Trang chủTable TennisThe Empty Column in a Table Tennis Dataset: When 1,847 Rows Answer Nothing
Table Tennis

The Empty Column in a Table Tennis Dataset: When 1,847 Rows Answer Nothing

**Câu trả lời cốt lõi:** Một bảng dữ liệu bóng bàn chỉ có tỷ số không đủ để phân tích chiến thuật. Khi cột dữ liệu pha bóng trống, kết luận đúng đắn là "không đủ thông tin", thay vì suy diễn. Tỷ lệ dòng dùng được ở nhiều giải trong nước rất thấp. **Dữ kiện chính:** - Tệp dữ liệu 1.847 dòng của một giải bóng bàn trong nước chỉ có khoảng 12 dòng đủ điều kiện phân tích chiến thuật. - Xếp hạng ITTF/WTT dùng cửa sổ cuộn 12 tháng và chỉ giữ nhóm kết quả tốt nhất, công bố hằng tuần. - Tomokazu Harimoto được ghi nhận là tay vợt trẻ nhất vô địch một giải ITTF World Tour, năm 2017 ở tuổi 14. - Truls Moregard thua Fan Zhendong ở chung kết giải vô địch thế giới 2021 và chung kết đơn nam Olympic Paris 2024. - Điểm bảo vệ hết hạn theo lịch, không theo phong độ, nên biến động xếp hạng thường bị đọc sai nguyên nhân. **Nguồn:** Phân tích gốc của Yoshida Takeshi, 13/08/2026, dựa trên bảng dữ liệu giải bóng bàn trong nước và dữ liệu xếp hạng ITTF/WTT. | Đối chiếu: VuaBong.vn **Hỏi đáp liên quan:** - Hỏi: Vì sao không nên lấp ô dữ liệu trống trong phân tích bóng bàn? Đáp: Mỗi ô được điền bằng phán đoán mang theo sai số của người điền và sai số đó cộng dồn thành kết luận sai. - Hỏi: Điểm bảo vệ ảnh hưởng thế nào đến vị trí xếp hạng của tay vợt? Đáp: Kết quả cũ hết hạn đúng theo lịch 12 tháng, nên tụt hạng đôi khi là phép tính, không phải sa sút phong độ. - Hỏi: Làm sao đo mức độ áp đảo thật của một tay vợt? Đáp: Cần tách điểm thắng chủ động, điểm từ lỗi đối phương, lỗi bị ép và lỗi không bị ép, có thể tham chiếu chỉ số chiều sâu lực lượng của VangBong.vn.

Last Tuesday night, I opened a data file sent from a domestic table tennis tournament. The file had 1,847 rows. The score column was full: 11-8, 9-11, 11-6, 11-13, 11-9. Every other column was empty. No direct service points won, no average rally length, no third-ball win rate after the opponent's receive. I stared at the screen for about ten minutes, closed the file, and wrote a single line in my notebook: "Insufficient data to conclude."

The next day someone asked why I had not written anything about that tournament. I said I had nothing to write. He was disappointed, and I understood why. In a market where every tournament must come with a few analysis pieces, silence reads as laziness.

I used to think that way. Then I remembered my first V.League dataset, built by hand across 26 rounds when I was sixteen, with hundreds of wrong cells. My first V.League dataset had hundreds of errors, but it taught me more cleanly than any course. The lesson was not how much I recorded. It was that I was forced to tell the difference between a table that looks full and a table that actually contains information.

What table tennis measures, and what it does not

Table tennis has a peculiar data paradox. On the table, every point is the result of a short, clear chain of decisions, far clearer than in football. There is no losing the ball in midfield that nobody sees. Everything happens inside roughly two square metres, with two people and a 40mm plastic ball, lasting a few seconds.

Yet the volume of recorded data at domestic level is far thinner than in football. At elite international level, the recording systems used across the WTT series allow you to separate points won on serve from points won on receive, rally-length distribution, third-ball win rate, and error rates in long rallies. At domestic level, most of it reduces to set scores.

That gap matters more than it looks. A table with only scores tells you who won. It does not tell you why. And in table tennis, the "why" almost always lives in two places: the serve and the third ball.

