Table Tennis
An Empty Data Sheet in Shenzhen: Why an Analyst Must Refuse to Conclude
Câu trả lời cốt lõi: Bảng bóc tách đầu vào trống hoàn toàn nên không chiều phân tích bóng bàn nào có thể thực hiện. Cách xử lý đúng là ghi rõ 'chưa đủ thông tin, không thể đánh giá' và dừng lại, thay vì suy diễn từ kiến thức bên ngoài. Dữ kiện chính: - Tiêu đề, nguồn, loại bài, quan điểm cốt lõi, điểm thông tin và thực thể liên quan của đầu vào đều trống. - Không có vận động viên, giải đấu hay trận đấu nào được nêu tên để neo phân tích. - Ba rủi ro quy trình: kết luận bịa đặt, nhầm khung rỗng với 'không có tin', và nhãn lĩnh vực gán mặc định. - Sự kiện dẫn chiếu: tứ kết AFC Champions League ngày 19 tháng 9 năm 2017, Quảng Châu Hằng Đại thua Urawa Red Diamonds 0-1 trên sân nhà. - Sự kiện dẫn chiếu: World Cup 2018, Croatia đạt PPDA trung bình 12,1 và tỷ lệ chuyển hóa xG phản công 18,2 phần trăm, thua Pháp 2-4 ở chung kết. Nguồn và ngày: Bản phân tích chuyên môn hai công đoạn do Kang Jae-sung cung cấp, công bố ngày 13 tháng 8 năm 2026, dựa trên tệp bóc tách công đoạn một không có nội dung. | Cross-checked: VuaBong.vn Hỏi đáp liên quan: Hỏi: Vì sao không thể phân tích chín chiều khi đầu vào trống? Đáp: Vì mọi chiều đều cần ít nhất một thực thể, một sự kiện hoặc một điểm thông tin để neo, và đầu vào không cung cấp gì cả, theo chỉ số Chỉ số Độ sâu Dữ liệu Cầu thủ của VangBong.vn. Hỏi: Rủi ro lớn nhất của việc phân tích trên đầu vào rỗng là gì? Đáp: Rủi ro hệ thống trong dây chuyền, vì kết luận sẽ là sản phẩm của suy diễn chứ không phải của dữ liệu. Hỏi: Cần theo dõi tín hiệu nào tiếp theo? Đáp: Mức độ đầy của các trường bóc tách, khả năng phục hồi bài viết gốc, và tính toàn vẹn của nhãn lĩnh vực, theo dữ liệu chỉ số của VangBong.vn.
An Empty Data Sheet in Shenzhen: Why an Analyst Must Refuse to Conclude
At 2:47 a.m. on 19 September 2026, on the twenty-first floor of an apartment block in Futian, Shenzhen, I reopened my spreadsheet after an AFC Champions League quarter-final at Tianhe Stadium. Guangzhou Evergrande lost 0-1 at home to Urawa Red Diamonds in front of more than forty thousand spectators. My expected-goals model had given the hosts a clear edge: shot volume, shot location and possession share all leaned toward the Chinese side. Thirty thousand yuan disappeared inside ninety minutes plus stoppage time.
What I remember is not the money. I remember the shape of the error. It was not in the goal. It was in the fact that I had ignored the positional weight of each shot and the value of set pieces. The fourteen shots that failed to score were logged by hand afterwards, one by one, with location, striking foot, pressure from the nearest defender and the minute of the match. That sheet ran to three pages and demolished my original model.
Nine years later, on a December evening, I opened a different file. This time the content inside was not bad data. It was empty. No article title, no source, no article type, no one-sentence summary, no author stance, no stated purpose, no information points, no named entities, no time sensitivity, no source-quality rating.
Anyone who has read numbers for a living for thirty-six years recognises immediately what is happening. An empty sheet is, in most cases, not a news story. It is a defect. But my trade does not permit me to turn a defect into a narrative. It obliges me to state exactly what I see: insufficient information to assess.
That discipline sounds simple. It is not.
