An Empty Cell Is Not a Clearance: Auditing a Null Result in Tennis Analytics

27 matches. 4,312 points. Eleven days. And one column out of six that never got filled.

January 2026, Chennai. The Chennai Open was playing its final edition at the SDAT Nungambakkam Stadium. I had a borrowed media pass, a laptop, and a six-column chart template: serve placement, return depth, rally length, returner position on second serve, break-point outcome, and entries. The first five columns filled 27 times across eleven days. The sixth stayed empty. In the final, Roberto Bautista Agut beat a 20-year-old Daniil Medvedev 6-3, 6-4, and the line in my chart that said the most was this: Medvedev was returning second serves from roughly 1.4 metres behind the baseline and winning only 38 per cent of those points.

I published that chart as 'The Last Chennai Open' on a bare-bones site of my own. It drew 40,000 reads. After that I stopped writing match reports as narrative. Every claim now carries a denominator — per 100 points, out of 27 matches, out of 81.

Last week a nine-dimension tennis analysis framework landed in front of me. All nine dimensions returned the same answer: 'N/A — insufficient information'. No title. No source. Article type unclassified. The information-point array empty. No player, no tournament, no date. The only surviving field was the domain label: tennis.

My first instinct was to delete it. My second instinct stopped me, because deleting it would have been exactly the error I flag every week in injury reporting — reading a blank report as a fitness certificate.

So this is an audit of that null result. And it is, more usefully, an audit of three ledgers where tennis's own record rests on empty cells that coverage has never bothered to count.

What a null result is, and what it is not

The pipeline had two stages. Stage one extracts facts and viewpoints from source text; stage two applies domain expertise to those facts. Stage one extracted nothing. Stage two had nothing to apply.

A word on vocabulary, because tennis coverage treats the two cases as identical. A null result means 'no assessable evidence is available'. A negative result means 'evidence exists, and it points to absence'. The gap between them is the whole distance between a scan and a press release that omits a sentence.

The framework refused to rate overall risk, and gave its reason in one line: no information is not the same as low risk. That sentence does more work than most injury columns published in a season. The absence of an ITIA case in a story is not compliance. The absence of a medical-timeout controversy is not evidence that tactical MTOs have stopped. The absence of any mention of the 52-week points-defence cliff does not mean no player is standing on its edge.

The framework did one thing correctly: it did not launder its own emptiness. It refused to convert 'N/A' into 'normal', which is what every reader-friendly desk would want. And it logged its real failure — not that tennis told us nothing, but that a fully populated template can look finished over an empty input. That is the live danger. A clean nine-section report reads like an analysis has happened.

I will not write further about the framework. I will apply its question to tennis's own record: where have we been reading empty cells as green flags?

The first ledger: the empty injury cell

Injury information in tennis is private property, not public documentation. A small fraction of what passes between a player, his team and the tour reaches a press conference, and that fraction is usually shaped by the commercial interest of that particular week. This is not conspiracy; it is the ordinary structure of incentives. The week of a sponsor announcement, the same injury is 'minor' that becomes a 'precautionary withdrawal' seven days later.

The consequence is that what appears in coverage is description, not diagnosis. Strapping on the left thigh. The physio walking on at 3-4 in the second set. First-serve speed falling from 198 km/h to 184 across a set. Second-serve percentage climbing as a player protects a shoulder. First-serve speed suddenly returning in a tiebreak. These are observed facts. 'He is fit' is a conclusion, and it does not follow from them.

I charted India's Tokyo Olympics tennis campaign point by point: 19 break points, 2 converted. That number is a report in itself. But my columns did not record whose wrist was taped that week or whose knee was swollen, because that information does not come from the court, and I do not fill it in by inference. The honesty of my chart is that it states what it does not know.

There is a second layer readers never see. The gap between a return announcement and actual fitness is the least measured interval in tennis. No counter exists, no public register records how many players retired within their first three events after a declared comeback. If it did, comeback stories would carry their real failure rate beside them.

This empty cell distorts more than headlines. A player quietly losing serve speed holds the same seed, and the opponent across the net has no way of knowing whether the upset he just produced was an upset at all.

The second ledger: 21 editions, then nothing

I keep a dated ledger I update every year: ATP or WTA-level events staged in India.

The Chennai Open ran from 2026 to 2026 — 21 editions, one venue, Nungambakkam. After 2026 the licence moved to Pune, which ran for a few years as the Maharashtra Open. Since then the calendar carries no ATP or WTA event in India. At the start of every new season I write zero in that column. The zero is not a mood. It is a count.

This is where coverage stopped counting. When a tournament leaves, it does not become news. The news happens the day the licence is sold; the emptiness that follows enters no headline. I remember standing outside the stadium on the last night of 2026 wondering where the next edition would be held — and nobody asked, because the story that night was the final.

This absence has an owner and a date, and both should be on the record. 'Indian tennis is in crisis' is a feeling. '21 editions between 2026 and 2026; zero on the current calendar' is a fact anyone can check.

