Null Return in Table Tennis Analysis: A Complete Framework With Zero Evidence

Eleven at night in Shenzhen, a forty-one-page report sat on my screen. I opened it after the coaching-staff meeting, and for the first twenty minutes everything looked immaculate: nine analytical layers, ruled tables, bold headings, every cell filled with text. By page three I noticed something wrong. Every cell carried the same sentence — “insufficient information to assess.”

Forty-one pages. More than four thousand cells. Not one player’s name. Not one tournament. Not one metric. Not one date. The only real piece of information sat on the first line of the file: the domain label read table tennis.

The editor called at half past eleven and asked whether it could be turned into a three-thousand-word piece. An intern on the team was even keener, proposing we “fill it out” with a few stories about famous players and a couple of estimated figures. I said no. This article explains why.

What a report with no data actually is

In sports analysis there is a type of output almost nobody teaches you at school: the null return. It is not the same as “nothing worth saying.” It is not the same as “weak data” either. A null return means the pipeline ran its full course, assigned labels, split layers, built the frame — and found not a single piece of evidence to put inside it.

That report was a textbook null return. Its structure was flawless. Title fields, source notes, data cells, conclusion cells, risk cells — all present. But the contents were empty. The article-title field was left unresolved. The source field was left unresolved. The article-type field was left unclassified. The core-viewpoint field was empty. And the most important field of all — the list of information points — was entirely empty.

What stands out is that the domain label was still populated. Someone, or some system, had managed to identify that this document belonged to table tennis. But immediately afterwards, not a single entity was recognised. No player. No coach. No federation. No event. The entities field was left with the instruction “identify from the information points above” — while above there were no information points at all.

To someone who has worked in this trade for thirteen years, this is a familiar signature. It usually does not mean the original article was empty of content. It means the extraction stage failed, or the source was blocked, deleted, or truncated, or only the opening section came back without the body. But whatever the cause, what I was holding was a file that could not be used for analysis.

Nine layers and the cost of formal completeness

The framework in that file was designed around nine layers. I have worked with similar frameworks for years, so I can read the intent of whoever built the frame, even when the contents are blank.

The first layer is technique, tactics and equipment. To activate it you need at least one of four things: a named player with a style descriptor; a tactical review of a single match with scoring structure; a description of how a coach deployed the line-up; or an explicit statement of an equipment change. That file had none.

The second layer is player data and head-to-head records. It needs at minimum a named player, a current ranking, and either a head-to-head table or a set of recent results. Here, the WTT rolling 52-week points deduction can only be applied with a points ledger. No ledger, nothing to calculate.

The third layer is the event system and points rules. It needs a named event with a date, so it can be located within the Olympic cycle and mapped onto the points table. The current tier structure is fairly clear: the Olympic Games, the World Championships and the World Cup form the three majors; below them sit the Grand Smash, Champions, Star Contender and Contender tiers. With no event name there is no tier to compare against.

The fourth layer is the competitive landscape, especially the balance of power between strong associations. To discuss that you need at least two entities at association or athlete level placed side by side. The entities field was built to capture associations, yet it was empty.

The fifth layer is rules and governance. It only works when there is a clear rule trigger: a reform proposal, a selection dispute, a disciplinary precedent, or a change in governance structure.

The sixth layer is coaching staff and the talent pipeline. It needs a named coach or team, plus a signal of change: a hiring cycle, an expiring contract, a retirement wave, or internal trial results.

The seventh layer is the risk surface. It needs a concrete subject and at least one triggering event: a match, a ranking shift, an injury report, a selection decision.

The eighth layer is public narrative and expectation. It needs an identifiable claim or framing in the source text, plus a credibility rating for the source.

The ninth layer is industry transmission, from upstream equipment, youth development and training; through midstream events, associations and clubs; down to downstream broadcasting, commerce and derivative markets. It needs at least one named commercial actor or one policy signal.

Nine layers. Nine minimum input requirements. Not one of them was met.