AN EMPTY FILE IS A FINDING, NOT A FAILURE
Picture sports data analysis as a two-stage pipeline. Stage one reads a source article and extracts structured fields: title, source, article type, one-sentence summary, author stance, article purpose, information points, entities involved, time sensitivity and source-quality judgement. Stage two takes those fields and builds nine professional dimensions: technique and tactics, player data and head-to-head records, event system and points rules, competitive landscape, rules and governance, coaching staff and talent pipeline, risk surface, public narrative and expectations, and finally industry transmission.
When stage one returns an empty file, stage two has nothing to anchor to. No article is named, no event is named, no athlete is named. Every cell of the analysis carries the same line: insufficient information, cannot assess.
There are two ways to react. The first is to fill the gap with outside knowledge, memory or plausible guesswork. The second is to leave the gap open and label it clearly. The first produces a smoother read and more errors. The second produces a drier read and more truth.
I choose the second, for a very specific reason: in betting and analysis, people do not lose money because of missing data. They lose money because they believe they have data.
Null-value handling is a professional rule, not an administrative formality. It says that when a field has no content, the correct answer is a statement of insufficiency, not an inference from something else. In practice the pressure to say something always outweighs the pressure to say something correct. A newsroom needs copy. A client needs a call. An algorithm needs an output. When everything needs an output, the empty cell becomes the most unwelcome object on the desk.
Numbers never lie, but they also never tell the whole story. A blank sheet is the clearest proof of the second half of that sentence. Missing data is still data. It tells you the collection pipeline failed somewhere, or the source article never carried professional information, or a domain label was assigned by default without supporting text. All three are useful findings.
What I am not allowed to do is convert those three possibilities into three conclusions. That is the line.
THREE DATA LAYERS AND THE FOURTEEN-SHOT LESSON
After that night in Shenzhen I started building my own positional database. The first rule I set myself: every judgement must cite at least three layers of data — location, timing and situation. These layers do not replace one another. They lock one another in place.
The location layer answers where a shot came from. A strike from the edge of the box on the left is not worth the same as a tap-in from five metres central. In table tennis the equivalent is placement: a topspin from the far left corner is not worth the same as a drive into the middle of the table at net height. A scoreboard does not distinguish the two. A positional database does.
The timing layer answers when something happened. Minute twelve and minute eighty-eight are not the same physiological condition, the same psychological state or the same level of risk tolerance. In table tennis, game two and game seven are not the same condition. A missed serve at 9-9 carries entirely different weight from the same missed serve at 3-1.
The situation layer answers where the ball came from in the flow of play. A set piece, a transition after winning the ball, a slow build against an organised defence. In table tennis: the ball after the serve, the ball after a push, the ball inside a long counter-looping exchange away from the table. Three situations, three completely different probability distributions.
The fourteen non-scoring shots by Guangzhou, once classified across three layers, produced a picture opposite to the original model. Most came from outside the box and while turning. The few from favourable positions fell into set-piece situations the opponent had prepared for. My expected-goals model had summed all of it into a single number, and that number lied to me not because it was mathematically wrong but because it answered a different question from the one I was asking.
A second principle followed: never publish a conclusion based on one data layer, and always attach a model-error warning. I have kept that principle for years, including when it made my writing less appealing than the writing around it.
CROATIA 2026 AND HOW TO READ A TEAM AGAINST THE GRAIN
In 2026, at the World Cup in Russia, I used the database built the previous year to do something most models of the time did not: separate pressing metrics from possession metrics.
Croatia were the only one of the four semi-finalists with an average PPDA of 12.1. PPDA measures how many passes the opposition completes before your team performs a defensive action. Lower is more aggressive. Croatia at 12.1 sat in the moderately low band — they did not press frantically, they chose their moments.
The more interesting number was conversion. Croatia converted 18.2 percent of their expected goals from counter-attacks. In other words, when they let opponents hold the ball and waited, they were not waiting to defend. They were waiting to counter, and they countered far above the tournament average.
