The Empty Report: Twelve Correct Sections, Zero Lines of Data

At 2:47 in the morning, the second monitor in a Los Angeles apartment lit up with a JSON file twelve sections long. The section headers were correct. The section order was correct. The date format was correct. And every data field carried the same two words: not available.

I sat and looked at it for four minutes. Seventeen years of reading basketball reports, and I was used to files missing numbers, missing names, missing even the name of the league. A file with a perfect template and a hollow core was a first. It looked like a box score printed out by a machine: every column ruled, every cell divided, and the one thing that makes a box score mean anything was absent — who scored.

If I pushed that file to the next stage, the system would not raise an error. It would read, classify, and produce an analysis with nine full sections, three conclusions per section, two hidden insights per section. Beautiful, like a published article. And not a single true word in it.

That twelve-section skeleton is what I build for every analysis: tactics and technique, player data, team operations and the salary cap, league landscape, rules and governance, coaching staff and locker room, risk, media narrative and expectations, and the industry ripple. Each section has its own table, its own criteria, and one mandatory line for evidence citation. The skeleton was built to defend against the very thing it had just produced: a report that looks complete.

The pipeline runs in two stages. Stage one reads the source text and extracts atomic information points — every sentence, every number, every name becomes a citable unit. Stage two takes that list and runs it through the twelve sections, and every conclusion must point back to a specific information point. With no information points, stage two has nothing to point at. It has only the skeleton.

That night, stage one returned a schema-valid, content-empty object. No source title, no source, no one-sentence summary, no entity identified, and an information-point list with exactly zero elements. Four possible causes, ranked by likelihood: the source text was never ingested, whether because of a paywall, a login wall, geo-blocking, or a JavaScript-rendered page returning an empty body; the extraction model returned null through a parsing failure or a mid-stream truncation; the wrong object was passed from stage one to stage two; or the source was video, podcast, or an infographic and needed transcription before any analysis.

The interesting part sits elsewhere. A real basketball article almost never yields zero entities. There are teams, players, coaches, scores. Absolute emptiness is the signature of a pipeline fault, not of an article with no content. Put differently: the emptiness here is itself information. It tells me where to go back to, not what to write.

I have met three other kinds of emptiness in this job, and each taught a different lesson.

In 2026, aged 24, I sat at Summer League with a spreadsheet in hand. Dillon Brooks, an undrafted free agent, posted a defensive rating of 98.3 across five games; Troy Williams, competing for the same slot, posted 104.2. The data was sufficient. I had evidence, I had a conclusion, I had a probability model nearing completion. I held the piece back three weeks to refine the model. Another blog published a piece praising Brooks three days before me. Mine went unread. That was emptiness by hesitation: the data existed, and nobody received it.

In 2026, aged 27, I sent a forty-page report to the LA Clippers medical staff on Kawhi Leonard's hamstring re-injury risk under a dense schedule following a long layoff. The model's number was 1.6 times. The report was ignored for being too verbose. In August, Kawhi went down exactly as forecast. That was emptiness by format: the data was right, and nobody read it.

In 2026, aged 29, I wrote two pages on Enzo Fernández for a Premier League sporting director. Progressive passing at 11.4 metres per 90, a 78 percent success rate under pressure, the best among under-23 midfielders at the Qatar World Cup. I recommended a 30 million euro valuation. In January 2026, Chelsea paid 120 million euro. Those two pages leaked onto a data forum. That was emptiness by leakage: the data was right, was read, and no longer belonged to its author.

All three share one feature. I was not wrong about the numbers. I was wrong at the handoff.

In a statistical table, a zero has two entirely different origins. A player with no field-goal attempts in a first quarter because he was smothered, because nobody passed to him, because he stood in the wrong spot — that is a fact about the game. The same zero, if the tracking camera lost him for four minutes, is a fact about the camera. Two facts, two conclusions, two actions. Collapsing them is the most common error in sports analytics.

The largest blind spot in basketball data is not a shortage of numbers. It is the failure to separate an absent number from a zero.

A more concrete case. When a medical staff reports no injury, the sentence can mean the player is healthy, or it can mean nobody ordered an MRI. Across an 82-game season, those two readings differ by weeks of recovery. An early reader tells them apart through one small detail: the imaging history. If the player has not been scanned in six weeks, the phrase no injury is a statement about process, not about a knee.

The same logic applies to the empty report that night. Each blank field across the twelve sections could mean there was nothing worth saying about this game, or it could mean the system retrieved nothing to say. Stage two cannot tell the difference on its own. The reader has to do that work.

Basketball has a concept called empty stats — a good line on a losing team, or numbers accumulated in garbage time. The empty report is the paperwork version of the same disease: hollow content, full form.

The damage mechanism sits on the reader's side. When a document has clear headers, tables, and an evidence-citation line, the reader lowers the level of scrutiny. Format acts as a licence. A report without headers makes people ask where this number came from. A report with headers makes people ask what the conclusion is, and skip the first question.

