The False-Completeness Trap: When a Basketball Analysis Sheet Is Full of Cells but Empty of Facts

Two in the Morning, One Heavy File

At two in the morning I opened a heavy file. Nine sections. Every section had tables. Roman numerals on some, pale blue shading on others, a line of italicised text reading 'Confidence: Medium' at the foot of each conclusion. The structure was tidy enough that I assumed I was reading a genuine analytics department's summary, the kind EuroLeague clubs still exchange after every round.

I scrolled down looking for a player's name. Nothing. A team's name. Nothing. Any number at all. Nothing either.

All nine analytical sections had been written out in full. Each had a conclusion. Each conclusion carried an 'Evidence' line. The pages looked thick. Inside, they were exactly as thick as the amount of air they contained.

What kept me up until nearly dawn was not the emptiness. It was the fullness that troubled me. A blank document deceives nobody. A document stuffed with cells, tables, subheadings and professional vocabulary deceives almost everyone who lacks the time to read to the last line. Every result is a deliberate lie. The empty analysis is the most sophisticated member of that family: it does not invent numbers, it simply makes you believe the numbers exist somewhere.

One File, Nine Sections, and a Name Nobody Went Looking For

I have spent ten years in the business of taking sports analyses apart. It started with two-thousand-word blog posts on forums, one of which drew forty-seven reads, roughly ten of them my own. Now I host the podcast Vùng phủ sóng, twenty-five minutes an episode, dissecting one tactical situation at a time. My daily work is reading documents like last night's: post-game summaries, transfer dossiers, assistant coaches' notes.

The volume of these documents is growing fast. Partly because automated content platforms can now push out thousands of 'post-game analyses' in a single round of fixtures, each several thousand words long, each with a proper hook, proper arguments and a proper closing. Partly because of readers. We have grown used to a form of text that has the shape of analysis. One glance tells you which template it was built from. The eye slides across it quickly, and during that slide it registers the text as 'having content'.

A real analytics room at a EuroLeague club works in the opposite order. It begins with a question, then a subject, then a number. If the question has no subject, the work stops. Nobody would dare send up a nine-section document that cannot name a single team, because the head coach would hand it back within thirty seconds.

Last night's document did the reverse. It started with the frame, then went looking for the filling. When it found no filling, it kept the frame and filled the frame with the frame itself. Where the tactical section should have said 'drop coverage on the pick-and-roll', it said 'N/A — insufficient information'. Where the player data section should have carried a name, it said 'No player identified'. Where the industry impact section should have carried a trigger, it said 'N/A' six times running, with a note stating that any claim produced there would be fabricated from start to finish.

So all nine sections, in the end, answered correctly. But they answered correctly in a way that made the document four thousand words longer without adding a single new fact. That is what I want to address here: there is a kind of emptiness the naked eye cannot see, and it is spreading through Vietnamese sports content faster than any ordinary error.

Empty Stats, at the Level of Content

In player evaluation there is a concept the numbers people call empty stats. A player scores twenty points in a thirty-point loss, all twenty in the fourth quarter after the opponents have pulled their starters, and the shooting percentages look immaculate. The box score is full. The impact on the result is nil.

I have spent most of my career stripping that kind of number back. Only last night did I realise it has an upper-storey version: empty stats at the level of content. An article with a headline, subheadings, three-part structure, tables, advanced terminology, an open-ended conclusion. Count the words and it is long. Count the columns and there are many. Measure what the reader carries away — nothing.

It is more dangerous than a player's empty stat in exactly one respect. With a player, you can rewind the tape and count for yourself. With content, the reader has no tape. The reader has only a feeling.

The Four Things a Post-Game Analysis Needs

After years in this trade, I have settled on four minimum requirements.

First, a named subject. Not 'this team' — the team, the player, the coach. Without a name there is nothing to analyse, because every tactical conclusion depends on who runs it and who defends it.

Second, a quantitative anchor. One number, any number, traceable to a source: points, possessions, metres of spacing, percentages, contract years. The number is the hook that stops the story from drifting away.

Third, a time frame. Which game, which quarter, which stretch of the season. Basketball in the first quarter differs entirely from basketball in the final four minutes, and an analysis that fudges time cannot be tested.

