Null Results: When the Table Tennis Data Sheet Refuses to Speak

On the night of March 15, 2026, at Phu Tho Arena, I sat in front of three monitors and stared at the data sheet for the national table tennis championship semifinal. The club I serve as data consultant had just won 3-2 against a strong opponent from Da Nang. The stands applauded. The coaching staff shook hands warmly. I stayed quiet. Because the data sheet in front of me was empty. Not almost empty in the sense of missing a few scattered data points. Not a temporary glitch requiring a restart. Truly empty. Three independent collection systems returned the same thing: nothing. Verified metrics equal to zero. Players logged equal to zero. Direct service-point rate did not exist. In 36 years of observing the industry, I have learned to name this phenomenon: the null result. And the most interesting thing about my profession turned out to be not when the data sheet dances. It is when it goes silent. I joined the sports writing and data analysis profession in 2026, when I began working for the Daily Mail, contributing for a full 25 years. In 2026 I hosted broadcasts of several major events including the Table Tennis World Cup and the Sudirman Cup in badminton. In 2026 I began serving as a data consultant for a club in Ho Chi Minh City; that year a 1,500-word analysis of V.League PPDA drew public attention because it showed that our club's winning streak came from luck more than from structure. All those stages taught me something few in the industry want to hear: most of the data we collect, in a great many matches, says nothing at all. The majority in the analytics field are often haunted by the feeling that they must always have a conclusion. Every match must yield a finding. Every data sheet must produce a story. If not, the analyst feels useless. That is an extremely dangerous pressure, because it turns the data worker into a fabricator of data. In table tennis, where I invest most of my professional time, this problem is even more serious than in football. A singles table tennis match lasts on average 30 to 45 minutes with about 4 to 5 games. The number of ball contacts per game typically ranges from 60 to 120. But the number of points that can genuinely be analyzed independently in tactical terms is far smaller, sometimes only 8 to 12 points across an entire match. What does that mean? It means the sample is too small to support firm conclusions, unless you collect data across dozens of consecutive matches. And that is exactly when the null result appears. Not because the match has nothing to say. But because we have not yet collected enough for it to speak. Let me tell three stories about null results. The first comes from the 2026 season. I was tracking a young player in a northern club's U21 squad, call him T. T's coach sent me video of six consecutive matches and asked me to analyze why the player had lost four of six. I received the files, opened them, and immediately saw a problem: the video had no close-up angle on the player's racket hand, no angle from behind the opponent, and the audio had been muted to avoid copyright. Which meant I could not evaluate the spin serve, the very thing all six matches revolved around. I told the coach honestly: with this material I cannot reach a conclusion. He was disappointed. He said: but you are an expert, you must be able to say something. I said: an expert is not someone who invents conclusions from data that does not exist. An expert is someone who knows when to say I do not know. This is a very small story. But it represents hundreds of similar situations I have encountered over more than three decades of work. What is striking is that none of the people who asked me to analyze ever came back afterward to say they should have prepared better material. Instead, most of them moved on to another analyst willing to produce conclusions from a six-match sample missing camera angles. The second story comes from that very night of March 15, 2026, which I described at the opening. After three systems returned a null result, I had two choices. One was to write a short report to the coaching staff saying the data was insufficient and asking for recolleciton. Two was to try to fill the gap with inference from video, memory and feeling. I chose the first. The coaching staff replied that they needed something more concrete. I answered that precisely because they needed something concrete, I was not permitted to fabricate. A week later we discovered why all three systems were empty: cameras were placed at the wrong angle and nobody had noticed. The lesson here is not that it turned out to be just a technical fault. The lesson is: if I had fabricated a conclusion from empty data, I would never have discovered the camera fault. The null result is itself a signal. The third story comes from 2026, when I worked with a young table tennis national squad preparing for a regional tournament. We had very complete data, more than 40 matches recorded meticulously. But when we ran the statistical models, the correlation between the PPDA metric (passes the opponent completes before your side regains the ball, borrowed from football but adapted to table tennis) and win rate came out at just 0.09, essentially no correlation. I could have ignored that metric and moved to another. But instead I stopped and asked: why is it so? The answer turned out to be fascinating. In table tennis, PPDA does not reflect defensive strength as in football, because table tennis has an entirely different service structure. A skilled player may deliberately concede the service initiative to gain advantage on the third ball. Which means high PPDA does not mean weak defense; it may mean proactive tactics. The null result here was not nothing. It was that what we are measuring does not measure what we think it measures. And here is where I must say plainly what the majority in Vietnam's sports analytics field do not want to hear: the compulsion to always have a conclusion has become an occupational disease. It manifests in three ways. First, analysts tend to select convenient data samples. They skip matches that do not yield pretty results and analyze only the matches that produce compelling narratives. This is a sophisticated form of data fraud, and it is so common no one calls it fraud anymore. Second, analysts tend to turn correlation into causation. If metric A is high and the team wins, they say metric A causes the victory. The majority cannot distinguish between two things happening together and one thing causing the other. In table tennis this is