Indian Basketball's Largest Dataset Is the One That Was Never Recorded

Two files. One from 2026, one from 2026.

The first has 1,180 rows in it. Fourteen Chennai Slam games, the team's possessions counted one by one, every pass outcome, every second-chance point, every turnover with its cause — all in a single spreadsheet, because nobody at the arena could tell me how many second-chance points the team had conceded. I filled those rows watching a buffering web stream. A few months after the season ended, the league folded, and its play-by-play died with it.

I opened the second file this month. There is a header row, there are columns — possession, lineup, points, shot zone — and below them, nothing. No match ID, no date, no team, no player. The game whose record had been requested was played. Its scoreline ran in a newspaper. Its possession log was simply never recorded anywhere.

An empty sheet is the normal condition of Indian basketball, not the exception. The data that was never recorded is a bigger story than any scoreline, because scorelines are remembered and possessions are forgotten.

It is worth being precise about what a possession log actually is, because in India the phrase triggers an immediate reflex: that is NBA stuff. In fact it is the cheapest data product the sport has. One scorer, one tablet, five fields: possession number, team, the five players on the floor, the outcome — shot, turnover, foul or offensive rebound — and if it was a shot, the zone. Those five fields yield offensive rating, defensive rating, pace, eFG%, second-chance points and turnover rate. A game contains roughly 150 to 170 possessions across both teams. That is 150-odd rows per game. Across a twenty-game season, three thousand rows.

Three thousand rows are outside the budget of every Indian basketball institution that exists. The sentence sounds like a joke and is entirely true.

When the UBA launched in 2026, it briefly looked as though domestic basketball would finally get a record of itself. There were franchises, television, arenas, foreign coaches. The league ran from 2026 to 2026 and then stopped. It had a broadcast and never an official play-by-play. The surviving footage became the only source — and it became the basis of my sheet. The names that followed — 3BL, EPBL, INBL — were small, disputed and short-lived. None left a database that a researcher could open three years later.

Every vanished league leaves a dataset that remembers its shape — and in India that dataset has been incomplete, scattered and ownerless every single time.

The Basketball Federation of India's own events are no different. The Senior National Championship runs every year with Railways, Services, banks and state units in the draw, and that is the real pyramid of Indian basketball. The scorebook there is made of paper. A box score is produced sometimes, not always. The question of a possession log does not arise. Having watched these tournaments for fifteen years, my conclusion is that the problem is not an absence of information. The information exists. It simply belongs to nobody.

There is a contradiction here that almost nobody notices. FIBA keeps complete live stats for its own competitions. When India's men's or women's team plays an Asia Cup qualifier or any FIBA event, possession-level data for that game sits on FIBA's platform — with lineups, with a shot chart, with plus-minus. The Indian player's international performance is fully coded, while the domestic system that produced him has not one coded game. The data exists at the exit and is missing at the entrance — that inverted pyramid is the clearest structural signature Indian basketball has.

The consequence is that we judge players by their international output and can never measure their domestic development. When a nineteen-year-old is picked for the national team, the selection committee does not have his offensive rating across forty games. It has a coach's opinion, two trial notebooks and a state association's form. That is not a broken selection process; it is the only shape a selection process can take when no data exists.

I return to the sheet with 1,180 rows not out of nostalgia. I return to it because it remains the most detailed public possession record in Indian basketball, and it belongs to a league that no longer exists.

That sheet showed something the scoreline never did. Chennai Slam's fourth-quarter offensive rating for the season was 108.6. The moment opponents flipped into a 2-3 zone, it fell to 89.2. Nineteen points per hundred possessions — eight to ten points in a close game. What was happening on the floor was this: no patience to reverse the ball over the top of the zone, kick-outs arriving a beat late after guard penetration, and the shot clock expiring the moment the ball entered the high post. All of it is in the sheet. None of it ever appeared in a newspaper.

The sheet also records its own defects, which matter more to me. The stream died in the second quarter of three of the fourteen games. I reconstructed those possessions from the scorebook and a local reporter's tweet thread — estimates, not measurements. My coding rules were simple: an offensive rebound did not start a new possession, free throws from a shooting foul belonged to the same possession, and deliberate end-of-game fouling was excluded from team defence. I wrote those rules down afterwards, not before. That was my biggest methodological error and I do not hide it.

I trust the possession count more than the final whistle — but only when the count is published alongside its coding rules and its list of missing games.

That is why every chart I publish ships with three things: a method note, a list of absent games, and a link to the raw sheet. If someone recounts my 1,180 possessions and proves me wrong, the field moves forward. If someone merely cites them, it stops.

I ran the same method against a different question in 2026-21. The entire Indian Super League season after Covid was played behind closed doors in Goa — 110 matches, no crowd, no noise to smother a coach's voice. I set those 110 matches against the three pre-pandemic seasons and used the NBA's Orlando bubble as a control. Home win rate fell from 45.8% to 38.1%, and away-team bookings dropped 19%.

The wrong way to read those numbers is as proof that home advantage disappeared. The right way is to decompose home advantage into its parts — travel, officiating, crowd, surface. In the closed-doors season travel remained, the surface remained, the officials remained. What was removed was the crowd. The 7.7 points that fell were, almost entirely, the crowd's share. Any club needs to know which slice of home advantage it controls — and the crowd is one of them, the officiating is not.

I held that study back eleven days because I wanted to add a referee-assignment variable. When it did not materialise I published at ninety per cent and versioned the rest. The news cycle had come and gone inside those eleven days. The lesson was plain: in Indian sports journalism, the price of completeness is irrelevance.