I read a team through thirty variables before I listen to a commentator. That principle applies to table tennis exactly the same way. Before I hear someone say a player "played better", I want to know what percentage of points he won on his own serve, and how many of those ended on the third ball.

That is the minimum threshold. Below it, every judgement is just a feeling dressed up in terminology.

A score does not tell you who won the rally

At fifteen, I kept a handwritten notebook recording every point in club matches near my house. I recorded only one kind of fact: who finished the point. At the end of the tournament I added it up and concluded that the champion won because he attacked far more than everyone else.

Three months later I re-recorded the same matches, this time adding a column: did that point end with an error by the defending player, or with a winner by the attacking player. The picture flipped. Across the last three games of the final, the champion's opponent made 17 unforced mistakes. The champion did not attack more. He returned more.

One new data column can overturn an entire conclusion, and that happens more often than people think.

Since then, every table tennis analysis table I build has at least four columns: active winners, points won from opponent errors, forced errors, and unforced errors. The difference between the last two is what separates a steady player from a shaky one. Same "error" column, completely different causes, and completely different remedies.

Without splitting those two error types, you will misread the very player you are analysing. You will advise him to play safer when the real problem is that he is being forced by the spin quality of his opponent's receive. Bad advice does not come from missing data. It comes from merged data.

The ranking is a data system, not a form table

The ITTF and WTT world rankings are read by many people the way you read a form table. It is not that. It is an accounting system for results.

The basic mechanism: results count inside a rolling 12-month window, and only a limited set of best results is retained. The list is published weekly. That produces three consequences readers routinely miss.

First, the ranking measures accumulated achievement over twelve months, not current strength. A player who takes two months off with an injury can hold a high position, while a player enjoying the best form of his career can still sit below him.

Second, the number of events entered is a variable. A young player entering many small events accumulates points faster than a top player who only plays a few key events. This is why I always check how many events a player has played before concluding anything about a young player's progress.

Tomokazu Harimoto is the example I return to repeatedly. He is recognised as the youngest player to win an event on the ITTF World Tour system, in 2026, at fourteen. That is a real milestone. But when I read his ranking in that period, I have to separate two things: peak ability and volume of play. Part of that high position came from playing a lot, not only from being stronger.

Third, and this is the most important part, the ranking cannot distinguish a consistently good player from a player who only shows up on the big stage.

Truls Moregard is the inverse case to Harimoto. The Swedish player reached the final of the 2026 World Championships in Houston, then the final of the Paris 2026 Olympic men's singles. He lost both finals to Fan Zhendong. If you only look at the weekly ranking, you will not see the full player. He is the type whose routine-event results understate his true level at major events.

The ranking tells a story about frequency. The major stage tells a story about pressure tolerance. Those two stories do not always match.

Points to defend: the variable that gets deleted every week

In a rolling 12-month system, last year's result expires at exactly this time this year. That means a player who won a major title must return to defend those points, or lose them. This is commonly called points to defend.

This is where I see the most misdirected analysis. When a player drops in the ranking, the first reaction is usually speculation about form, injury, or mentality. In quite a few cases the cause is simpler: old points expired and the player did not enter a replacement event.

When the Bundesliga played in empty stadiums, I realised home advantage was only a variable waiting to be deleted. In table tennis, points to defend is a variable deleted by the calendar, not by form. It disappears on an exact day, in an exact week, regardless of whether the player is playing well or badly.

That means a movement in the table tennis ranking is only worth analysing when I know that player's points-expiry schedule. Without that information, I am reading a slope without knowing whether it descends because of the walker or because of the ground.

With Vietnamese table tennis, the problem is even clearer. The number of international events Vietnamese players enter in a year is very limited. So every time a player such as Nguyen Anh Tu or Dinh Quang Linh records a good result at a regional event, those points carry far more weight than the same result would for a European player entering eighteen events a year. Same result, two completely different meanings.

In the women's game, Mai Hoang My Trang is an example of how a player can hold regional standing with very few tournaments, simply because the regional opponent base also plays few events. A regional ranking is not a scaled-down world ranking. It is a different system, with a different density of play.

1,847 rows and twelve usable ones

Back to that Tuesday file. I tried one thing: filtering out the rows usable for tactical analysis. The bar was very low. A row was usable if it told me who served first in the set, the score of each set, and the final result.