Fourth, a source tier. Who recorded this, which body published it, or did I count it myself in my living room. Those three tiers carry completely different weights.

Last night's document satisfied all four — formally. It had slots for a subject, slots for numbers, slots for timing, slots for sourcing. It lacked precisely the one thing that makes the other four anything more than slots: content.

Kevin Love, Game Five, and a Post with Forty-Seven Reads

Based on my own tracking experience watching games, the clearest example of numbers that lie beyond what the eye can catch is Game Five of the 2026 NBA Finals, Cleveland Cavaliers against the Golden State Warriors. The Warriors won 129-120 and took the series 4-1.

I was seventeen. I spent seventy-two hours rewatching the final fourteen possessions. In my hand-kept chart, Kevin Love's effective field goal percentage in that game came out at 38.5%. Looking at that line, any commentator has licence to say Love shot poorly.

At the same time, I counted six occasions on which Love stretched the defence. Those six did not produce points for him, appeared in no box-score cell, and on my reading led directly to ten LeBron James points. Read only the box score and you see a player performing badly. Watch the tape and count, and you see a player performing his role and carrying an influence the box score does not pay for.

The two-thousand-word post I wrote afterwards, headlined around invisible value, drew forty-seven reads. Almost nobody cared. But it taught me the rule I still work by: open with a jarring number, then use tactical context to explain what that number means, rather than narrating the game chronologically.

What matters here is that 38.5% was a real, measured, methodologically sound, sourced number. It still leads to a wrong conclusion if you stop at it. Last night's empty document did not even have one real number to stop at. It was empty at a deeper floor.

Özil, 0.4 xG, and Two Hundred Comments

The same mechanism recurred at the 2026 World Cup, when I first trialled expected goals on a major tournament.

I calculated Mesut Özil's xG across Germany's three group matches and arrived at 0.4 in total. Measured against his Arsenal season, that was a drop of about 41%. Television did not mention it. Commentators talked about spirit, about desire, about a former champion no longer being itself.

I wrote a forum post arguing that Özil had been abandoned inside a system that moved too slowly under head coach Joachim Löw, and that his numbers had collapsed not because he had declined but because the team's structure no longer created positions for him to receive the ball.

That post drew two hundred comments. Many disagreed. None produced counter-evidence.

I bring this up not to boast, but because it shows something important about the market: readers react most violently to posts that contain numbers, even when they disagree, even when they cannot verify. A number generates force. An empty document generates none. In a market where force is measured in views, the empty document is at a disadvantage — so it learns to impersonate force.

Olympiacos, 4.7 Metres, and 63%

Three years later, during the pandemic shutdown, I retreated into an archive of old data to cope with the anxiety. I spent nine weeks rewatching eight Olympiacos games in the EuroLeague.

I measured the average distance between the two defenders in pick-and-roll situations: 4.7 metres. I counted that they forced opponents to move the ball to the right side 63% of the time. Those two numbers combine into a very specific tactical story: they accept a shot from one side, and they stand further apart than normal to close the driving lanes.

From there I recorded thirty podcast episodes myself, twenty-five minutes each. Episode twelve, on drop defence, was spotted by a basketball podcast producer who invited me to collaborate. That was the turning point from hobby to work.

The professional lesson is simple: a number has value only when it travels with an observable behaviour. 4.7 metres is not knowledge. 4.7 metres plus accepting the shot from the right side is knowledge. Had I published 4.7 metres and then written two thousand words interpreting it, I would have produced a full-cell, hollow document. A podcast is not born in the studio, it is born in the silence of the world — and that silence is where I sat counting, not where I sat writing.

Italy, 4.2 Metres, and a Three-Hour Call

The final example is the time I nearly fooled myself.

At the Euro 2026 quarter-finals, held in 2026, I dug into how Italy defended under Roberto Mancini. I analysed the 120 minutes against Belgium and measured the average distance between the five defenders at just 4.2 metres — nearly a metre tighter than in the group stage.

A clean number. A tidy number. A number telling an appealing story about a team suddenly compressing into a block. I finished the podcast script and nearly published it.

Then I called an Italian assistant coach I knew from a forum. The conversation ran three hours. He asked one question that collapsed my conclusion: was that distance a tactical intention, or a situational reaction after Belgium took the lead and Italy had to push up?