When football analytics budgets froze in 2026 I moved to basketball full time. The 3x3 debut at Tokyo arrived as a clean mathematical sandbox — ten-minute games, a twelve-second shot clock, one ball, eight teams. I coded all 36 men's games and ran a possession-value model across them. Latvia won gold, and the win was not about scoring; it was two-way possession efficiency, where the average value of every attack stayed constant regardless of the score.

I turned that model on Princepal Singh. The NBA Academy India graduate played for G League Ignite and later for Sacramento's Summer League side. Indian media looked at his points per game. I looked at his rim-protection and offensive-rebound value, both in the top decile among Ignite bigs. His scoring rate was irrelevant. His role was the relevant fact.

One rule came out of that piece and I apply it to every player report now: name the role, name the mechanism, then name the number — never the reverse. Two agents still quote the line back at me, and I suspect they remember it because it makes their job easier rather than harder.

Back to the empty sheet. Everything available about that game amounts to a scoreline, some scorer names and a photograph. No lineup data, no possessions, no shot zones. That does not mean the game did not happen or that nothing can be said about it. It means that whatever gets said about it will not be analysis of the game — it will be a report written from outside the arena.

And here is my professional complaint. A large share of Indian basketball coverage is exactly that: a report filed from outside the arena containing not one measured fact about the play. We assemble a story out of "brilliant win", "disappointing loss" and "outstanding performance" — a story nobody can verify. When there are no numbers, adjectives become the only currency left, and adjectives are never proven wrong because nobody measures them.

My most useful habit as a reporter has been publishing the things I did not get. The eleven-day-delayed study, the sheet with three incomplete games, the chart that outlived the Chennai league it was built to explain — these are not weaknesses in my record, they are the method. The chart survived the Chennai league it was built to explain, and that is not a lament — it is a warning about models that fall in love with their own elegance.

This is where I have to disagree with myself.

The obvious conclusion is that Indian basketball's problem is a data gap, and that if we start logging every game the improvement begins. I do not buy it. The absence of data is a symptom, not the disease.

Consider what logging requires. A scorer, a tablet, a training session, and — most importantly — an institution that still exists next year and still owns the data. The third condition has never been met in India. When the UBA folded its data folded with it, because nobody owned it. 3BL, EPBL and INBL folded the same way. Until venue ownership, broadcast money and federation posture are settled, there will be nobody to keep the log. This is not a data problem. It is an ownership problem.

Second, and this cuts against my own work. I counted 1,180 possessions by hand, so I carry the instinct to treat my count as truth. That is dangerous. Indian games have no official play-by-play, which makes my count the only count, and scarcity breeds confidence. A single observer's figure is an estimate, not a measurement. I write that every time and every time it gets read past.

Third, the most uncomfortable point. The Indian basketball stories that travelled furthest in the last decade were built on hope rather than data. Draft night in 2026, when Satnam Singh went 52nd; the NBA Academy opening in Greater Noida in 2026; the NBA India Games in Mumbai in October 2026. All three were real events. None of them left a durable possession dataset. All three built a narrative in which the centre of Indian basketball is external validation, not the guard playing for the Railways team who is twenty-eight and played thirty-four games last season.

A sport that points all its attention at validation from outside never builds its own internal yardsticks.

The inverse is just as wrong and I want to be clear about it. I am not saying the Academy or the India Games were worthless. I am saying that treating them as the sport's centre of gravity is a geographic delusion. The Academy is one pathway; it is not the only pathway and it is not the base of a pyramid. A country whose Senior National Championship has no digital record cannot call an academy a pyramid without ignoring the ground beneath it.

So I also reject the conclusion my own argument produces — that because there is no data, analysis is impossible. Analysis is not impossible. It is expensive. It has to be built by hand, possession by possession, and shared with anyone willing to recount it.

Which brings me to what I think matters most over the next two years, and it is not a franchise league.

FIBA runs a genuine 3x3 ranking system tied to individual players and teams, and Olympic qualification runs through it. That is a route for an Indian player that waits on no franchise, no broadcaster and no promoter's budget. It needs a court, three players and a tournament registration — and every tournament issues points that are publicly recorded. It is one of the rare places where Indian basketball data builds itself without an institution's blessing.

This does not mean 3x3 replaces the five-a-side game. It means a structure exists in which ranking, performance and opportunity sit on the same public register — and no other structure in India does that.

The second route is the one nobody covers: the women's Division B campaign. There is no media scrum there and no sponsor, but there are games — and games mean possessions. Anyone who sits through one season and codes it will hold the first genuine performance record in Indian women's basketball. For a researcher trying to occupy unique ground, I think that is the best investment available, because the competition for it is zero.

The third route is the least exciting and the most necessary: the bridge from school to Senior Nationals. Tamil Nadu and Punjab school systems play hundreds of matches a year and retain a record of none of them. Log the district finals of those two states alone with five fields, and in five years India will have its first player-development chain — and a selection committee will have something other than opinion.

One last thing about the empty sheet. I am not tearing it up. I am filing it in the same folder as the 2026 sheet, because they are two ends of one story. One holds 1,180 rows and no league. The other holds columns and no rows. In nine years we added facilities, improved broadcast, multiplied cameras — and the state of the record did not move.

If not one Indian domestic tournament publishes a complete possession log by the end of next season, take it as settled that we learned nothing about the game in nine years. We only talked about it more.

The question is not why Indian basketball has no data. The question is: when every arena already has a scorer and every game already generates 150 possessions, whose pocket do those rows go into — and why do they never come out?