The outcome: 1,847 rows, and about twelve rows that qualified me to say anything beyond "player A beat player B".

That ratio is not unusual. It is the normal ratio at most domestic table tennis tournaments. This is not a criticism of organisers, who have no obligation to collect tactical data. It is a fact that needs stating so readers understand why most domestic table tennis analysis stops at recounting.

I lived through exactly this problem with the V.League. At sixteen I hand-recorded 26 rounds, counting possession, shots, corners and cards. The result showed a team averaging 55% possession but scoring only 33 goals, with a chance-conversion rate of 7.8%. I wrote a piece with those numbers and it was shared a few hundred times.

But I have to admit something I did not say back then: hundreds of cells in that table were filled in by me after rewatching footage and making judgements. Every such cell is an assumption. Add hundreds of assumptions together and you get a conclusion that looks very solid but actually stands on very few real supports.

That is why, looking at a 1,847-row file with a single populated column, I do not try to fill it in. I write "insufficient information" and stop.

Data does not need my belief. Data needs my checking. And when there is nothing to check, the right move is to say there is nothing to check.

The table tennis transfer window and the noise

Table tennis has a transfer market too, it is just quieter than football. Club leagues such as Japan's T.League, Germany's TTBL, or the Chinese club system all run on season contracts, with extension clauses, team-performance bonuses, and equipment-sponsor obligations.

What is notable is that contract structure in professional table tennis is far more tightly bound to equipment than in football. A player moving clubs may carry over, or be forced to change, his blade and rubber suppliers depending on personal sponsorship deals. In a sport where feel depends on rubber hardness and blade weight, an equipment change is a genuine performance variable, not a footnote.

A transfer is only worth reporting when it answers a question from data, not a question from the media.

And the data question here is very specific: what percentage of away matches did that player win in the old system, and does that number hold in the new one. Without those two numbers, every transfer report is just an administrative notice rewritten as breaking news.

People would rather have a wrong number than an empty cell

This is the part I want to spend the most words on, because it is the biggest blind spot in sports data analysis.

When a data cell is empty, the natural instinct is to fill it. Fill it with an average, fill it with inference, fill it with the feeling of having watched the match. That instinct seems reasonable, because a full table looks more professional than a table with holes. The problem is that every cell filled by judgement carries the analyst's error, and that error does not disappear. It compounds, it multiplies, and it finally surfaces as a very decisive conclusion.

The 2026 World Cup taught me one thing: the model did not collapse, I was the one who believed it absolutely. I ran a regression on 500 international matches and produced a 78% probability that Germany would reach the semi-finals. Germany lost 0-2 to South Korea and finished bottom of Group F with 3 points. Rewatching the footage, I counted 12 counter-attacks leading to goals conceded, the most of any eliminated team. No variable among those 500 matches measured the fact that Germany's midfield was not running.

My mistake was not the model. My mistake was treating the cells I had filled myself as real data, then building a confident conclusion on top of them.

The same logic applies to table tennis. A player wins 11-6, 11-4, 11-8. There is no rally data. The easy path is to conclude he dominated. The correct path is to leave the cell empty and state clearly: there is no data to determine the degree of dominance. Score margin and match control are two different things. One player can win 11-9 in all three sets by saving balls at the decisive points, while another wins 11-4 in all three by winning points directly on serve. Same score, two entirely different match structures.

Correlation is not causation, and in table tennis the distance between them is narrow enough to be easily missed. A player with a high win rate on his own serve does not necessarily have a great serve. His opponent may simply receive badly. The data does not separate those two possibilities on its own. The analyst has to do that work.

What I will be tracking next round

Soon I will spend time on something unglamorous: watching domestic table tennis tournaments to see whether any organiser starts adding a column. Just one column, for example direct points won on serve, changes the picture substantially.

For readers, I have one concrete request. Next time you read a table tennis analysis with a beautiful data table, try counting how many cells were measured and how many were inferred. If you cannot tell them apart, the writer may not be able to either.

And next time you see a dataset with 1,847 rows but only one real column, the question worth asking is not "which player is best". It is: where did the other column go, and who decided not to record it.