Small Samples and the Myth of 'Steel Nerves' in Elite Table Tennis

One morning in Guangzhou, I opened my spreadsheet and found a blank page. Not because I had forgotten to save. The raw data feed for a match I intended to dissect had come back empty: no point-by-point scores, no serve statistics, no rally-win rates. Just empty cells stretching out like a deserted grandstand.

My first reaction was not "there is no data yet." It was a familiar temptation: to fill the gaps with inference. The table tennis world does this every day. A player wins three tight matches, and a legend of "nerves of steel" is born instantly. A player loses two in a row, and a form crisis is staged across the news sites. The biggest trap in this profession is not reading a number wrong. It is reading a number that does not exist at all.

Numbers do not lie, but the people who read them do.

Take a recent example. At the Paris 2026 Olympics, Wang Chuqin — then world number one in the ITTF men's singles ranking — fell to Truls Moregard in the knockout rounds. Within hours, a story spread across every forum: he was a player with "weak mentality," who "could not handle big-stage pressure." An entire career of hundreds of wins on the WTT circuit was compressed into a single afternoon.

I do not deny that loss was real. I simply ask one question: what was the sample size behind that conclusion? To call a player "mentally weak," you need to measure him across many tense matches, not one. That is the point nearly every table tennis analysis skips.

A sport that lives on small samples

Picture one season for a top player. He plays a few dozen matches on the WTT circuit, plus continental and world championships. That sounds like a lot. But if you care only about the tense moments — deciding games, match points — the number is genuinely tiny. A player might experience only ten or twelve such situations in an entire year.

Ten observations. That is all you have to build a legend of nerve.

With ten trials, the confidence interval on any conclusion is so wide as to be useless. If a player wins seven of ten deciding games, he is instantly crowned a "big-match king." But statistically, that rate is perfectly compatible with a player whose baseline win ability is only 55%. In other words, most of what is called "nerves of steel" is just an ordinary probability sequence projected onto a big screen.

What is frightening is not wrong data, but correct data that is too little to tell a story.

This is where I borrow thinking from football. In football, people use xG — expected goals — to separate outcome from process. Table tennis has no such universal standard, but it has equivalent and even stronger metrics: rally-win rate, serve-win rate, and first-three-ball win rate.

First-three-ball win rate is table tennis's xG. It is not a flashy yardstick; it is the confession of the match.

When you look at the first three balls, you see what the scoreboard hides. A player who wins 4-2 may have been thoroughly beaten in long rallies, yet won thanks to tricky serves at the right moments. Conversely, a player who loses 2-4 may have dominated in rally quality, only to squander the decisive points. The result is obvious; the process is the truth.

Based on my experience watching matches, I have noticed a fairly durable rule: when a player wins deciding games in a row, the media praises his spirit. But when I check his first-three-ball data, usually nothing has changed from before. It is simply that the 50-50 points fell to one side. And 50-50 points, over time, always revert to the mean.

If the rally-win rate does not change, then a run of deciding-game wins is variance, not nerve.

In table tennis, the most durable legends are not the ones with the most miraculous moments, but the ones who rarely need them. Fan Zhendong won men's singles gold at Paris 2026 not through a divine shot, but through a high, stable rally-win rate across the whole tournament. Sun Yingsha holds the women's singles number one spot not because she saves more match points than others, but because she rarely lets matches reach match point. That is the difference between a great player and a remembered one.

When the stands are empty, I see the truest player

In 2026, when the pandemic turned arenas into empty grandstands, I collected data from hundreds of international matches. In football, home advantage all but evaporated: the home-team win rate dropped sharply. In table tennis, the crowd effect is far smaller, because the table is an almost closed space — applause cannot change the trajectory of a serve.

But the pandemic data revealed another variable, more important and less discussed: schedule density. When tournaments were compressed and players had to play many matches in a short time, what decided matters was no longer "nerve" but recovery capacity and stability of movement. Players with compact technical foundations and low energy expenditure suddenly rose. Stars who lived on inspiration declined.