Data is like a book. The crowd looks at the cover, the wise read every page. But there is a kind of book with a beautiful cover, a table of contents listing twelve chapters, and a blank sheet in every chapter. Even the wise have to turn the pages to find out.

Three failure layers, three different fixes.

Emptiness by ingestion: the source text never entered the system. The symptom is an empty response body with a 200 status code. The fix sits on the collection side — capture raw HTML or a rendered screenshot, then extract text before re-running. Re-running without changing the collection method produces an identical result, because the cause was never touched.

Emptiness by extraction: the text exists, but the model returned null or was truncated before the information-point generation step. The symptom is a file of unusual size, or an empty summary field while the title field is populated. The fix is to check input length and add an information-point count before the stage handoff.

Emptiness by medium: the source is video, podcast, or an infographic. There is no text to extract. The fix is to add a transcription or OCR layer, a layer that does not currently exist in my pipeline, and that is a genuine hole.

Those three layers map onto three familiar scouting situations. A player who does not appear on tape may be injured, may be out of the rotation, or the tape may be from the wrong game. Three different conclusions, three different actions, and an inexperienced scout will fold all three into nothing to see here.

In roster construction, you place the checkpoint where it is cheapest. A bad contract blocked in the scouting room costs an afternoon; blocked at the cap sheet it costs three years. Same principle for a data pipeline: an empty file blocked between stages costs one re-run; caught at the final draft it costs an article and some credibility.

That night, my pipeline accepted a schema-valid, hollow object and raised nothing. The only checkpoint that could have stopped it was a hard conditional: if the information-point list is empty, or no entity is identified, refuse the handoff. One line of code. Without it, the system will always produce something, because the system was designed to always produce something.

The point I want to stress: this incident is not stage two's fault. Stage two did its job — it honestly reported that there was nothing to analyse. The break was that nobody made it stop before walking into the twelve sections.

In scouting, an insufficient information report is a valid and valuable document. It carries a signature, a date, a reason. Its value is that the next person does not have to redo the work. An NBA team may receive hundreds of such reports in a season. They never make the news. They are what keeps a scouting department from fooling itself.

The empty report that night could be handled the same way: stamped, dated 11 February, the four possible causes listed, the fields needing re-ingestion listed. One page. It is not an article. It is the invoice for a failure, and invoices must be filed.

Every finding needs a moment before it becomes a truth. The empty report is no different. Today it is a technical fault. Three weeks from now, if someone publishes a nine-section analysis from that same source with not one information point behind it, it becomes evidence of something more serious: that sports analysis is producing articles with no provenance at all.

I apply this principle to people as well. A scout who files a report saying this player has potential without a single number, a single possession, a single timestamp — that report should be returned like an empty JSON file. A transfer item saying the two sides are in talks with no named source, no timeline, no figure — the same kind of emptiness.

This industry has an odd habit: it forgives vagueness and punishes precision. A sentence saying it could happen is never wrong. A specific number can be wrong and will be remembered. That is why reports grow vaguer, more formatted, and emptier by the year.

Data that is correct and ignored is not data — it is a debt owed by the person who refused to read. And an empty report presented beautifully is a debt owed by the person who wrote it.

What I write today may be forgotten. The system it builds will not be. Four failures in seven years — Brooks, Kawhi, Croatia, Enzo — taught me one thing, differing only in where I erred: sometimes I held the data too long, sometimes I presented it too densely, sometimes I let it leak. All three were handoff errors, not analysis errors.

The empty report is the fifth, and the first where the fault occurred before I had any data at all. It forced me to look at the earliest link: collection. There, no amount of analytical skill rescues a page blocked by JavaScript.

Croatia did not reach the final by accident. They were led by people who could read numbers. But those people also had to have numbers in hand first. In 2026 I had 74 percent of possession in the middle third and 12 key passes from Modrić across the cup matches. The piece was buried because the byline was too small, then shared 3,000 times in a single night when Croatia reached the final. Had the data come back empty that night, I would have had nothing to bury, and nothing to dig up.

This industry worries about missing data. The larger risk is data that arrives too complete, too correctly formatted, too ready to publish.

An empty report is honest. An empty report filled with guesswork is a liability. And that liability is always paid by the reader, never by the writer — the writer was paid in advance.

There is a paradox in how newsrooms operate: a null result is not news. Nothing happened does not go on the front page. So the null result gets pushed aside, and the gap gets filled by someone with less information and more confidence. That is the mechanism behind most of the wrong basketball stories of the past decade: not that somebody lied, but that the gap was not allowed to stay empty.

I once believed an analyst's value lay in how often he was right. I now believe it lies in how often he dares to say not enough. A report saying not enough never reaches the front page, and precisely for that reason it is the most trustworthy thing on the desk.

Emptiness is not the enemy of analysis. The enemy is manufactured completeness.

Next time an analysis of a player you follow is published with every section filled, count the information points inside it before you read the conclusion. I have set my own verification date: 30 June 2026, when the season closes, I will reopen the JSON file from the night of 11 February and write beside it the number of empty reports I stopped before they became articles. If that number is zero, my system still is not working — even if the pieces keep coming out on schedule.