cricket_asia: The Ball-by-Ball Data Exists. The Structure to Read It Doesn't.

At 11:40 p.m. last week I opened a file on my laptop. The name was stage_1_extraction.json. Inside there were twelve fields. Eleven were empty. The twelfth contained two words: cricket_asia.

No title. No source. No publication date. The article type was unclassified. The summary was blank. Author stance was not recorded. The purpose line was empty. The list of information points contained not a single point. And directly beneath it, an instruction was printed: "Identify entities from the information points above."

There were no information points above.

I opened the Kochi notebook and found the match had already drawn its own arrows — passing lanes, pressing triggers, the empty space on the left flank. In that notebook every arrow had a target. In this file the arrows were present and the target was missing.

That emptiness did not strike me as strange. It struck me as familiar. It is precisely the shape that Asia's cricket data architecture takes every day — plenty of counters, almost no classifiers.

Cricket in Asia is the most densely logged sport on earth. That claim rests on numbers, not sentiment.

Count the domestic structure of the Board of Control for Cricket in India. The Ranji Trophy, the Duleep Trophy, the Syed Mushtaq Ali Trophy, the Vijay Hazare Trophy — and above them the Indian Premier League and the Women's Premier League. Every ball of every match in these competitions passes through a scoring system. The 2026 IPL season ran to 74 matches, and each match produced roughly 240 to 250 deliveries. That is close to eighteen thousand deliveries in a single season. Every delivery carries at least nine recorded variables — bowler, batter, over, ball, runs, wicket, extras, shot direction, field setting.

That scale exists outside Asia too. In Asia its density is different. Asian Cricket Council tournaments, the Pakistan Super League, the Lanka Premier League, the Bangladesh Premier League, the UAE's International League T20, the T20I calendars of Nepal and Oman — all of them keep their own ball-by-ball files.

There is no shortage of data. There is a surplus of data.

The failure is not in counting. The failure is in classification.

A delivery log records what happened on the third ball of the fourteenth over. It does not record what contest that ball belonged to. "Yorker, 142 kilometres per hour, dot ball" is a record. "This ball was bowled because the bowler had seen the batter's back foot open in the previous over, and the fielder was moved from deep midwicket to long-on" — that is a claim.

Records accumulate by themselves. Claims do not. Claims are constructed.

In that JSON file I was face to face with the empty shelf. cricket_asia was not a conclusion. It was a filing decision — a system placed a document in a drawer and reported the drawer's label. What was inside the drawer, nobody wrote down.

Asia's cricket data architecture has three layers, and each has a different owner.

The first layer is scoring. It is official, it sits with the board, and it is publicly available. A scorecard belongs to no one and to everyone.

The second layer is tracking. Ball-tracking systems such as Hawk-Eye log the speed, dip, swing, seam position and bounce height of every delivery. This layer runs on licence. It belongs to no board. It belongs to a contractor.

The third layer is analytics. Databases such as Statsguru, analysis platforms such as CricViz, aggregators such as Sportradar and Stats Perform. This layer looks the richest and is the most incomplete, because here data is sold, not understood.

The fourth layer does not exist at all. That layer is interpretation. What a given ball meant, which field setting forced which batter into which shot, which team changed its strategy in which tournament and why. This layer is not stored on any server. It is stored in a notebook.

Based on my years of watching matches, I can write this: Asia's cricket data problem is not a problem of scarcity. It is a problem of ownership. The data that matters sits locked with someone. The data that is public is partial.

First structural claim: the inequality in Asia's cricket data is vertical, not horizontal.

To see this you have to place two scorecards side by side. On one side, an India–Australia Test, where seam position, release point, bounce height and even the batter's foot position in the crease are recorded. On the other, a match from the ACC Premier Cup, where the scorecard is complete and nothing beneath it exists.

Both matches are official. Both carry T20I or List A status. Both have data. The difference is that one set of data is read and the other is merely stored.

Take Nepal. Nepal received T20I status in 2026 and ODI status in 2026. Since then it has played hundreds of international matches, and the crowds that gather at the ground in Kirtipur are the largest for any associate nation in Asia. The data on those crowds sits with nobody's board. Whoever holds it does not publish it.

This is where vertical inequality sits. Full members have the whole apparatus — tracking, archives, analysis teams. Associates have a scorecard and nothing else. Both are members of the same council. Both are counted in the same rankings system. But when it comes to analysis, one team has ten years of data and the other has ten years of record.

Data and record are not the same thing. A record states what happened. Data states why it happened.

Second structural claim: the data that decides crores is never made public.

I have always read the IPL auction as an engineering problem rather than a sporting contest. Retention arithmetic, purse allocation, and the Impact Player rule that quietly rewrites XI construction — all three are pure data problems.

A franchise decides whether to retain four players or three, and to carry the rest of the purse into the auction. On what basis? On a model that weighs age, injury history, strike rate over the last three seasons against powerplay strike rate, economy in the death overs, and the behaviour of the home pitch.

That model lives in a file. That file is not public.

The consequence is that what the viewer sees — a young batter sold for crores — is not the output of any public analysis. It is the final frame of a private calculation. We see the outcome, not the process. And then we mistake the outcome for the analysis.