The ranking is the summary; the raw data is the testimony.

That is why I never start an analysis from the ranking. A ranking is the output of a complex points system, shaped by who enters which events, how many points they defend, and which events they skip. It is like reading the summary at the back of a book and thinking you understand the plot. To understand a player, you must read his every rally, not the number next to his name.

Take an illustrative model. Suppose a player has a baseline rally-win rate of 53% — a very good figure at the elite level. If that player faces a run of even points, the probability he wins three decisive points in a row is only around 15%. But if it happens once, the media calls it "a champion's moment." The truth is: with hundreds of players and thousands of point sequences each season, such 15% streaks will appear constantly. Everyone is a champion sometimes. The question is who can repeat it.

And here is the hardest part: data does not teach us who will win. It only teaches us who is lucky right now.

The noise does not come only from the court

In recent years, table tennis has begun to have something like a transfer window. Players move to play in foreign national leagues, sign with clubs, and change personal coaches. And with that comes a ceaseless stream of rumors: who is about to leave, who is about to join, who is unhappy.

The problem is that this noise distorts the very data we are trying to read. When a player is rumored to be moving clubs, people start interpreting every loss as a sign of distraction. That is a causal inference drawn from correlation, and it violates the most basic principle of the profession.

I learned this from my own mistake. Years ago, I attributed a poor run of form to an off-court event, simply because the two coincided in time. Later I checked again and realized the poor run had begun before that event. Correlation is not causation — and in table tennis, where the sample is already small, it is even easier to mistake one for the other.

The new generation and the successor's trap

Whenever a young player like Lin Shidong wins a few titles in a row, a wave of commentary appears: "a new dynasty has begun," "the old generation is finished." I understand the appeal of that story. But it too is a small-sample conclusion.

A young player can win a big title because he peaked at exactly the right time, because opponents had fitness problems, or simply because the draw was favorable. To claim a genuine generational handover, you need to see it repeat across many tournaments, many tables, many different opponents. One season says nothing. Two seasons start to mean something. Three seasons become credible.

That is why I always question sustainability before offering praise. Not because I doubt the new. But because I have seen too many "golden generations" dissolve within six months.

The contrarian angle: when people correct a mistake with another mistake

A backlash is spreading, and I want to warn about it. After the data-analysis cohort showed that "nerves of steel" is largely noise, some fans and even some young analysts went too far: they declared that mentality does not exist, that every match is pure mathematics. That is another mistake, and no less dangerous.

Mentality exists. It simply is not where people look for it. It is not in the win-loss outcome of deciding games — a thing muddied by too many random variables. It is in the quality of decision-making under pressure: which serve a player chooses when trailing, whether he dares to change tactics mid-game, whether he can hold the length and spin of his serve while his heart races.

That is the measurable variable. That is what I call "pressure data" — not the deciding-game win rate, but the degree of change in technical choices between a normal state and a state of being pressed.

And here is the lesson from that blank spreadsheet at the start of this piece. The absence of evidence is not evidence of absence. When data is empty, the only correct conclusion is "unknown." Not "nothing," and not "something." A poor analyst turns the silence of data into a voice. A decent analyst lets it stay silent, then goes looking for another source.

My own limits

I must confess one thing. My profession has a blind spot. The more I trust the numbers, the more easily I forget that every number is born in a specific context — a table, a ball type, a serve rule, a schedule density. The same 53% rally-win rate can mean something entirely different if it comes from a small tournament against weak opponents, compared with a world championship quarterfinal.

That is why I write less, write slower, and always include a "limits of the data" section at the end of every piece. Not to appear modest. But because that is the professional truth.

What to watch in the next cycle

The next round will answer one specific question: whether the players now praised for "nerve" can hold their form once the 50-50 points revert to the mean. Instead of rewatching the beautiful rallies on social media, I will track three numbers: the first-three-ball win rate, the match density over the two adjacent weeks, and the number of deciding games each player must go through. Whoever plays the fewest deciding games and still wins is not lucky. He is good.