What a real information point looks like

To see how large the gap really is, compare it with a genuine information point in table tennis. The rule-reform history of this sport is a dense, verifiable dataset.

In 2026, the International Table Tennis Federation (ITTF) decided to increase the ball diameter from 38 millimetres to 40 millimetres, effective October of that year. In 2026, the scoring format changed from 21 points per game to 11. In 2026, the service rule was tightened, requiring the server to keep the ball visible and outside the cover of the body and the free arm. In 2026, the ITTF banned speed glue containing volatile organic compounds, effective 1 September. In 2026, celluloid balls were replaced by plastic balls at major events.

Every one of those lines is a complete information point: a subject, an action, a date, a technical consequence. A player who attacks fast close to the table is affected differently from a player who plays away from the table when ball diameter increases, because a larger ball spins less and travels more slowly, reducing the efficiency of strokes built purely on speed. A player who lives on spin reacts differently to the speed-glue ban, because that glue once allowed faster spin generation over short windows.

Those forty-one pages contain not one sentence at that level. No dates, no subjects, no consequences. Only the frame.

Five signatures of a null return

Across many files of this kind, I have distilled five signatures. They do not replace checking the source, but they separate a weak document from an empty one.

The first is a populated domain label with empty content. A system able to assign a label has read something. But if no entities can be extracted afterwards, the likely explanation is that it only read the shell.

The second is interdependent fields going empty together. When the title, source, article type and core viewpoints are all unresolved, the information-point list is almost certainly empty too. This is a causal chain, not a coincidence.

The third is the survival of instruction lines in their original form. The phrase “identify from the information points above” left inside the file is the trace of a pipeline broken mid-run. Whoever built the frame wrote that line as a reminder to themselves, and it was never executed.

The fourth is an unassessed time-sensitivity field. For a table tennis analysis, failing to anchor to any date is a serious problem, because every judgement about form, ranking and cycles depends on knowing when the document was produced.

The fifth is an unattributed source-quality field. That file stated that source quality should be judged from the source fields of the information points — while those fields were empty. It is a loop with no exit.

When all five signatures appear together, the professional conclusion is clear: this is a null return, not a low-information article.

Pipeline risk is not competitive risk

There is a confusion I encounter often in young analysis teams. People read a file full of “insufficient information to assess” and assume they have discovered a competitive risk. They have not.

Competitive risk belongs to the game itself: an overloaded schedule causing fatigue, a technical overhaul stuck in a trough, fluctuation while adapting to new equipment, a style being decoded by opponents, energy dispersed across too many fronts. To screen for those, you need at minimum a player, a schedule and an equipment change.

Pipeline risk is something else entirely. It is a failed extraction stage, a blocked source, a truncated file, or a model misreading a format. It says nothing about table tennis. It says something about the machine.

The most serious risk that file creates right now is not a threat from any player. It is decision risk: someone reads it, believes they now hold an analysis, and starts acting on the silence of the data. The correct handling is to mark it as a null return, halt the analysis chain for that item, and send it back upstream to be re-run.

Where the transmission map breaks

In industry analysis I always draw the transmission map before writing. For table tennis the diagram has three tiers. Upstream is equipment, youth development and the coaching system. Midstream is events, associations and clubs. Downstream is broadcasting, commerce and derivative markets such as training machines, tracking apps and consumer equipment.

A map like that answers concrete questions. Does a star winning a title lift domestic rubber sales the following quarter? Does a ball-related rule change force academies to repurchase entire ball inventories and adjust their curricula? Does a tournament changing sponsor alter broadcast slots and therefore how coaches arrange training schedules?

That report could not draw a single arrow. No upstream actor, no midstream actor, no downstream actor. No brand, no broadcaster, no host city, no event. A map with no points is not a map.

When the head-to-head file is empty, you cannot talk about form

One of my favourite parts of table tennis analysis is the head-to-head record. It tells stories the ranking table cannot. Some players sit lower on the points list yet beat one specific opponent seven times out of ten, because their style exploits exactly the right weakness. Others sit very high but break whenever they meet a particular type of opponent.