This is why, in Asian cricket, data journalism and data science have become two separate professions. The journalist has the scorecard. The scientist has the model. Both are talking about the same player and both hold half the picture.

Third structural claim: the Ranji Trophy is the load-bearing wall nobody photographs.

In December 2026 a Ranji Trophy match was played in Rajkot. I read its scorecard in full. Every ball was logged. Every wicket carried its over. Every batter had a strike rate. It contained no less data than any major international fixture.

No analytical writing was produced on that match.

I have repeated this count many times, and the result is always the same: roughly 130 to 140 Ranji Trophy matches are played in India each year. Every match generates a complete ball-by-ball file. Perhaps ten per cent of those matches receive any structural analysis. The remaining ninety per cent go into an archive and stay there.

This is where the real economy of Indian cricket hides. India's second XI is the hardest team to make in any sport anywhere — I make that claim on the numbers, not on emotion. More than eight hundred players appear in a Ranji season. About twenty get a look.

The data from that filter exists. The interpretation of it does not.

Fourth structural claim: women's cricket is being measured on men's scales, and the scale is wrong.

On 2 November 2026, at the Dr. D. Y. Patil Stadium in Navi Mumbai, India beat South Africa by 52 runs to win their first Women's World Cup. Every match of that tournament was played in India. Every ball was logged. Every match's data sits in an archive.

The Women's Premier League began in 2026 with five teams. Mumbai Indians won the first title, Royal Challengers Bengaluru won in 2026, and Mumbai Indians won again in 2026. Three seasons, three finals, a full data set.

Now ask the structural question that actually needs asking. What are the retention rules? What is the purse? Where does the scheduling window sit — after the men's IPL or before it? How long is the broadcast commitment? These four questions are data questions, and their answers come from documents, not from sentiment.

Until those four answers are publicly stable, no coverage watershed survives two seasons. Measuring women's cricket against men's viewership numbers is like measuring T20 death overs through a Test session structure. The scale is borrowed, so the result is borrowed too.

Fifth structural claim: this pipeline does not fail loudly. It fails silently.

That JSON file was not corrupted. It did not contain wrong data. It contained no data at all. And that is the most dangerous failure mode, because the pressure to fill an empty field is always present.

When a system has no title, no summary, no information points — and is still asked for an analysis — what does it do? It fills in the most probable answer. It guesses. It manufactures a claim that appears to match the available evidence.

This disease is old in cricket writing. I call it unquantified momentum language — "the momentum shifted," "he wanted it more," "the pressure told." These are empty-field sentences. They contain no count, no field setting, no seam position. They are believed because they sound true. They are not true. They are merely incomplete.

An empty field does not lie. It says nothing. And in a system, particularly a machine-assisted one, silence is not tolerated. It gets filled.

One thing needs stating plainly, because I am asked it repeatedly. Large language models will not solve this. They will enlarge it. A model handed an empty list of information points will produce a convincing-looking list. The fix is not a bigger model. The fix is a smaller claim — small enough to be counted.

I counted 1,029 passes until control stopped being a statistic and became a structure. In that count I invented not one pass. I wrote only what could be counted. What could not be counted, I did not write. That is the discipline, and it is the cure for this entire data crisis.

Now the part where I have to go against my own habit.

My whole instinct says the answer is more charting. More notebooks, more clip budgets, more timestamps. That instinct is wrong. Charting is my method, not my argument.

Asia's cricket data problem is not a problem of volume. Eighteen thousand deliveries per season are already being logged. The problem is who owns which of those eighteen thousand deliveries, and who is permitted to write their interpretation. The board has the scorecard. The broadcaster has the tracking. The contractor has the licence. The analyst has the rent.

In this structure the analyst is the weakest link, and the loudest.

The second inversion is this: that empty file did not read to me as a failure of cricket. It read as an accurate mirror of cricket. The system solved what it could solve — the domain. cricket_asia. Everything else sits above the level a machine can reach, because everything else depends on the person who was watching the match and writing in the notebook.

The empty seats did not erase atmosphere; they exposed the pressing lines underneath it. In the same way, empty fields do not erase available information; they simply mark where information ends and inference begins.

And where inference begins, I have to fight my second habit. For every official statement I keep a standing question: who benefits from this version, and what would the numbers look like if it were false? When a board says its data architecture is strong, the question is not how many servers it runs. The question is how many matches have had their interpretation publicly written.

In February and March 2026, India and Sri Lanka will co-host the ICC Men's T20 World Cup. It will be the single largest block of data in Asian cricket's history — three countries' audiences, three countries' pitches, two host boards, one tournament.

In April 2026 I will not ask how many sixes were hit. That count will be done. I will ask how many of that tournament's deliveries had their interpretation written, who wrote it, and whether that interpretation could reach the data locked on a contractor's server.

If the answer is that the interpretive share stayed where it is today — a handful of people, a handful of matches — then Asian cricket will have accumulated the world's largest data block and never learned to read it.

Eighteen thousand deliveries a season. And one drawer, labelled: cricket_asia.