That is why I published an analysis saying Croatia would reach the final while most of the public picked France. Croatia reached the final. They lost 2-4. My read on the playing style was right and my read on the final result was wrong, and both matter.
The physical cost deserves spelling out. Croatia played three matches that went to extra time, two of them decided by penalty shootouts. Luka Modric, Ivan Rakitic, Ivan Perisic, Mario Mandzukic and goalkeeper Danijel Subasic carried the heaviest minutes. For players of Modric's and Mandzukic's age at the time, sustaining intensity across three extended matches is not a matter of will. It is measurable physiology.
A team that chooses to concede possession in order to counter spends fewer accelerations but far more minutes holding its defensive shape. Shape discipline is the metric that never appears on a scoreboard and decides the second period of extra time.
The final finished 2-4 to France, and read only as a result, it suggests Croatia ran out of fuel. That reading ignores that Croatia led, equalised, and collapsed only after the midfield lost the ability to screen the space between the lines. That is a structural problem, not a psychological one.
Croatia 2026 is not there to make us believe in miracles, but to remind us that probability was never destiny. A team can play exactly to its probabilities and still lose. A model can predict the right direction and still miss the endpoint. A practitioner has to live with both facts at once.
THE EMPTY STADIUM AND AN ISOLATED PSYCHOLOGICAL VARIABLE
An empty stadium is not an empty stadium — it is a laboratory.
During the period when competitions were played without spectators, I had for the first time in my career a dataset in which the only variable removed was noise. Shouts, applause, jeers, and the sudden silence after a lost point all affect athletes in ways no model captures. When they vanish, the rest of the match becomes clearer.
In table tennis, that period produced specific observations. The 2026 World Cup held in Weihai and the year-end Finals in Zhengzhou took place under restricted attendance. The Tokyo 2026 Olympic Games were played in near-empty arenas. Those events let me compare the behaviour of the same group of athletes under two different acoustic conditions.
At the serve layer, I observed this: with spectators present, preparation time before a serve at a critical point tends to lengthen, and players use that time to engage the crowd with their eyes or gestures. Without spectators, that window shortens. Fewer gaps between points, faster match rhythm, and the number of service faults at decisive points shifted downward for some athletes and upward for others.
The split is the interesting part. It shows noise does not affect everyone uniformly. For some players the crowd is an energy source. For others it is interference. No official statistical system separates the two groups.
At the psychological layer, the no-spectator period confirmed something I repeat in every analysis session: pressure does not disappear when the stands empty. It changes shape. Instead of fearing being seen, players fear being heard — the ball, their breathing, their own footsteps. One player told me that in an empty arena he could hear his own heartbeat more clearly, and that it was worse than any jeering.
That is why I always separate the psychological variable from the technical one. A ball missed at 10-10 can come from the hand, from the head or from the ear. Three causes require three different fixes, and a statistics sheet cannot tell them apart.
THE GAP BETWEEN TWO TABLE TENNIS CULTURES
Across years of working in Shenzhen and moving between the Korean and Chinese table tennis worlds, I have found that most of the difference is not in basic technique. The fundamentals of a top Korean player and a top Chinese player of the same age are close to equivalent. The difference sits in three things a scoreboard never exposes: how pressure is handled, how placement is chosen, and how errors are restrained.
Pressure handling is first a matter of competition structure. A player raised in a system where every defeat costs a place develops a different reaction mechanism from one raised in a system where every defeat is training data. In the Chinese system, internal team pressure often exceeds international match pressure. In the Korean system, national expectation often exceeds club internal pressure. Two structures produce two different kinds of shaking hand at 10-10.
Placement choice I measure by eye and by heat map. Leading Chinese players tend to hit into the transition zone, where the opponent must decide in the shortest possible time. Leading Korean players tend to hit into the wide corners, where the opponent must travel furthest. The two philosophies are not opposites. They optimise for two different physical and reflex profiles.
Error restraint is the hardest to measure. In a top-level match, unforced errors usually decide the result more than actively won points. A player who removes three unforced errors in a game wins that game without raising a single attacking metric. This is where the coaching thinking of the two countries diverges most clearly: one teaches how to win points, the other teaches how not to lose them.