To say any of that, you need a head-to-head table. You need overall win rate, win rate over the last two years, win rate at the three majors, and win rate in deciding games. You need performance against foreign opponents, consistency at major events, and the ability to handle pressure at clutch moments.

In that file the head-to-head table had a single row: insufficient information. Every other cell said the same. That means no player to compare, no ranking to analyse along an age curve, and no multi-event workload data to assess overload risk. In other words, the entire player-data section — the backbone of table tennis analysis — had vanished from the document.

The day I learned that thin data is not empty data

There is a large distance between thin data and empty data, and I learned it during the season played without crowds in Shenzhen.

That year, when the pandemic forced matches behind closed doors, the head coach asked me a question so simple it was hard to answer: does football change when the pressure of a crowd disappears? Together with a colleague I analysed fourteen home matches before and after the pandemic period. Fourteen matches is a very thin sample, and I knew it. But it was not empty.

The results were surprising: with empty stands the team pressed about 23 percent higher and played about 17 percent fewer long balls, because players no longer feared being jeered for losing the ball. I proposed switching to a high press. The club president objected, arguing there was no point pressing without a crowd. I still persuaded the coach to trial it in a friendly, and the team won 4-1 with 71 percent possession. He applied it for the final nine matches and the team climbed from twelfth to seventh.

When the stands are empty, football returns to its original form: a conversation between 22 people.

The professional lesson sits here. A fourteen-match sample is thin, and I said so plainly to the coach. I did not hide the wide error margin. But it was still data: a subject, a period, metrics, before-and-after comparison, a verification result. A thin sample with structure can still save a season. A complete framework with no sample at all saves nothing.

Croatia and the lesson about things you cannot see

In Croatia I learned that a midfield does not run after the ball. It runs after space.

I tell this story not to talk about football, but to talk about how a data gap can be filled the wrong way.

That summer I was paid by a sports outlet to write tactical analysis for a World Cup. The quarter-final between Croatia and the host nation ended level after extra time, and Croatia won on penalties. I lost three nights of sleep. What kept me awake was not the result but the way mainstream analysis explained it: everything converged on a single name in midfield.

I redrew twelve diagrams and found a very different structure. Two wide players repeatedly drifted inside, forming a rectangle in the middle, turning the starting shape into something else in attack. Those movements, not any individual, produced control of the ball.

Croatia had no Zidane, but they had a web of invisible passes.

A tactical wizard is not someone who sees more, but someone who looks where others forgot to look.

The thirty-five-hundred-word piece with twelve hand-drawn diagrams spread widely. A coach in Shenzhen shared it and wrote that this girl understood football better than several men he had worked with. That earned me my first paid analysis job.

But my point lies elsewhere. If there had been no match data that day, no footage, no diagrams, I would not have been allowed to write it. I would have had to return the assignment. And that is exactly what I just did with the forty-one-page report.

The first phone call and the habit of documenting everything

The first call came from a woman nobody names on the coaching bench.

I was twenty then, studying movement science and writing tactical blogs for a local women’s football club. The team lost a match by three goals, but what I saw in the footage was not three conceded goals. I saw a back four and a midfield four distorted into a different structure every time the team lost the ball, exposing the gap between the two centre-backs. I wrote a two-thousand-word piece about it.

The reaction was unpleasant. Many comments attacked me for being a girl. Someone wrote plainly that girls know nothing about pressing.

Then the head coach called me. She said I was right, and invited me to be an unpaid video-analysis assistant.

From then on I formed a habit I still keep: every judgement must come with a specific passage of play or a specific metric. No evidence, no writing. That habit costs me time, but it is also what has kept me from ever having to retract a conclusion.

Thirteen years later, holding a file of four thousand cells all reading insufficient information, I understood how valuable that habit was. It gave me the composure to tell my editor: this cannot be written.

Source grading: the most neglected job of all

There is one stage of the analytical process almost nobody wants to do: source grading. It is not glamorous, it produces no conclusions, and it is usually treated as paperwork.