Rohan Bopanna won the Australian Open men's doubles title at 43 with Matthew Ebden. That achievement does not stand alone; it is evidence of a pathway, and the pathway does not run through a domestic tour. It runs through academy years in Bengaluru, a European base, and repeated direct entry into Grand Slam draws. With no home tour, a player has no domestic ladder to climb; he has to grab a foreign one, and the cost of that climb is carried by a family, not a system.

On men's singles, my ledger puts the Indian ceiling at roughly No. 68. That figure comes from career-high lists, and I publish it with the same warning every time: it is an individual peak, not a system average. A career-high tells you nothing about how many players reached that level and slid back. That count does not exist anywhere.

The third ledger: 18, 6, 2 — counted but never stated

Leander Paes: 18 Grand Slam titles. Sania Mirza: 6. Rohan Bopanna: 2. Mahesh Bhupathi: the 2026 French Open mixed doubles with Rika Hiraki, India's first Slam title, from a partnership nobody was tracking at the time.

Those numbers are exact and should be stated exactly. And that is precisely where the 'India's singles dream' framing collapses — because it never states them. It runs on the language of dreams, which cannot be tested and therefore can never be wrong.

My claim is that Indian tennis is not an under-achievement story. It is a category error. In a sport where the country produced some of the most competitive doubles players in the world, coverage chose as its success metric the one discipline the country never invested in. That is not a failure of the sport. It is a failure of reporting, and failures of reporting can be corrected.

I have to apply my own warning here or I become the thing I audit. '18 versus 2' is meaningless without era, discipline, partner quality and draw size. Paes's 18 split into eight men's doubles and ten mixed, spread across four decades. Bopanna's two belong to a different era, when doubles draws and scheduling had already changed. Comparing them without scope conditions is exactly how GOAT arguments are built and why they never resolve.

Add one clean number: of those 18, how many came with an Indian partner and how many did not? That question either strengthens or breaks any claim about an Indian doubles tradition. I have counted it and I am deliberately not printing the answer, because a count you do yourself is a count you remember.

The contrarian angle: hand-charting fails silently too

I charted it by hand before I trusted the model. That is my habit and I treat it as my credential. Honestly examined, the habit carries the same disease it claims to cure.

My 2026 chart records 27 matches and 4,312 points. It accounts fully for the players who were on court. It accounts for nothing about the players who never entered the qualifying draw. My dataset suffers from survivor bias, and that bias sits inside the figures I trusted most. When I wrote that 38 per cent figure for second-serve returns from 1.4 metres back, I did not write how many players tried the same position and exited in the first round, never entering my chart at all.

And the chart has a second property that makes it unusual: the tournament no longer exists. I cannot re-measure it. My data is no longer verifiable, only archived. That is why I date and place every chart, and why I do not trust anyone else's chart without checking its date.

1,412 shots later, the noise finally introduced itself as a pattern — Russia 2026, when I logged every shot of all 64 matches into an open expected-goals model. Croatia scored 14 goals from 7.9 xG, and I wrote before the final that regression was due. France won 4-2. That was the model's best day, and I do not call it my victory, because the model could have been wrong and the same article would then be evidence against me.

The Wimbledon semifinal that gave me my pressure index — 6 hours 36 minutes, 99 games, 26-24 in the fifth — still rests on my weakest sample. The index showed ace rates climbing in the closing games rather than falling. I presented it as serving improving under pressure. An equally strong alternative reading exists: the returner was tiring, so the ace rate rose. Both explanations come out of the same data, and my sample is one match. An index built from one match is not an index. It is an observation.

In April 2026, with my graduate analytics offer withdrawn, I built the Empty Stands Dataset — all 81 Bundesliga matches played behind closed doors, hand-logged. My count put the home win rate at 33 per cent against a 43 per cent baseline over the previous 18 months, with home penalties per match roughly halved. I folded in fan-free tennis events and measured second-serve speed up 1.8 km/h.

The dataset is public, and its biggest weakness sits on the same page: the closed-doors effect cannot be separated from the other causes active in that window — restart rust, compressed fixtures, five substitutions. I recorded the difference between 33 and 43. I did not prove that difference belongs to the crowd alone. Anyone presenting it as 'the crowd effect' is claiming more than my data supports.

That is the discipline the framework showed in its null result and that daily reporting lacks: keep correlation and causation apart, and state where you stop.

What to track next

Three columns over the next twelve months, each with a number for an answer rather than an opinion.

First: the number of ATP or WTA-level events staged in India. If it is still zero at year's end, that is a recorded fact, not a complaint. Second: Indians inside the men's top 100, and of those, how many entered main draws directly versus through qualifying or wild cards. That ratio measures the health of a system, not the peak of one player. Third: retirements and mid-match withdrawals involving Indian players on tour, and how many were preceded by any public injury disclosure.

The third column will speak loudest, because that is where our empty cell costs the most. On the day a player walks off holding a wrist, the real question is not what happened to him. The real question is who recorded his serve speed across his last three entries, and who chose not to print it.

And when the next ATP event in India is announced — if it is — the first question in coverage should not be who is playing. It should be why this is the first since 2026, whose account the intervening eight years sit in, and how many Indian players reached the top 200 in that period without a home tour. On the day we count that, we will know whether we lost eight years or saved them.