Blank Cells in the Spreadsheet: Data Discipline and How We Analyze Southeast Asian Badminton

An office in Kuala Lumpur, 11:40 p.m. I open a file a colleague sent, expecting the usual columns: rally win rate, conversion at decisive points, distribution of shuttle landing zones. The file is empty. Not a single row. In that moment there is a very specific temptation that anyone who has sat in front of a spreadsheet understands: fill the blanks. Just place one number in an empty cell, and the sheet instantly looks 'complete'—and a complete sheet always looks like the truth.

I did not fill it. And that decision not to fill it is the subject of this article. In sports analytics, especially in the Southeast Asian badminton market, the most dangerous thing is not missing data. The most dangerous thing is false confidence built on blank cells.

Context: why badminton resists analysis more than football

I grew up with football, and I entered the profession through an xG model. But the market I live in—Malaysia—is a badminton market. Here, people do not just watch badminton; they live on it. International badminton events in Kuala Lumpur sell out in hours, and coffee shops in Brickfields or Cheras are packed during semifinal nights. That is why I moved part of my focus to this sport.

But badminton resists data models in its own way. In football, a match has 90 minutes, roughly 1,000 passes, and dozens of scoring situations that can be assigned value. In badminton, a match lasts 40 to 90 minutes but records only one number per rally—win or loss. No 'passes,' no 'penalty area,' no 'assists.' Everything subtle lives between two numbers: the score before and the score after.

That is why most public badminton data stops at three metrics: points, unforced-error rate, and match duration. For fans, that is enough. For an analyst, it is useless. You cannot say anything about a player just by knowing he won 21-19. You need to know where on the court he scored, with which stroke, in which situation, and—most importantly—whether he did it at 19-19 or while leading 18-9.

I started with lower-division xG, where people mock every number. That experience taught me one thing directly applicable to badminton: the truth is not in the big match but in the scorned data zone.

Core: building a measure for a sport that has no xG

When there is no ready tool, you must reframe the question. In badminton, the right question is not 'is this player good' but 'how was the value of this point created, and can it repeat?'

Contextual rally win rate

Not all points in badminton are equal. A point at 19-19 carries psychological and tactical weight utterly different from a point at 5-3. If you only count total points, you miss the entire pressure structure of the match. I divide each game into three zones: opening (0-8), middle (9-15), and decisive (16-21). Then I measure each player's win rate in each zone.

This approach exposes what the scoreboard hides. Some players win 70% of opening points but fall below 50% in the decisive zone. Reading the result, they are the winner. Reading the structure, they are fragile when the match enters its tensest phase. In betting—where I earn a living—that difference is worth more than the world ranking.

Shuttle landing value

This is the closest thing to xG I can build in badminton. Every stroke that ends a rally can be assigned a scoring probability, based on the hitter's position, the opponent's position, and the stroke type (smash, slice, drop, lift). From video data of roughly 200 national and international matches I coded myself, I calculated that a straight smash from mid-court carries a scoring probability of about 0.42, while a cross-court drop at the right tempo carries about 0.61. The absolute number matters less than the ranking. And that ranking shifts by opponent.

I once described this approach briefly: data is like a monk—the fewer the words, the more the truth. Three pillar numbers, wrapped in one sentence, beat three pages of tables.

The scorned sample

In 2026, while building an xG model for Malaysia's Super League, I found a young striker at a second-division club with an xG of 0.82 per match, against a league average of just 0.41. Nobody noticed him because his club was not rated. By season's end he scored 23 goals and the club won the second division. He was bought by a Thai club for two million ringgit.

I tell this story not to boast. I tell it to say that the structure of the problem is always the same, whatever the sport. The most valuable signal always sits where the fewest people look. In Southeast Asian badminton, that place is continental-level events, qualifying rounds, and the men's doubles matches the media ignores.

The Malaysian case: rereading what the scoreboard does not say

Take an example I have followed for years. In men's doubles, Malaysia's Aaron Chia and Soh Wooi Yik won bronze at the Tokyo 2026 Olympics and bronze at the Paris 2026 Olympics, plus the 2026 world title. If you only read results, you know they are an elite pair. If you read their match structure, you see something else: their decisive-zone win rate runs notably higher than their overall match win rate, and it is not season-stable.

That means what they show at the decisive moment is not an innate quality. It is a state that rises and falls. For an analyst, that is the whole point: a pair does not 'have nerve' in some abstract sense. They have, at a specific moment, a cluster of metrics that lets them survive pressure—and that cluster is measurable, comparable, and can vanish.

The same applies to Lee Zii Jia, the 2026 All England champion. Reading the scoreboard, people talk about 'form.' Reading the structure, people talk about which stroke he scores with in the final twenty points of a game. Those are two ways of describing two levels of understanding.

I stress this because the Malaysian market has a habit I dislike: using reputation and head-to-head history in place of analysis. In the transfer market, people pay for reputation, not performance. In betting, people wager on memory rather than probability. Both are mistakes with the same root: reading names instead of numbers.