But it determines almost the entire credibility of everything downstream. Information about a player’s injury from a federation is a different thing from information from an anonymous social account. Information about an equipment change from a manufacturer is different from information from a fan forum. Information about selection criteria from an official document is different from information from an unsourced article.

In that file the source field was unresolved. That means even the most basic step — placing the source into mainstream media, self-media, or fan community — could not be performed. And when the source cannot be graded, every judgement about the reach of the public narrative collapses with it.

I often tell younger colleagues: if you could keep only one field in the whole file, keep the source. Without a source, everything else is decoration.

The contrarian angle: when heat maps become the new astrology

Here I have to say something that may annoy a few colleagues.

In recent years a new generation of visualisation tools has flooded sports analysis. Heat maps, coverage zones, touch-density diagrams. They are beautiful. They make reports look weighty. And they are becoming something like a new astrology.

Why. Because a heat map tells you where a player was, not why they were there or where they should have been. A deep red zone can signal an excellent wide player hugging the touchline, or it can signal a player afraid to leave a safe position while the whole system needs them to move. The same image, two opposite readings, and the heat map cannot adjudicate itself.

More dangerously, heat maps create a sense of completeness. They fill the page. They make readers believe evidence exists, when what they are looking at is a presentation of raw data that has not passed through tactical interpretation. That is exactly the problem of the forty-one-page file, only in a more extreme form: complete form, empty content.

A nine-layer analytical framework can look very professional. A vivid heat map can look very persuasive. Both can conceal the fact that nobody really knows what is happening inside a team’s tactical system.

My argument is not to discard visualisation. It is to ask a question before believing it. Is this player in that position because of a tactical task, because of habit, or because the opponent forced them there? If you cannot answer, the map has not yet said anything.

The final trap: the market prefers a story to an N/A

There is one pressure I should name plainly, because it is the biggest reason null returns rarely get published.

The market does not like emptiness. Readers do not open a sports article to be told there is not enough data. Sponsors do not pay for a report full of insufficient information. Editors need headlines. Algorithms need content. And in that machinery, one “no” becomes a story, a story becomes a conclusion, and a conclusion becomes an action.

I once saw a case like this. An analysis of a player was missing the entire head-to-head section. Instead of leaving it blank, the writer filled it with a sentimental description: this player has nerves of steel at decisive moments. Nobody verified it. Three weeks later, when that player lost a deciding game, the sentimental line was dragged out as evidence for the exact opposite conclusion.

The price of filling gaps with sentiment is not one wrong judgement. It is a whole chain of beliefs built on sand, and when the sand shifts, trust in the entire analysis trade shifts with it.

So when my intern proposed filling the report with a few stories about famous names, I understood why. Big names generate enormous search volume. An article mentioning them will almost certainly get more reads than one saying there is not enough data.

But I still said no. Because if I filled it with those names, I would no longer be an analyst. I would be a storyteller, and a storyteller is not allowed to sign a data report.

ENFP in the analysis room: finding inspiration in the driest numbers.

My joy is not in having a thick report. It is in the moment a seemingly meaningless metric suddenly explains a decision on the coaching bench. That moment does not come from filling gaps. It comes from patiently waiting for the gap to speak for itself.

What I sent back, and what I am waiting for next match

I returned the report to the analysis unit with three lines of notes: re-run the extraction stage for this item; check whether the source is blocked, deleted or truncated; and if the source cannot be recovered, close the item as an official null return.

I added one line for myself: never mistake completeness of form for completeness of content.

The nine-layer framework in that file is not wrong. It simply had nothing to say yet. If the source is recovered, all nine layers must be re-run from scratch, and no conclusion from the empty version may be carried over. If it cannot be recovered, the most honest thing a professional can do is say they do not yet know.

There is one small detail I keep in mind. When extraction works and the file comes back with real data, the first thing I will do is not write conclusions. I will go looking at the edges of the frame: unnamed players, uncounted gaps on the table, rallies that never appear in the statistics table.

Because complete data was never the destination. It is only the departure point.

And the next match will answer.