The File That Came Back Blank: In Indian Cricket, the Missing Data Does the Selecting

Last week a file opened on my screen with every field empty. Title: N/A. Source: N/A. Article type: unclassified. List of information points: zero. Only one thing survived at the bottom — a taxonomy tag reading cricket_asia. No player's name, no match date, no format, no score. After twenty-nine years of watching and counting, I can write this plainly: a blank file is not an accident. It is a structural habit of cricket's information economy, and it sits exactly where teams get picked.

The most honest use of a blank file is to treat it as blank. Whatever gets written to fill the gap is not analysis. It is decoration.

Context: every ball is a recorded event, and still we know very little

Cricket is one of the few sports where every event is numerical by nature. A Test match carries roughly two and a half thousand legal deliveries; a T20 carries 240. Every ball logs runs, wickets, bowler, batter, over, fielding position. On the surface this looks like an analytics desk's dream.

But a ball-by-ball score and a ball-by-ball dataset are different objects. The score tells you what happened. The dataset tells you how — where the ball pitched, which shot the batter chose, where the fielder stood, in which over the bowler's pace dropped. The first is available everywhere. The second is available almost nowhere.

In the summer of 2026 I learned that difference on my own skin. In March I left a Delhi broadcast production house, where I had spent six years cutting stats packages for a Hindi sports channel, and joined a nine-person digital outlet with no archive and no budget. That October, India hosted the FIFA U-17 World Cup across six cities. England beat Spain 5-2 in the Kolkata final — I watched that match from the stands — and Rhian Brewster took the Golden Boot with eight goals. I logged all 52 matches by hand: shot maps, PPDA, build-up sequences. A model card went out within ninety minutes of every final whistle, errors included, correction log attached.

That routine installed a habit that still governs every file I touch: publish the sample definition before the conclusion. It sounds like self-discipline. It is actually defensive. Once the sample definition is already printed, it becomes very hard to hide what the sample left out.

Which is where the extraction layer enters. The software that reads news does not know cricket; it knows HTML. To that software, a five-day Test scorecard and a table-heavy live-score page look the same. A page with few sentences and many tables comes back empty. A report filed in Hindi, Urdu or Tamil gets stuck at language detection. A site built with JavaScript is invisible without a headless browser.

So a blank file does not mean nothing happened in the world. It means what happened in the world was not written in the language our pipeline reads.

Core: what can be seen is what gets bought — the real rule of selection economics

This is the part that has cost me the most arguments.

India's biggest team is its second team — the line has become proverbial. The standard explanation is talent surplus: too many players, too hard to choose. I don't accept it. This is not a question of talent scarcity or abundance. It is a question of selection process and pipeline. And the most powerful variable in that process is data, not talent.

Put two systems side by side. The first is the IPL. Every ball is tracked, every innings of every player produces a public, standardised, machine-readable record, and once a year an auction stamps a price on every player. The auction is a ledger: demand, supply, age, role, form all collapse into a number, and the number is public.

The second system is the Ranji Trophy. Around thirty-eight teams, hundreds of matches, played in every corner of the country, and for most of them only the scorecard is public. No ball-tracking, no pitch maps, no consistent video, no uniform format — some states file in English, some repeat the same typing errors, some credit the bowler for a wicketkeeper's catch.

The real point is this: when one system can be measured and another cannot, the market pays for the first. That is what happened after 2026. What replaced zonal selectors was not merely franchise scouting — it was a data-literate market that the IPL had manufactured itself. The franchise did not read Ranji. It read the IPL, because the IPL was readable.

In February 2026, at the Bengaluru mega auction, I built a wins-above-replacement board valuing forty players against a ₹90 crore purse. The board had one purpose: to show where auction price was running ahead of sporting value. I flagged the recency premium in advance — the market overpaying for the last result on record. Harshal Patel's 32 wickets in 2026 pulled ₹10.75 crore out of RCB. Ishan Kishan topped the room at ₹15.25 crore to Mumbai Indians.