especially dangerous because small samples make every correlation seem stronger than it is. Third, analysts tend to turn null results into meaningful results by adding words. Trend, likelihood, appears to be, notably. These phrases add no information; they add only counterfeit confidence. Old recordings are a mirror; only those who dare to look see themselves. I reviewed hundreds of analytical reports I wrote during 2026-2026 and realized that about 30 percent contained conclusions not supported by actual data. I did not fabricate deliberately. I was simply under pressure to have an answer. That is an uncomfortable truth to admit at 52. But if I do not admit it, I will repeat it. My data cafe is busiest when the stadium is empty. When everyone is talking about the match just finished, I sit alone with the data sheet and wait. Numbers know how to hold their breath, and I wait for them to exhale. So how do you recognize when a null result is truly occurring, rather than mere intellectual laziness? I have three principles. Principle one: check the sample before checking the conclusion. If the sample is under 20 matches for a specific metric type, be careful with any conclusion you draw. If it is under 10, consider that you have nothing at all. Principle two: distinguish between data that does not exist and data that exists but does not correlate. These two situations demand entirely different handling. If data does not exist, you need to collect again. If data exists but does not correlate, you need to question your very method of measurement. Principle three: name the silence. In my reports I always include a section titled What Cannot Yet Be Concluded. This section matters no less than the conclusions. It forces me to point out exactly where my reasoning still lacks data. The crowd watches the score; I watch the forgotten pass. In table tennis, the forgotten pass is usually a serve returned in a boring manner, so boring that the camera does not bother with a close-up. But those boring serves often carry the most important tactical information. Since 2026 I have applied a method I call three-layer analysis. Layer one is raw data, what you can count: points, games, service-point rate. Layer two is interpretive data, what you can infer from video: positions, movement direction, tempo. Layer three is contextual data, what you can only understand if you understand the circumstances: psychology, head-to-head history, playing conditions. Null results usually appear at layer two. At layer one, data is always available because you can count anything. At layer three, data is always fuzzy because people are complex. But layer two is the layer many analysts skip, because it demands both video-reading technique and patience. There is one example I will never forget. In 2026 a club asked me why a player they had signed was failing to meet expectations in his first 5 matches. The leadership wanted a concrete conclusion to decide whether to let him go. I took the data, worked for two weeks, and concluded that I could not conclude. The reason was simple: 5 matches, 12 games, about 180 points. Statistically, that is too small a sample to distinguish poor form from normal fluctuation. These two possibilities demand entirely different responses, but a small sample does not allow distinguishing them. The leadership was unhappy. They said: we do not pay to hear cannot conclude. I replied: if I produce a conclusion from a 5-match sample, you will use it to make a signing decision. That decision would rest on a number I myself know is insufficient. That is far worse than hearing cannot conclude. By season's end, this player played 22 more matches and won 15. The null result had been confirmed as a null result. But there is something more worth pondering here. If the leadership had decided based on 5 matches and let him go, the story would forever close with a wrong conclusion, and all of us would believe that conclusion was right. This is what I call the truth manufactured by a small sample. It is more dangerous than fake data, because it looks right. Every number is a puzzle piece, but I do not assemble by habit. The majority's assembly habit is to assemble in the direction the story has already been written. Mine, after 36 years, is this: if the story is too smooth, that is a sign I am fabricating it myself. In Vietnamese professional table tennis this pressure is especially evident at club level. A team competing in the national championship typically has 15 to 20 athletes across age groups. Each season each athlete plays only 10 to 25 matches depending on role. If a player plays 12 matches, their data will hold about 30 games and 400 points. This sounds like a large number, but in reality it is not enough to distinguish an important technical change from natural form fluctuation. Modern statistical models require a minimum sample of 50 to 100 games to reach reliability above 90 percent. With a national table tennis system of only a few dozen senior athletes, reaching such a sample requires accumulating data across many consecutive seasons. Almost no club in Vietnam maintains data collection discipline at that level across seasons. That is why most club-level analysis in Vietnam, however beautifully presented with charts and detailed tables, still carries a hidden null result inside. Readers do not see it because it is masked by confident language. If you are a club leader reading this, there is one question I recommend asking every data analyst you hire: in this report, where is the section on What Cannot Yet Be Concluded? If they point to an empty section or there is none, you are not receiving analysis. You are receiving reassurance. On that night of March 15, 2026, after writing the data-insufficient report, I stayed at Phu Tho Arena another two hours. Not to do anything. Just to sit with the silence of the data sheet. And in that silence I realized something: my club had won 3-2, but that victory was supported by no measurable metric. That does not mean the victory had no reason. It means the reason lay in a layer of data we do not yet know how to measure. If you work in sports analytics, and every data sheet you produce has a clear conclusion, ask yourself: are you seeking data, or seeking comfort? A null result is not the failure of analysis. It is proof that the analysis is being honest. And in an environment where everyone wants an answer within 24 hours, honesty may be the most counter-intuitive thing. I once feared the microphone; now I let the data speak for me. And when the data chooses silence, I learn to be silent along with it.