The Empty Template: What Indian Cricket's Data Returns, and What It Silently Withholds

On a Tuesday afternoon in early August 2026, a JSON file opened on my laptop with twelve fields in it. All twelve were empty. Title: N/A. Source: N/A. Type: Unclassified. Core viewpoints blank, list of information points blank. Exactly one field carried a value — Domain Label: cricket_world. The system knew it was looking at cricket. It knew nothing else.

Had this been the first time, I would have called it a bug and moved on. I have been pulling ball-by-ball data for eight years, and this is the fourth time a feed has returned structure without substance. The first was in 2026, when a series page rendered through JavaScript and my scraper captured an empty scorecard table. The second was in 2026, when a paywall returned an empty body block and my script read 'zero deliveries' as fact. The third was last year, when a state association's website printed the scorecard but left the bowling overs blank.

All three times the error was mine. All three times the script stayed silent. No exception, no red flag, just a well-formed emptiness. That silent failure is the most honest portrait of Indian cricket's data system: missing information is never logged as an error, it is always logged as a default.

Cricket data lives in three layers, and the three have different owners. The first layer is the scorecard — runs, wickets, overs, catches. Public, free, and enough to compute a team average. The second layer is the delivery log — which ball landed where, who hit it, who missed. Half-public: it exists inside a broadcaster's commentary feed but is not lodged in any open registry. The third layer is the origin data — ball tracking, stump mic, match referee reports, national academy workload registers, physio notes. Entirely private.

My work sits on the second layer, and that is its structural weakness. I can only measure what a broadcaster decided to log that day. Matches without broadcast, stadiums where the commentary feed crashed, associations that never submitted a scorecard — those matches do not exist in my database. Not as unknowns. As absences. And absences are never counted.

The Ranji Trophy has been played since 2026, well over four thousand matches. There is no open delivery-level archive for any of it. The direct consequence is that no independent workload record for India's domestic fast bowlers can be built. When a 24-year-old quick breaks down, we cannot say how many deliveries he sent down in the previous eighteen months. We can say how many matches he played. Matches are an administrative count. Deliveries are a workload.

Football solved this, which makes the comparison useful. In Europe's top leagues, event-level data — every touch, every pressure, every pass — is commercially available, and any independent researcher can buy it and check the work. Cricket's equivalent is fragmented per series by broadcaster, and in India the fragmentation is worse: internationals in one place, domestic cricket nowhere, women's cricket barely anywhere.

My own pipeline runs in four stages: extract, clean, tag, analyse. That Tuesday, all four ran and all four reported green. A pipeline that calls an empty dataset a success is functionally identical to an institution that stamps 'satisfactory' on a report it never received.

I pulled every post-pandemic death over. The spreadsheet started confessing — but only after I forced it to admit how many balls it was counting. Death overs are four overs. Twenty-four deliveries. A season-long 'death specialist' tag is routinely assembled from three good overs, one evening of reverse swing and a broadcast highlight reel. When I began publishing the sample size in the first line, half the claims shrank on their own. Then I pulled every bowler's estimate toward the league baseline, on the assumption that until the evidence is heavy, a bowler is average rather than extraordinary. Few names survived that. One of them was a name no auction purse could dismiss as narrative.

I built the Jasprit Bumrah death-overs work during the lockdown, from the 2026 season. The interesting finding was not that his yorker share rose. The interesting thing was that I had no official instrument for counting it — I was relying on a commentary log, which meant the right to decide whether a given ball was a yorker or a low full toss belonged to whoever was on the microphone that night. That is a classification decision. It is not data. It is an opinion wearing a data jacket.

Three hundred-plus transfers. One pattern nobody in the room wanted to publish. In the mega auction I coded every purchase by role, age, nationality and price. The cleanest pattern was not the badness of any single buy; it was a repeatable blind spot in the market — a premium on uncapped seamers, paid without checking their death-over economy in domestic T20. The second pattern was name-based retention: players whose role was undefined stayed because their name sold.