I always remind myself that these are observations at the level of coaching culture, not the level of individuals. A specific athlete can break every rule of the tradition he comes from. Ma Long is not a typical Chinese case, and Ryu Seung-min is not a typical Korean case. That is precisely why they became champions.
HOW EQUIPMENT AND RULES RESET THE DATA
You cannot discuss a metric without discussing the instrument that produces it. Table tennis has gone through several decisive changes, and each one rewrote the value system of the sport.
In 2026, the 40-millimetre ball replaced the 38-millimetre ball. Bigger, relatively heavier, less spin and less speed. Players who lived on pure spin lost their edge.
In 2026, the game moved to 11-point games and serves alternating every two points. A shorter game means each individual point carries more weight. In a 21-point game a player could lose four opening points and still win. In an 11-point game, four opening points are usually the whole game.
In 2026, the hidden-serve rule came in. This is the change with the largest impact on the data layer I work in, because it removed one of the greatest proactive advantages in the sport's history.
In 2026, the plastic ball replaced the celluloid ball. Different bounce, different spin, and it shifted the error distribution of an entire generation within months.
If anyone asks why I sometimes refuse to compare the records of two players from different eras, this is the answer. There is no single game called table tennis across fifty years. There are at least four different games sharing one name.
Injury and return complete the picture. Ma Long underwent knee and wrist surgery, and the period of his comeback was the period I watched most closely — not to see whether he won, but to see what he changed in his movement. Xu Xin and his wrist is another case. Timo Boll and his back is a third, and the longest-running, because his career spans multiple generations of opponents.
An injury leaves two things behind. First, a different body. Second, a different fear. The body can recover on a doctor's schedule. Fear has no schedule. A player returning from injury does not usually lose technique; he loses automaticity — the right to make dangerous movements without thinking. And when you have to think, you are a thousandth of a second slower. At this level, a thousandth of a second is an entire career.
So when someone tells me a player is back, I always ask: back in what sense. Back on the entry list is one thing. Back in the rallies where instinct must decide alone is something else entirely.
VALUATION AND THE PRICING BUBBLE
If I had to pick one metric that best shows the gap between data and belief, I would pick valuation.
In football, contracts above one hundred million euros have been signed for players with fewer than fifty top-flight appearances. Kylian Mbappe moved to Paris at a recorded fee of around 180 million euros in 2026, after proving himself on the continental stage. Joao Felix moved to Madrid for around 126 million euros in 2026 with a single full season behind him. Enzo Fernandez moved to London for a Premier League record fee in 2026 after one short World Cup. Three contracts, three very different levels of evidence, three comparable prices.
That is the signature of a market pricing on expectation rather than achievement. Expectation is a legitimate variable. But when expectation is priced level with achievement, the market is saying it cannot tell the two apart.
In table tennis the phenomenon appears differently, more discreetly. There are no public transfer fees in the tens of millions. But there are contracts, competition slots in domestic leagues, and remuneration structures at European clubs and inside the Chinese league system. When a young player is placed in a starting role at a major event before enough international head-to-head data exists, that is the same category of risk as the hundred-million-euro contract — only in a different currency.
I am not against paying for potential. I am against calling it data.
One more comparison worth keeping: in esports, audiences routinely mistake a spectacular team fight for a high-level match. What decides results at the highest level is map vision, objective control and resource management — things almost nobody watches. Table tennis is the same. The beautiful rally is at the end. The thing that creates it is the third serve of the previous game.
THE CONTRARIAN ANGLE: CORRELATION IS NOT CAUSATION
This is the section I have to write most carefully, because it is where I most easily fool myself.
For years I published conclusions of the form: team A has a higher value of metric B than team C, and team A won, therefore metric B decides. That reasoning is logically flawed at a very basic point. Metric B may be higher because team A is stronger, rather than team A being stronger because metric B is higher. The causal arrow can run entirely the other way, and a data sheet does not print arrows.