I repeat those two numbers not because they shock. I repeat them because they exist. There is no equivalent board for a Ranji season, because the data that builds a board does not exist at that resolution.

Here is my first real conclusion, and it is not convenient. If domestic performance in India is not recorded by age and by season, then every selection doctrine — zonal or data-driven — is necessarily incomplete. What we call unseen talent is often not unseen talent at all. It is unrecorded performance.

And this is not only about numbers. I have watched matches for years with an old print-desk habit — notebook, pen, one line after every over. What goes into that notebook never surfaces in a scorecard: which bowler was dropping pace in his third spell, which batter was reworking his footwork against spin, which fielder's throw had slowed since last season. In domestic cricket that observational layer is entirely informal, which means unverifiable. And what cannot be verified never reaches the auction table.

Contrarian: a null result is also a finding

16 May 2026. The Bundesliga returned to empty stadiums. At the time, crowd was a valid constant in my model. Across the remaining 81 matches of that season the home-win rate fell from 43 percent to 21 percent, and I cut the home-advantage coefficient from 0.31 goals to 0.09.

That was not my best forecast. It was my most useful error, because it got published.

Cricket coverage is missing exactly this. We remember our correct calls and forget our empty results. Most stats packages cut for a channel in a given year die because they proved nothing — and often those were the most valuable ones.

In my own ledger there has been at least one case where the claim I set out to disprove survived contact. "India's middle order collapses under pressure" gets repeated on television far more often than it gets defined. Falling wickets are not evidence of pressure; falling wickets are falling wickets. Pressure becomes measurable only once you fix the over, the target, the sequence of wickets and the base rate you are comparing against. Without that, the debate is not measuring. It is reciting.

And here I have to admit my own biggest weakness. The ledger method is powerful, so the temptation is to keep applying it to the same window. After 2026 the Impact Player rule entered Indian domestic T20 and changed the arithmetic of batting and bowling contribution itself. Any domestic ledger from before that rule is now as stale as an auction price from before it. The window's expiry date must be printed inside the piece, or the ledger becomes a forecast and the forecast becomes superstition.

People ask me whether domestic cricket's data can be improved. The technical answer is yes, expensively. The professional answer is that the question is wrong. The question is not whether it can be improved but whose account pays for it. Putting cameras on the IPL is investment, because the return shows up in broadcast rights and auction prices. Putting cameras on Ranji is cost, because the return does not land directly in anyone's ledger. Data therefore gets built where the market is, not where the selection happens.

Two things that are not broken

First, taxonomy. The single tag that survived in the blank file read cricket_asia, not Cricket. A small thing that reveals a large one: the classification system and the analysis system are not speaking the same vocabulary. Where the schema language shifts, data drops out silently, without an error message. It is exactly like a Ranji scorecard — the mistake is not visible, something is merely missing.

Second, the writing habit. What I learned two decades ago as a features editor at Mid-Day in Mumbai still applies: what you do not know, you do not write. Journalism calls it integrity. Analytics calls it null handling. They are the same discipline, and both are in highest demand when the pressure to answer is highest.

What to watch next

Three signals over the next two or three seasons.

One: whether domestic scorecards acquire a standard, machine-readable schema — not ball-by-ball, at minimum a consistent player-level format. The day that happens, the whole frame of India's selection debate changes, because the excuse of the word "unseen" expires.

Two: whether auction valuation boards become public. Right now they stay inside franchises. Until they come out, nobody can test the claim that one franchise bought well and another bought expensive.

Three: whether any institution publishes its correction log. That is the hardest signal and the most necessary.

My ledger does not predict outcomes; I price the uncertainty. And the largest uncertainty in front of me right now belongs to no match and no player. It is which dataset we will fight the next great selection argument on — or whether we will sit in front of another blank file and invent a story to fill it.