Here is where I have to report the wound in my own dataset. I only looked at players who were bought. What happened over the next two years to the twenty who went unsold was not in my model. That is survivorship bias, the most common disease in auction analysis. Tilak Varma is a beautiful story precisely because the outcome was good — Mumbai Indians bought him for ₹1.70 crore at the 2026 mega auction and he made 397 runs in his debut season. The story is true. The story is not evidence. It becomes evidence only when the eight of the twenty players my model selected who failed are published too.

The autopsy began with a single scorecard line I could not change. My opinion was supposed to dismantle it, column by column; it dismantled my opinion instead. The 2026 Champions Trophy final at The Oval. India lost by 180 runs, bowled out for 158 chasing 338. Every channel that evening ran the same sentence: a bad day. I spent the following month writing a script that parsed the tournament ball by ball. India's top three were running a powerplay strike rate well below Pakistan's, and the middle-overs dot-ball percentage sat close to half. The conclusion was that it was not a bad day. It was an old template, caught in the brightest possible light.

What that analysis taught me connects directly to today's empty template. I could do that work because the tournament's delivery log was public. Had the same match been played where no commentary feed was archived, I would have had 158 and 338 and nothing else. And you cannot test a template with 158. You can only grieve with it.

In women's cricket the gap is sharpest. The WPL has run since 2026 and its broadcast rights carried serious money. The question is not whether the money arrived. The question is whether the data scaffolding arrived with it — the open record that will define the league for the next five years. If a researcher in 2029 wants to ask what the death-overs economy trend was across WPL seasons, will there be an open file, or will they too be reverse-engineering a commentary log? I do not know the answer, and that is the worrying part. A league that does not publish its data is not keeping accounts of its own legacy.

A pattern emerges, and I take it to be the true shape of Indian cricket's data problem. The problem is not scarcity of data. It is asymmetry of data — dense where a broadcaster has commercial interest, empty where a state association holds control. Every ball of international men's cricket is logged somewhere. No ball of the Ranji Trophy is. For a 19-year-old fast bowler that gap is the difference between a career and a footnote, because selection runs on the data that exists, and what does not exist is not read as talent.

Now the part where I have to write against myself.

My first instinct that Tuesday was to blame the institution. The argument was ready: the BCCI does not publish workload data, state associations do not disclose selection inputs, there is no public injury surveillance register. All true, all publishable. But by that evening I had to admit my own pipeline had committed the exact sin I was laying at the boards' door — it found nothing and said everything was fine.

A demand for transparency carries weight only when the person demanding it applies the same standard to his own system. I have added a minimum-validity assertion: if the list of information points is empty, the system throws an error rather than returning a blank template. Applied to Indian cricket, the equivalent rule would be this — a state association that submits no workload log for a season is not a default zero. It is a logged exception, and its name should be published. Non-existence is itself a data point and should be recorded as one.

There is a second trap I fall into repeatedly: turning correlation into mechanism. Toss-winning teams win. Teams using the Impact Player win. Their net run rate leans the same way. All three move together, so naming any one of them as the cause is easy and almost certainly wrong. My rule now is that wherever I suggest a mechanism I attach a test to it, and I name the alternative explanation. The real danger with data is not that people will quote wrong numbers. It is that a right number will prop up a wrong story and end the argument.

So I pre-registered the question for this piece: is Indian cricket's data gap a shortage of resources or a shortage of decisions? The answer did not favour my thesis. The resources exist. The cameras exist. The scoring software exists. What does not exist is a decision — the decision that this information is public infrastructure rather than private advantage. That is a policy choice, not a technical limit. Until it is treated as a policy choice, every season will return the same empty template with a different title on top.

What is the signal for the next round? I do not think the next big step is a new metric. I think it is a public registry that records which data did not arrive. On the day every domestic Indian match leaves behind an open delivery-level file, a 21-year-old quick's career will stop depending on his coach's memory. Until then, every time a scorecard looks complete and is hollow inside, I will keep asking the same question — who removed what is missing, and who benefited?