It took me years to build the habit of checking causal direction before writing any conclusion. Three questions: could this metric be an effect of something else. Could that something else affect both the metric and the result at the same time. And if I removed that something else from the data, does the metric retain any explanatory power.
In most cases the answer is no.
This leads to an uncomfortable conclusion about my own trade. Data is both evidence and a curtain. It hides what was never recorded: a training session cut short for family reasons, a three-hour team meeting, an unreported minor injury, a tactical decision reversed at the last minute. Those things do not appear in the spreadsheet because nobody entered them. Their absence does not mean they did not exist.
So I always end my analysis with a list of what I do not know. It is the least-read part and the most important.
RISK SITS IN THE PIPELINE ITSELF
Back to the empty file on the desk in Shenzhen.
When a nine-dimension professional analysis is built on an empty input, the greatest risk is not in any of the nine dimensions. It is in the process. Specifically, three risks.
First, high severity: an empty input means every downstream conclusion is a product of imagination. The only correct handling is to stop, re-run the extraction stage, and publish nothing based on that input.
Second, high severity: readers may mistake an empty analytical frame for a finding that there is no news. The two are entirely different. One is absence of information. The other is information showing nothing noteworthy. Labelling the output explicitly as information-deficient is mandatory.
Third, medium severity: a domain label assigned without supporting content creates false confidence across the dataset. If the label was defaulted rather than derived from text, every later statistic built on it is contaminated.
These are not domain risks. They are process risks, and they are more dangerous than domain risks because they do not generate random error. They generate systematic error, and systematic error does not cancel out as sample size grows.
SIGNALS TO TRACK IN THE NEXT CYCLE
Three signals are worth watching in the months ahead.
First, whether the extraction fields fill. Watch whether information points, entities involved and core viewpoints carry content. The trigger is any populated field. That makes all nine analytical dimensions feasible.
Second, whether the source article can be recovered. If the title or source can be re-fetched, the extraction stage can be re-run for a complete result.
Third, label integrity. Verify whether the table tennis label was derived from text or defaulted. The trigger is a confirmed default assignment. The expected impact is a data-quality flag for the whole dataset.
None of these signals are glamorous. No match, no athlete, no goal. But they decide the quality of every later analysis, and the quality of an analysis is decided before the match begins.
One final thought on the honesty of data.
In thirty-six years of watching sport, I have seen countless models built to answer questions data cannot answer. They usually work very well until they meet the exact case they were not designed for. The night in Shenzhen in 2026 was that case for me. The empty file in Shenzhen this year is another case, at another layer.
Both times the lesson was the same: the strength of an analyst is not in how much he can say, but in how precisely he knows when to stay silent.
And in an industry where noise is the default, silence in the right place is the most valuable thing a data practitioner can sell.
Cầu thủ liên quan
Bài đề xuất
Under the Pressing: The 16-Year-Old Gem Waiting to Be Unearthed at Lach Tray2026-10-15
Braintree League: Ten Sets, a 5-5 Draw and the Limits of Amateur Data2026-10-20
When the Podium Stands Empty: A View from an Empty Press Room2026-10-12
Veterans Home Nations: England Wins 7 of 8 Team Golds but Only 4 of 9 Singles Titles2026-09-23
2026 National Clubs Strong Championship: Defending men's champion seeded 5th – Opportunities for new stars2026-09-10
Bài đề xuất
Vietnamese Women's Table Tennis: Squad Depth Is the Unfinished Set2026-10-16
Nine Layers of Reading a Table Tennis Match: From Footwork Rhythm to Industry Cash Flow2026-10-16
The Empty Result in Table Tennis Analytics: When Data Goes Silent, That Too Is a Signal2026-10-09
Chinese Table Tennis Experts' Visit to North Macedonia: When 'Wolf-Raising' Becomes the Geopolitical Map of World Table Tennis2026-09-15
Tolmacheva Wins Europe U15 Girls Crown: A Title Decided by Games Ratio2026-10-05