We never resolved it. I rewrote the entire episode, three and a half thousand words, presenting both opposing possibilities instead of settling on one. That episode became the most downloaded content of the month, passing five thousand listens.

The lesson was not 'do not conclude early'. The lesson was: a measurable number is not necessarily a meaningful number, and the gap between those two is where the writer has to sit. Last night's empty document shares one strange feature with my Euro 2026 mistake: both were highly confident about their own shape. The difference is that I had three hours on the phone in which to be contradicted. It had nobody to call.

What the Empty Document Provides

After those four examples I built myself a small comparison table.

An analysis with substance provides a subject, a measurement, a time frame, a source, and a hypothesis that can be falsified. It gives the reader something to argue with. It puts itself in danger.

An empty analysis provides structure, subheadings, confidence classifications and sentences describing the process of its own production. It puts itself in no danger at all, because it asserts nothing falsifiable. It merely describes its own inability to describe.

And this is the part that made me write this piece. Read by a professional, a document like that is harmless. Professionals see the letters 'N/A' and know immediately. But read by someone lower in the content chain — a new writer, an editor chasing volume, an automated system harvesting data from it — it will be consumed as a genuine analysis. It looks like one. It has all nine sections. And it will flow downstream.

A Null Result Is the Most Honest Result, and the Market Hates It

Now the counter-intuitive part, and I have to admit it makes me uncomfortable.

If you asked me which analysis was the most valuable of everything I have read, my answer is last night's file. Not because it was informative — it contained exactly nothing. But because it was honest in a way almost no sports document dares to be.

When there was no player name, it stated that no player name could be determined. When there was no data, it stated that there was no data. When every further inference would require invention, it stated plainly that every further inference would be invention. Analysts call this a null result, and it differs fundamentally from a neutral result. Neutral means 'not enough data to conclude in either direction'. Null means 'no data exists on which to say anything at all'.

The problem is that the market hates a null result. Here I have to argue against myself. It is easy to write a piece praising the honesty of the null result and to turn that into a very elegant moral position. But if I am honest, in my first three years in this trade I wrote plenty of empty pieces. Pieces about games I did not watch, built on stat sheets I downloaded. Pieces with full structure, full opinions, full open-ended closings, and not one observation of my own. I wrote them because they sold easily. I wrote them because nobody caught me.

Which is why I do not treat this as a story about machines. Machines only accelerated what people had already been doing for years. What changed is scale. One hollow writer produced one hollow piece a day. One hollow system produces ten thousand hollow pieces before lunch.

And there is one more point I have to make, uncomfortable as it is. A winning machine is only an illusion until somebody is willing to break it. In sports content, what gets called performance is really willingness to consume. Every time you click on a piece because its headline has air in it, every time you share a handsome table without checking its source, you are paying for the exact mechanism that produced last night's file. I do not exempt myself from that sentence.

So the genuinely counter-intuitive point is not 'write honestly'. It is this: if you want to remove empty content, do not start on the production side. Start with reading habits. Producers only respond to the signal readers send. For ten years the signal we have sent is: thicker is better, fuller cells are more credible.

There is a test I apply to everything I publish, and I invite readers to apply it to my work too. Count how many details appear that only someone who actually watched would know. If that count is zero, the piece is empty, however long it is. I cut three of my own podcast episodes last month using that test.

Variables for the Coming Round

Before I close, here are a few things worth tracking, in the way I do at the end of every Vùng phủ sóng episode, because basketball does not end with the whistle. Basketball never ends with the whistle, it ends with a question.

First watchpoint: the completeness rate of the automated analyses you encounter. If you read ten pieces in a week and none names a specific player, that is a systemic signal, not an individual one.

Second: the label-without-substance signature. A piece carrying topical tags, classifications and keywords, yet not one number traceable to a source.

Third: source quality. How many pieces you read today state who measured the number and when? If that share falls, an entire market's ability to separate news from rumour falls with it.

And the question I leave behind, not to answer now but to carry into next week: if an analysis was written and nobody can point to a single player inside it, what exactly was it analysing? I have my hypothesis. I will save it for the next episode.