Table Tennis Has No xG: The Data Problem of WTT and the Limits of the 52-Week Ranking

The match ended after seven games. I stayed behind in my apartment in Chengdu, staring at a spreadsheet with only two columns I could actually fill: the score of each game and its duration. The other three — maximum ball rotations, point-win rate on short serves, distance covered per point — were completely empty. Not because I was careless. Table tennis, at the level of publicly available data, has never produced them.

I have followed professional table tennis for more than fifteen years, from handwritten notes at junior events to the present, when every point flows across a screen. One thing has not changed in all that time: this is a sport with one of the most transparent scoring systems in the world, and yet the poorest analytical metrics among individual combat sports. Football has xG. Basketball has step-by-step tracking data. Table tennis has a ranking.

In 2026, when global football stopped, I built a dataset of 2,471 matches from five European leagues between 2026 and 2026 to measure home advantage, then compared it with 494 matches played in empty stadiums. The average home points dropped from 1.54 to 1.21. I mention that not to talk about football. I mention it to place next to another reality: in table tennis, no equivalent dataset exists, and if I want one, I have to record every point myself.

What the 52-week ranking actually measures

The WTT system — the commercial arm established by the ITTF in 2026 — operates on a rolling 52-week mechanism. Points from a tournament expire after exactly one year, and a player's ranking is the sum of their best results within that window. This mechanism has a consequence few ranking readers notice: it rewards participation, not only victory.

A player who enters twelve tournaments a year with a 60% win rate can rank above a player who enters seven with a 78% win rate, simply because the first has more chances to accumulate points. I call this the workhorse effect: the number reflects workload, not quality. The public ranking offers no metric that separates the two.

Points-defence pressure complicates things further. Mid-cycle, when points from a major event are about to expire, a player must choose between competing to bank new points and resting to recover. That choice appears in no data row, yet it shapes the season.

Eleven points and the variance problem

In 2026, the ITTF cut each game from 21 points to 11, with serves alternating every two points and, at 10-10, every point. This is the best-documented rule change in the sport's history, and also the one with the largest statistical consequence.

In an 11-point game, an edge ball or a lucky net cord accounts for nearly a tenth of the game's total points. In a 21-point game, the same event accounts for less than a twentieth. The new rule deliberately doubled the noise probability to make the sport more attractive for television. The result is that every prediction model built on a best-of-seven format is working with a far smaller sample than viewers assume.

That is why I always place game win rate and point win rate side by side. A player who wins 55% of points but only 48% of games has a problem at decisive moments. A player who wins 51% of points but 62% of games is the opposite — he lives on moments. Those two profiles require entirely different readings.

The ball changed three times; the data did not

In 2026, the ball grew from 38mm to 40mm. In 2026, speed glue was banned. In 2026, celluloid was replaced by plastic. All three changes moved in the same direction: less spin, less speed, more rallies.

For an analyst, the consequence is not technical but comparative. A spin metric measured in 2026 cannot sit next to one measured in 2026, because they were born from two physically different sports. Anyone who tells you a player today spins harder than a player twenty years ago is comparing two different balls.

I call this the era variable. It sits silently inside every long-horizon dataset, and it is why I never publish a cross-era model without cutting at 2026, 2026 and 2026.

The seeding loop

There is a mechanism rarely discussed: seeds feed seeds. A player inside the top four seeds usually draws an easier early path, meets fewer strong opponents in the third round, and therefore has a higher probability of going deep. Going deep means more points. More points means keeping the seed. The loop closes.

The correlation here is strong. Causation is far murkier. A world No. 6 may be stronger than a world No. 3 in a specific match, but the number does not say so, and the draw follows the number.

That is why I track a metric of my own: a player's win rate against seeds 1-8 over the last 18 months. Based on my experience tracking matches, this figure diverges from the world ranking far more often than fans expect.

China and the rest

At the Paris 2026 Olympics, China won all five table tennis gold medals. That figure, standing alone, carries almost no information. Placed beside Tokyo 2026, where China lost exactly one final in mixed doubles, it begins to tell a story.

