Asian Cricket's Empty Database: Where Every Match Isn't Logged

In an eleven-person office in Chennai, in the summer of 2026, a CSV file sat open on my laptop. Sixty matches. Fourteen thousand four hundred deliveries. Four columns against every ball — over, batter, bowler, and death-overs exposure, a metric that weighs required run rate against wickets in hand. That file did not have a single empty cell. Six months later I tried to run the same template over a domestic tournament. More than three hundred cells stayed blank, because nobody had ever logged those matches ball by ball. Asian cricket's biggest data crisis is not on the field; it is in the cell nobody felt the need to fill.

Something strange landed on my desk in recent weeks. An analytics pipeline had been run on an Asian cricket subject. The output carried N/A everywhere. Title N/A, source N/A, information points empty, entity list empty, and the only surviving mark was a category tag: cricket_asia. My junior asked whether it could be populated. The answer was no — populating it would be the dangerous move. I remembered the evening a franchise analyst phoned to prove my death-overs model wrong. He was calling because he had the same data, had reached the same conclusion, and had not liked it. An empty database is often less dangerous than a wrong one, because at least an empty database does not lie.

Data does not grow wild. Every delivery that survives in a record was entered by somebody — a scorer, a feed operator, a coder, or a machine somebody trained. The question is not whether the technology exists. The technology is cheap now. The question is who pays the bill for logging that ball.

In the IPL the bill is easy to pay. Every match is a product, so every match carries thirty to forty cameras, four or five data operators, and a broadcaster's graphics team sitting on the live score feed. In 2026 I logged all fourteen thousand four hundred deliveries of the sixty-match IPL season myself. Sixty matches, one feed, and a growing list of doubts. My largest doubt that year was that my data covered one tournament while I was speaking about the whole sport.

Look at Asian cricket through coverage and it splits into three visible layers. The top layer is where every ball generates at least four metrics — the IPL, the Asia Cup, World Cups, and now the Women's Premier League. The middle layer is where the full scorecard exists but ball-by-ball context does not: the Ranji Trophy, the Dhaka Premier League, the Lanka Premier League, the domestic one-day tournaments. The bottom layer is where matches are played, appear in newspapers, and never enter a database — Plate group fixtures, associate-nation domestic finals, and most of women's domestic cricket.

The numbers clarify the gap. An IPL season carries sixty-odd matches and nearly every one of them is fully logged — coverage close to a hundred per cent. A Ranji season carries more than a hundred matches, and the publicly available data on them usually stops at the scorecard: runs, wickets, overs. There is no public route to knowing which batter scored how quickly against which bowler in which phase. That knowledge may sit in a private file inside a club. Publicly, it does not exist.

The gap sharpens in women's cricket. The Women's Premier League runs twenty-plus matches a season and receives roughly the feed standard of the men's IPL. The senior women's domestic tournament plays more than a hundred matches, and a large share of them falls into the category where the scorecard survives and the delivery-level record does not. The consequence is that when a young woman's selection is debated, the argument rarely rests on what she built across domestic cricket — it rests on the two or three games somebody watched in the ground or on television.

The gap is not uniform inside Asia either. India, Pakistan, Sri Lanka and Bangladesh each run at least one fully operational domestic structure, even if its data is partial. Associate members sometimes do not. A four-day domestic final gets played in an associate nation, one newspaper carries a report, and three months later the scorecard search returns a dead link. It looks like a technical problem. It is a memory problem: a competition nobody paid to remember does not stay in history.

All of this rests on one plain economic fact. Logging a first-class match ball by ball takes four days, a laptop and one trained person. Logging an IPL match to the same standard takes four hours, because the infrastructure already stands and an advertising-funded product is running it. The difference is incentive, not capability. The deliveries of a match that returns nothing to anybody are nobody's priority.

This is where the real issue begins. Selection decisions are made where the data is thinnest. India's second XI is the hardest team to make in world sport — I have written that sentence hundreds of times and it has held every time. But the evidence used to pick that team runs far shallower than the evidence available for players already inside it. Once a player is in, every innings, every delivery, every injury gets recorded. The decision happens before that, and it happens in the dark.

Take Sarfaraz Khan. He debuted in Tests at Rajkot in February 2026, sixty-two in the first innings and a half-century in the second. The story is usually told as a tale of selectors' stubbornness. To me it is a data story. His record was built in the one competition where public data stops at the scorecard. That does not mean the record was weak — it means nobody held the instruments required to test or defend it. When instruments are missing, arguments decay into opinions, and opinions that have decayed into arguments deliver justice to no one.

Something else shifted over the last decade that few people credit as a cause. The work zonal selectors used to do — travelling to regional matches, watching players through weather and pitch — is now done by franchise scouting departments, and done better. The difference is who they do it for. A franchise wants a player who fits one specific twenty-over role immediately. A national team wants a player who can hold a fifth day of a Test three years from now. Both needs are real. The trouble is that the data serving the first need and the data serving the second need are not of the same quality, and the institutions that select often read first-need data and decide second-need questions.

The Impact Player rule has added another layer to the distortion. Trialed in the Syed Mushtaq Ali Trophy and permanent in the IPL from 2026, it splits every player's role into two halves — a half that fields and a half that arrives only to bat or only to bowl. The effect is that an all-rounder's numbers never form in the shape Test cricket demands of him. When the rule changes, opinions built on old data go stale — but at the selection table a stale opinion is still placed like a fresh one. This is not a moral issue, it is a comparison error. And comparison errors become institutional because no single person can be blamed for them.