The contrarian angle: correlation is not causation

Here is the part where I must warn myself most.

When you have a good model, it is easy to fall into the trap of hunting data to confirm a hunch you already held. The data boom has produced a generation of analysts who can run a model but cannot doubt their own. I have been there. I have faced the consequences of being right—and that made me more careful.

In 2026, I published an analysis arguing that Germany, the reigning world champion, would be eliminated in the group stage. I relied on qualifying data: their defense allowed average opponents more than 120 passes in dangerous areas per match, and their pressing metric was suspiciously low. The piece was mocked. When Germany lost its final group match and went out for the first time in eighty years, the piece was shared thousands of times.

The 2026 World Cup taught me that a team is never invincible—but it also taught me the opposite, the thing few want to hear: being right once does not mean the model is right. If I repeated that result ten times and was wrong seven, I would not be a genius; I would be a lucky lottery-ticket seller. Public verification is not a tool for showing off wins. It is a tool that forces me to expose the times the model collapsed.

In badminton, this trap appears in another form. There is a widespread belief that home advantage in team events like the Thomas Cup or Sudirman Cup is undeniable. I do not trust claims that cannot be measured. In 2026, when European football stadiums closed during the pandemic, I compared pre- and post-data and found home-team win rates fell sharply. When the stadium stood empty, I realized home advantage is only the echo of the crowd. The crowd is a quantifiable player—and when that player is absent, the number disappears.

In badminton, the crowd has its own technical signature: high noise, strong lighting, and a list of strokes the audience reacts to most fiercely. If home advantage exists, it lies not in 'spirit' but in the rallies where noise shifts an opponent's reaction timing. That is a testable hypothesis, and I am gathering data to test it—not to prove it right.

The difference between those two things is the difference between an analyst and a prediction salesman.

Data infrastructure: the part few discuss

There is a dry fact I must state, because it determines everything above. Southeast Asian badminton data depends on infrastructure, and that infrastructure is thin.

At small events, scoring is done by one person sitting courtside. No tracking system, no independent analysis cameras, no international data provider sending raw data. That means most data must be created by hand. I have spent months coding match video myself—logging every rally, every stroke, every landing point—just to get a sample large enough for a single analysis.

This is why promises of 'AI models for badminton' sound glamorous in headlines but are often meaningless in practice. You cannot feed an algorithm dirty data and expect truth. Garbage in, garbage out—but garbage packaged in a shiny interface.

For a serious badminton model, one principle is absolute for me: every prediction must come with the conditions that would break it. If I say a player will win more in the decisive zone next tournament, I must write clearly: 'this is wrong if his unforced-error rate exceeds his recent-three-match average, or if he must play three three-game matches in two days.' A model is right only until the shuttle flies; after that it is the story of probability. Without breakup conditions, a prediction is just a claim dressed up with numbers.

Decoding the public narrative around Southeast Asian badminton

Badminton in Southeast Asia is not merely sport. It is identity. In Malaysia, Lee Chong Wei's wins over long-time rivals are remembered as national memory. In Indonesia, badminton is loved more than football. In Vietnam, where I was born, badminton is growing fast but its data system is still young.

Public storytelling about badminton has a repeating rhythm: a young player appears, wins a few matches, is called 'the hope,' then loses a big one, and the story shifts to 'mentality not yet solid.' This rhythm appears across the region, with every player. It is appealing because it is simple—it turns a complex structure into a story about a character.

But mentality is not an independent variable. Mentality is the output of fitness, schedule, and opponent difficulty. A player with a 'fragile mentality' may simply be someone coming off three three-game matches in four days, facing an opponent with a smash that is probability-superior in the decisive zone. Calling that mentality is a way of dodging analysis.

The life cycle of a narrative is something I track closely, because it is the most mispriced thing on the market. When a story about a player peaks, the odds usually already reflect that excitement, not the facts. Finding the gap between crowd expectation and objective assessment is my job. Sometimes that gap is positive, sometimes negative, and most of the time it is zero—and in that case, the correct answer is to bet nothing at all.

Takeaway: what the next round's signal is

I write this to restore honesty to a blank cell.

The spreadsheet I opened that night had no data, and the right thing was not to fill it with numbers I did not have. The right thing was to say I do not know. In a market crowded with people who want a confident prediction, the ability to say 'I don't know' is the hardest-won and most valuable professional skill.

For the coming season, the signal I am watching is not match results. It is three things: first, the pressure distribution of Southeast Asian players in the decisive zone when playing consecutive days; second, the shift in shuttle-stroke value when a player changes court surface—something bookmakers price very poorly in Asian events; and third, the measurable effect of crowd noise on error rates on service returns. These are data columns nobody keeps yet. They sit in the scorned zone, and as always, that is where the truth lives.

So the question I leave you: when you watch a badminton match, are you watching who wins—or a probability structure repeating itself? If you cannot answer that question with numbers, you have not yet seen the match.