At the 2026 World Championships in Doha, Wang Chuqin won the men's singles final against Hugo Calderano. A Chinese player beating a Brazilian is a familiar result, but Calderano reaching the final is a more notable signal than the scoreline itself.

The current competitive picture has three tiers. The leading tier is China, with squad depth no federation can replicate inside a four-year cycle. The second tier includes Japan, Sweden, Brazil and France — federations capable of producing a player good enough to beat anyone on a given night, but not yet good enough to win four consecutive matches against four Chinese opponents. Truls Moregard's run at the 2026 World Championships is the clearest illustration of that ceiling.

The third tier is everyone else, and this is where the ranking misleads most. Some players ranked outside the top 30 can still beat a top seed on a given evening. Table tennis, with its high point density and small samples, allows that more often than people think.

The survival of a table tennis nation is not decided by its No. 1. It is decided by its No. 5 and No. 6.

What is never measured

The table, the rubber, the sponge hardness, the number of wood plies — these variables directly shape feel, and almost all of them sit outside public data. A player who changes rubber before a tournament usually needs two to six weeks to recover the feel, and during that window their results fall below their true level.

No public dataset records the date of a rubber change. No manufacturer publishes the full configuration a player uses at each event. So I track equipment news as a separate variable, reading interviews, cross-checking public photographs, and taking notes. That work consumes roughly a quarter of my analysis time in a season.

This is a structurally opaque information market, not because anyone hides things, but because table tennis media has never treated equipment as data.

The industrial chain

WTT launched in 2026 with the goal of turning table tennis into a television product that sells at a high price. That path has produced real results: a denser calendar, brighter staging, bigger prize money. But it has also pushed operating costs to a level most events cannot sustain on ticket revenue alone.

The economics of broadcast rights in individual sports has a structural weak point: rights fees rise faster than viewership, because streaming platforms buy rights to capture market share, not to generate profit. I watched that spiral in football, where platforms paid high prices and then withdrew one by one. Table tennis, with smaller revenue and a more fragmented audience, has no shield against the same mechanism.

The signal to watch is not prize money. It is the number of events cancelled or downgraded on the calendar over the next two years.

The contrarian angle: analysts enter the locker room

In recent years, the number of data specialists working directly with national teams has risen. That is a step forward, and also a risk.

The risk is that a model built from historical data always lags reality by one beat. It knows Player A wins 68% of points when serving backspin to the left, but it does not know that the plastic ball at this event is softer, or that Player A just changed sponge hardness and the feel has shifted. The model's conclusions are right about the past and wrong about the present.

A good analyst is someone who knows when to stay silent. Emotion writes the script; data draws the map. I only draw maps. And a blank map is, in some cases, more honest information than a map drawn from assumptions.

What I learned in my early years is this: when a data column is empty, do not fill it with guesswork. State clearly that it is empty. Readers forgive a blank space. They do not forgive an invented number.

In 2026, I analysed Croatia before their World Cup group match against Argentina, calculated an average PPDA of 9.2, and wrote that Croatia would strangle the midfield. The match ended 3-0 exactly as written, and my piece drew 1,200 reads, while a piece mocking the other side's star drew 50,000. The number had spoken first, but people only listened once the truth had become legend. The lesson was not that the data was wrong. The lesson was that honest data needs storytelling strong enough to be read.

Signals for the next cycle

Three scenarios, with probabilities I assign and answer for myself.

Scenario one, roughly 55%: the ranking remains the only tool, WTT maintains a dense calendar, and the gap between ranking and true strength holds at its current level. Amateur analysis will lean ever harder on the ranking, and be wrong more often in individual matches.

Scenario two, roughly 30%: a point-level dataset is commercialised, similar to how tracking data reshaped basketball in the 2010s. At that point the analyst's value shifts from collection to interpretation, and people in my line of work will have to relearn from scratch.

Scenario three, roughly 15%: the calendar compresses under cost pressure, and the most important variable becomes the ability to choose which events to play rather than the ability to play them all. A season is a sequence; crowds watch the match, I watch the pulse of the market.

The thing I am most certain of, after more than fifteen years, is this: a sport never lacks data. It only lacks someone willing to record it. Trusting data is like an early cold morning: few people wake up in time to see it.