The Empty Cell in Badminton Analytics: When the Spreadsheet Has Nothing Left to Say

In March 2026, in a small meeting room in Nanshan District, Shenzhen, I placed on the table a spreadsheet with 47 columns and nearly two thousand rows of situations coded from video of a domestic badminton tournament. The most important column, the one recording how each rally ended, was empty in 38 percent of the rows. The head coach skimmed it and said: "Just read the rest." I did not. I circled those 38 percent of empty rows and asked back: if we treat the blank space as neutral, we are teaching our players to compete against an opponent who does not exist.

The meeting ended twenty minutes later with no tactical conclusion. Those were the most valuable twenty minutes of the month. This story repeats itself many times in my work, in many forms, and it explains why I grow increasingly suspicious of analyses that read too smoothly.

Context

Professional badminton has entered a phase where data is no longer a luxury. The Badminton World Federation introduced the Instant Review System using Hawk-Eye technology at top-tier events in 2026, turning every line call into a retrievable data point. The 21-point rally scoring system, adopted in 2026, gives every rally life-or-death value: a lost rally is never compensated for. Leading national teams all have their own analysis units, tablets placed on the coach's chair, and someone behind the stands logging every shuttle trajectory.

Data infrastructure does not automatically produce conclusions. A decent analytical process must pass through several layers: extracting raw data, cross-checking against video, building a model, and only then reaching a tactical conclusion. If the first layer fails, every later layer is merely well-presented prose. Based on my experience watching matches, I have read thirty-page reports with colourful charts whose data source was nothing more than an empty column filled in with feeling.

During the current transfer and squad registration period, the pressure grows. Clubs must finalise rosters, foreign-player slots and wage budgets. When time is short, people lean toward accepting an incomplete analysis, because a flawed analysis is easier to live with than an empty decision.

Core analysis

Three types of error appear when a data cell is empty, and all three are dangerous in their own way.

The first is defaulting to neutrality. Treating a missing value as "no effect" is an assumption, not an observation. In my sheet, the 38 percent of rows missing a rally-ending value clustered around rallies longer than fifteen strokes, precisely the physically decisive group in the third game. Removing that group from the analysis removes the entire story about endurance, which is exactly what my team was trying to assess.

The second is filling the gap with reputation. When data on a player is missing, people substitute ranking, past results, memories of a good match. An empty court strips away reputation. What remains is discipline. A player who once reached the semi-finals of a Super 1000 event does not therefore have a higher net-rally win rate right now.

The third, and most subtle, is using the wrong proxy metric. I once saw a report treat the short-service rate as a proxy for "attacking intent". Those are different things. A short serve is a starting choice; attacking is the chain of decisions that follows. A player who serves short 70 percent of the time can still play counter-attacking defence for an entire match. In Shenzhen, I have watched data replace intuition. The results are not always prettier.

The metric set I use for men's singles analysis has four groups, and I always state the margin of error for each. The first is rally-length distribution, measured in strokes, split by game. The second is the ratio of self-inflicted errors to errors forced by the opponent; these two figures are merged in a great many public statistics packages, and merging them is a serious diagnostic mistake. The third is net-area efficiency, calculated as points won over approaches to the net. The fourth is attack-return rate, measured as rallies surviving beyond the third stroke.

Each group carries clear trade-offs. Rally-length distribution speaks powerfully about fitness but stays almost silent about technique. Net efficiency reflects decisiveness but depends heavily on the quality of the opponent's shuttle. The attack-return rate is the metric I trust most, because it is hard to fake: to survive past the third stroke, a player needs positioning, reflexes and the decision to read the shuttle's direction, all at once.

Three weeks of comparing video with the dataset taught me something no software can teach: the true value of a metric lies in the rows it forces me to leave blank. Numbers do not lie. But they are extremely good at selecting which truths to show. An empty data column means something entirely different from a weak one. It is a statement that we do not yet understand the phenomenon well enough to measure it.

At the elite level, players such as Viktor Axelsen of Denmark and An Se-young of South Korea all work with dedicated analysis teams, and the gap between them and the rest lies largely in the quality of the questions they ask, not in the volume of data they collect. The figures cited in this article come from my personal tracking sheet, compiled from publicly available match footage and cross-checked twice before entry.

The contrarian angle

Most failures in badminton data analysis do not come from algorithms. They come from ignoring the blank zones. We are taught that more data means better conclusions. Few people teach that the quality of a model depends on how it handles missing data.

The execution blind spot also sits on the coaches' side. Their intuition is often treated as the opposite of data. I think that framing is wrong. Intuition is a variable; it is simply an unmeasured one. In my personal tracking sheet, one coach in a domestic league made the correct substitution in seven of nine matches by reading his player's breathing during the interval. No column in any spreadsheet recorded that breathing. My job is to build that column, not to dismiss it.

The 2026 World Cup taught me that every system can be dismantled. In badminton, the local version of that lesson is this: every statistics table can be dismantled, including the ones I build myself. The dismantling always begins with the empty cells nobody wants to look at.

Forward-looking note

Over the coming months, as teams finalise rosters for the next phase of the season, I will track one thing only: which clubs openly publish the data they do not have. The teams that admit their blank zones tend to be the better-prepared ones, because they know exactly what they need to collect at the next tournament. Positional discipline on court and discipline inside a spreadsheet are, in the end, the same kind of discipline.