricket.