From SEA Games to WTT: Vietnamese Table Tennis and an Empty Data Sheet

From SEA Games to WTT: Vietnamese Table Tennis and an Empty Data Sheet

On my computer in Hai Phong there is an Excel file called VTTC_DATA. It has fourteen columns: player, opponent, tournament, round, set-by-set score, points won on serve, points lost on serve, average rally length, win rate in the first three balls, win rate when leading in a deciding set, number of serve-pattern changes, receive errors at 9-9, match duration, and a notes column.

Three rows contain data. Seventy-two rows are empty.

I open that file every Monday morning, not to work but to remind myself that the hardest part of sports data analysis is not the model. It is finding someone willing to sit down and count.

Table tennis is a strange sport. Every point has a server, a receiver, a rally length, a way of ending, and an umpire recording the result. There is no conceptual argument the way there is over possession in football, no phase of play that gets redefined depending on who is watching. Everything is discrete, countable, writable into an empty cell. Yet in Vietnam I cannot find a single public point-by-point dataset for a domestic league, a SEA Games campaign, or a match involving the national team.

This is the paradox I want to dissect in this piece: the sport that is easiest to record is the sport with the least public data in Vietnamese sport.

Context: a perfectly structured sport and an empty library

I began following table tennis seriously after years of building football datasets. My first V.League spreadsheet contained hundreds of errors, but it taught me more than any course could. I once hand-recorded an entire 26-round season, cross-checking possession, shots, corners and cards, and discovered that a team with 55 percent possession had scored only 33 goals. That manual process taught me that sports data is not born from algorithms. It is born from people counting.

With table tennis, I assumed the work would be far lighter. Each set has 11 points, each point ends in seconds, and the camera at a SEA Games or a national championship sits at an angle wide enough to see the whole table. A five-set match can finish in 45 minutes and produce 80 to 100 points. That is a sample size a football match would need eight rounds to accumulate.

But when I went looking for sources, I found only two kinds of information. The first is the final result: who won, by what score, in which round. The second is impression: this player is raw, that one is mentally strong, this match was unbearably tense. Between those two lies a large gap where numbers explaining why the score looked the way it did should sit.

At international level, the WTT system since 2026 and the Grand Smash events since 2026 publish some match data on their platforms: set scores, a few aggregate metrics, basic service statistics. The International Table Tennis Federation publishes rankings and points periodically, publicly searchable. But that data stops at the international tier and at the aggregate level. Nobody publishes point-by-point data for a quarterfinal at the Vietnamese national championship. Nobody records which serve a player chose at 9-9.

The economic cause is obvious. Point-by-point recording needs a person sitting in front of a screen, a unified form, and someone to double-check. All three cost time, and time is what domestic organisers do not have spare. But I would argue the real cost is not money. It is habit. A football tournament has sponsors who want to see possession stats on the evening news. A table tennis tournament has nobody demanding any metric at all, so nobody produces one.

Four variables you can measure starting today

If I were handed a camera and a blank form for a SEA Games, I would start with four metrics. No more, because stuffing too many columns makes the recorder quit after three matches.

The first is points won on serve, the points a player wins while serving divided by the total points they served. At elite level, based on data I have compiled from Grand Smash-level WTT matches, this figure usually sits around 55 to 60 percent in men's singles. If a player falls below 50 percent, they are losing the phase of the game they control.

The second is points won on receive. This is where Vietnamese players are often weaker than the leading group in Southeast Asia. Receiving in modern table tennis is no longer a defensive act. With the 40+ plastic ball producing less spin than the old celluloid ball, the receiver has more chances to attack early, which is why this metric keeps gaining importance.

The third is average rally length. Based on my own tracking, most points in elite men's singles end within three to five shots, and the match average usually lands around four or five. Rally length says a lot about style: it rises when both players are playing away from the table, and falls when one side ends points early.

The fourth is win rate in the first three balls. This is the metric I rate highest and the one least mentioned in domestic coverage. The first three balls decide most of the shape of an 11-point set, because in a format that short, a player has little time to correct a mistake.