Now the part that made me write this. After the 2026 IPL final — Mumbai Indians beat Rising Pune Supergiant by one run in Hyderabad — I published a piece arguing the one-run margin had been set forty overs earlier. The figure was this: the sixth and seventh batters of that playoff side absorbed thirty-eight per cent of its death-overs deliveries, while its top three faced the fewest pressure balls of any qualifier's top three. Nine hundred thousand people read it, the franchise's analyst called to argue, and then the franchise hired me. The database started empty; the first real finding arrived as a fight.

The lesson in that result was not the number. It was that the team had given its most experienced batters the least pressured deliveries and its least experienced batters the most. Calling that a team error is easy. Calling it a structural outcome is accurate: the top three batted in the powerplay, where risk was low and runs came easily, and numbers six and seven absorbed the stretch where every ball concealed a wicket. Nobody in that setup thought wrongly. They simply had no instrument telling them the load was unbalanced.

That method does not transfer to domestic cricket, and this is the most important point in this piece. In the IPL, four pieces of context travelled with every delivery: wickets in hand, required rate, which batter stood at which position, and which phase the ball belonged to. The Ranji scorecard will not give you most of that in clean form. You will know who scored how many and who took how many. That information is good. It is not enough. A batter who arrives at eight behind the tail and makes thirty an innings and a batter who makes the same thirty from the top four look nearly identical in a scorecard, and are not identical to a Test team.

I carry one other experience into this. Working with live xG in football analytics taught me that you watch the second screen and the room — the number in front of you, and the people who are or are not trusting it. That habit does not translate directly into cricket, because cricket's uncertainty punishes where the football model rewards: low-scoring states invert the logic. So I changed the vocabulary — phase run rate, wicket probability, NRR scenarios, DLS-adjusted targets, in place of xG. The habit survived: read the number with its context attached.

One clarification matters here, because this is where people slip. The absence of data is not evidence of the absence of talent. If not a single ball of a Plate group match was logged, that does not mean no good player stood in it. It means we do not hold the tool to find him. The distinction is enormous, and it decides how the coverage crisis gets framed. Those who say domestic cricket lacks talent are usually saying they could not find a list of it. A model is only as honest as the questions it refuses — and an analyst only as honest as the questions she leaves open by saying 'I don't know.'

The last stage of this chain is broadcast, and that is where the oldest habit survives. Attacking the commentary box is easy. I have done it. In that Chennai startup I learned that empty tables make loud enemies — when you hold a claim and nobody is sitting in the room, your voice rises and your argument weakens. What I learned later is that the box's real shortage is almost always tools, not habits. In live commentary you have one ball's worth of time, no feed, and a producer in your ear telling you two seconds earlier to get to the break.

So the graphic that worked thirty years ago remains the cheapest one. 'His fiftieth hundred' needs an intern, an archive and two seconds. 'His strike rate drops against left-arm spin between overs fourteen and twenty' needs a live database, a producer willing to trust it, and a commentator able to place the line on the ball. The first option wins every time, because the first option demands no infrastructure at all. Broadcast still sells milestones because selling a milestone requires nothing to be measured.

Now the strongest argument against me, because not raising it myself would be complacency. Somebody in the room will say the data-shortage line is stale: every league has ball-by-ball feeds today, Hawk-Eye exists, every match is on video. The answer is that volume and coverage are different objects. An IPL season logs more than fourteen thousand deliveries — a very large number inside a very small universe. How many domestic first-class deliveries were bowled across Asia in the same window? Against that count, fourteen thousand is a handful of bullets in an ocean.

The second argument is the eye: 'I have watched him in the ground, I don't need the data.' I respect it, and I know its answer. The eye is a sampler, and its sample is biased. A scout who returns from a Plate match having seen one eighty remembers the eighty and not the nine other innings of that season. That is not a bad scout. That is human memory. Data is not the eye's opponent; it is the eye's assistant. The real question is never eye versus data. The real question is which one the institution listens to when they disagree — and the usual answer is whichever one speaks first.

One more caution, the one I give every team. More data does not automatically produce better selection. Over the last decade the volume of data in Asian cricket has multiplied several times. Has selection quality multiplied with it? The honest answer is that we have no method of measuring it, because we never measure what the players who were not picked could have done. That is counterfactual and always will be. Where the outcome cannot be measured, a sound process is the only honest claim available — and a sound process means naming your coverage gap out loud rather than hiding it behind a statistic.

Part of the reason for writing this is that naming the coverage gap has become expensive. Write that a board's domestic data is not public and you stop being convenient for an official of that board. That is not bravado, it is a cost. And there is a reason to pay it: what you get in return is a question that becomes embarrassing for a board to refuse.

So what would this look like in practice? A plain index, publishable by any data institution or university group in three columns: tournament, matches played, and matches with a publicly available ball-by-ball record. The ratio between the last two is the coverage index. Split it by gender and you have a map that speaks louder than any opinion. If the Ranji index came out near thirty per cent and the women's domestic index near eight, the argument would change subject. It would stop being about who deserves a chance and become about whom we have stopped watching.

The next signal will come from an unexpected place. I want to see which board is first to admit its coverage deficit in public — 'we do not hold data for this many of our domestic matches' — because that day honesty about data stops being a weakness and becomes a position. And if no board does it, the question stands: fifty years from now, when a historian wants to know what happened in Asian domestic cricket in the 2020s, what will be there? An empty cell. And the worst thing about an empty cell is not that it is empty. The worst thing is that somebody will fill it soon, and nobody will ever be able to tell whether what went in was true.