None of these four require expensive software. They require a person counting, one form used consistently across tournaments, and clear rules for ambiguous situations such as an edge ball or a point replayed after an umpire's call. I once spent two weeks doing nothing but fixing my own recording rules, and I would argue those two weeks mattered more than any predictive model I have written.

Five rule reforms as five natural experiments

Table tennis has created a remarkable number of natural experiments for itself. Within twenty-five years, competition rules and equipment have changed repeatedly, and each change is a rare chance to observe what actually decides results.

In 2026, the ITTF moved from the 38-millimetre ball to the 40-millimetre ball, effective from October that year, after the Sydney Olympics. A larger ball reduced speed and spin. The measurable consequence was longer rallies and an advantage for players with better physical conditioning and movement.

In 2026, the game moved from 21-point sets to 11-point sets with service changing every two points, effective from September. Shorter sets mean every point carries more weight, and the ability to overturn a match through one long run of points dropped sharply. A player leading 8-4 holds a completely different position than in the 21-point era.

In the same period, from 2026, the ban on hiding the serve was tightened: the server must let the opponent see the ball from the moment of the toss. This is one of the most consequential changes that casual followers rarely notice, because it directly erased the advantage of a group of players who lived on serves that were hard to read.

In 2026, after the Beijing Olympics, the ITTF banned speed glue, effective from September. Speed glue once allowed the ball to fly faster and spin harder for a few hours after a rubber was glued. The ban removed a form of technical advantage tied to equipment and forced teams to look for edges elsewhere.

In 2026, the 40+ plastic ball replaced the celluloid ball, mandatory from July. The plastic ball has different spin and trajectory characteristics, and I consider it the single most disruptive equipment change for the current generation of players.

When the Bundesliga played in empty stadiums, I realised home advantage is just a variable waiting to be deleted. The five rule reforms of table tennis taught me the same lesson on a larger scale: much of what we call a player's identity is in fact a temporary advantage created by a specific rulebook. When the rulebook changes, the advantage vanishes, and people call it a decline in form.

The ranking mechanism and points-defence pressure

Since 2026, the ITTF ranking system has counted a player's best eight results over a 52-week window, with points expiring after twelve months. This mechanism has a consequence fans often overlook: a ranking reflects not only strength but also calendar.

A player who enters many events, travels many legs and accumulates many results will rank higher than a stronger player who rarely goes abroad. Since 2026, the WTT system with Grand Smash, Star Contender and Contender tiers has created a clear points ladder, and that ladder rewards consistent attendance.

For Vietnamese players such as Nguyen Anh Tu, Dinh Quang Linh, Nguyen Khoa Dieu Khanh or Mai Hoang My Trang, this is a worrying closed loop. Travel and accommodation costs for an international leg are a genuine barrier, not an excuse. If you cannot travel, you cannot post counting results. Without points, you cannot enter higher-tier events. Without higher-tier events, you cannot earn points. I read a team through thirty variables before listening to a commentator, and the same principle applies here: before saying a player has stalled, check whether they were allowed to compete.

This also explains why young Asian players often climb the rankings faster than their true level early on, then get pulled back as their schedule thickens and old points expire. The points-defence pressure index, the share of points expiring in the next three months, is what I always calculate before assessing a player. It tells you whether they are playing to build or playing not to lose.

Vietnamese table tennis: the gap sits in the first three balls

Across several SEA Games and regional events I have followed, the pattern repeats fairly clearly. At Southeast Asian level, the gap between the leading group of Thailand and Singapore and the rest usually does not sit in long-range hitting or attractive rallies. It sits in first-three-ball efficiency and in the ability to receive under scoreboard pressure.

In other words, Vietnamese players usually do not lose because their basic technique is weak. They lose because at 8-8 their service choices vary less, and because at 9-9 they return a ball the opponent already predicted. This is the kind of difference no news report can describe without point-by-point data. A commentator can only say the match was tense. A recorder can say this player used four different serves across ten decisive points while the other used two.

At domestic level the problem is even more basic. The national championship and the youth system run regularly, with players, coaches and spectators, but there is no dataset accumulating across years. That means a 19-year-old joins the national team and nobody knows precisely what their strengths were three years ago. Evaluation shifts entirely to the coach's eye, and the eye is strong at spotting technical faults and weak at remembering the numbers of two hundred past matches.

I do not think data replaces a coach's eye. I think data is that eye's memory.

The transfer window and the value of a portfolio with numbers

In table tennis, the international transfer market runs mainly through national leagues and club systems: the T.League in Japan, the German table tennis Bundesliga, Chinese clubs and several other European leagues. These clubs recruit foreign players on very specific criteria, and those criteria almost always rest on match data.

When a professional table tennis club reviews a foreign player, it typically wants to know: their win rate against left-handed players, their performance in deciding sets, the variety of their serves, their doubles ability, and their consistency across a long season. These are questions data can answer, if the data exists.

For a Vietnamese player hoping to compete abroad, the biggest obstacle is not level. It is that nobody can prove that level with statistics. A European coach cannot watch every recording of domestic matches. They need a profile page with numbers. A transfer only deserves to happen when it answers a question posed by data, not by the media.

This is why I believe building domestic table tennis data has direct economic value, not just the appeal of a hobby for people who like counting. A player with a strong statistical profile has more options. A club that recruits with data buys fewer mistakes. And a table tennis ecosystem with accumulated data negotiates from a different position.

The counterintuitive angle: the problem is not missing data, it is missing questions

I once believed that with enough data, answers would simply appear. Before a major tournament, I ran a regression across hundreds of international matches and produced a very confident probability for a team I was following. Reality demolished that prediction. When I rewatched the footage and counted every phase, I realised my historical data could not measure something very simple: how hard the midfield lines were willing to run.

The lesson was not that the model was wrong. I was the one who had trusted it absolutely. Data does not need me to believe in it. Data needs me to check it.

In table tennis, the equivalent mistake takes a different shape. People import a football metric into table tennis without checking what it means here. Possession does not exist in table tennis because every point has a clearly defined owner of the right to serve. Home advantage in team table tennis is far smaller than in football, because a crowd cannot pressure an umpire the way it can at a stadium. And rankings, as I said, are a record of a calendar rather than a pure measure of strength.

The biggest blind spot in table tennis analysis is the kind of data nobody bothers to record. I call it dead data: information that exists for a moment and is never stored. Which serve a player chose after an opponent called a timeout. Which option they picked at 10-10. How they changed tempo after losing three points in a row. These things decide matches more than any aggregate metric, and they disappear the moment the match ends if nobody writes them down.

One more counterintuitive point: I do not think Vietnamese table tennis needs a predictive model. A predictive model is the final step, and it is only useful once clean data exists. The first task is to record well enough to explain why a match unfolded the way it did. Answering that question is already a major advance over recording only the score.

What to watch in the next cycle

Three signals I will be tracking, and which I suggest readers interested in Vietnamese table tennis track too.

The first is service performance in deciding sets among domestic players at regional level. If points won on serve in the final set stay level with earlier sets, that is a sign of settled competitive mentality. If it drops sharply, the problem lies in preparation for pressure points, not in technique.

The second is average rally length. A Vietnamese player who can lift rally length to the level of the leading Southeast Asian opponents will have more ways to win, because a longer rally means they have escaped being finished off early.

The third is the international calendar. How many events a player is eligible to enter over the next 52 weeks will shape their ranking more than any training block.

From a V.League spreadsheet to a Bundesliga model, my journey has been a journey of numbers that speak. Three rows of data in my VTTC_DATA file may sound thin today. But I started with a sheet containing hundreds of errors, and that sheet still taught me more than anything I have read. The job of a data writer is not to wait for someone to finish building the system. It is to start counting from the first point of the most recent match.