The Empty Report and the True Limits of Sports Data

On a Sunday morning in Surabaya, I reopened the analysis report for a domestic badminton tournament that had just ended. What struck me was not an unusual metric but emptiness. Every data field returned the same line: insufficient information to assess. No smash speed, no rally length, no net-point win rate, not even a player name. I sat in silence for a while. Fifteen years as a data consultant, I was used to models returning wrong or skewed results, but never silent ones. That silence forced me to write about itself rather than fill a tactical story to make the tables look good. My job is to read matches through numbers. I split metrics by court zone, cross-check with footage, place pressing measures beside shot origin. But the first principle I set for myself after 2026 was not "collect more data" but "know when data says nothing." That year, at 36, I used an xG model to advise the coach to push the line high in a promotion playoff. The model predicted 1.8 xG for us, but we lost 0-2 because the opponent sat deep and every shot was a harmless long-range attempt. I had ignored PPDA and the match context, looking only at the total number. The model was not wrong; I was wrong to let it speak in place of my own eyes. Today's empty report is a harsher version of that lesson. It is not skewed. It simply has nothing. For someone who likes order, who files everything into boxes, emptiness is more uncomfortable than a wrong result. A wrong number can be argued over. An empty cell offers only one honest choice: leave it empty. In badminton, data gaps appear more often than people think. A lower-tier BWF World Tour event may have no serve-speed sensors, no full Hawk-Eye system, no one recording each rally's length. Any comparison with a Super 1000 event is then a broken comparison. I have seen roundups citing "net-point win rate" for two players at two events of different tiers, different surfaces, different crowd conditions, then concluding one is better than the other. That is when data becomes decoration, not evidence. What I learned from watching matches live is that most of a badminton match's story is not in the stats table. It sits in the gap between a player's feet while waiting to serve. It sits in a player deliberately dropping short after the third rally to pull the opponent to the net before opening the cross-court angle. It sits in shortening breath in the third game, in surplus movement during a lost game that no table records. A player's true value lies where he runs and when he stops. That does not mean I turn my back on models. I still believe in them. But I distinguish two situations clearly: data not yet sufficient to conclude, and data that has said everything while readers demand more. In the first case, the kindest act is to say plainly "not enough." In the second, the duty is to rewatch footage, put numbers in context, split them by zone and timing. Many social media arguments about referees and video review stem from confusing these two situations. There is a temptation anyone in data work has faced: the temptation to fill empty cells. People call it "telling stories with numbers." But storytelling is different from fabrication. A small correlation, a sample of a few matches, an oddly prominent secondary metric — all can be inflated into a law. I nearly did that many times, and each escape came from one self-question: does this metric have a real mechanism deciding the outcome, or is it merely moving in the same direction. Croatia did not win the title, but they showed me a truth hidden in a number. Croatia's PPDA is the reward for anyone patient enough to pick up pass after pass. The lesson is not the 9.2 in the group stage or the 12.4 recoveries per match, but how it forced me to rewatch each phase to understand why Modric and Rakitic chose the right moment to press. The number only opens the door. You have to walk in yourself. Then in 2026, when the pandemic halted every league, I realized something else. I was retained during lockdown to forecast the team's form once football returned. I built a model from the first 15 rounds and advised the team to keep its possession style. The result: three straight defeats when the league resumed, because opponents pressed harder in empty stadiums and we lost the ball in our own half. My model entirely lacked two variables: the crowd and the spacing between players on the pitch. The pandemic taught me that data also knows fear — when the world stops, numbers are meaningless. Since then, I changed how I write. I no longer assert in one direction. Every analysis must carry at least two scenarios, and I always annotate the context of every number. Numbers are a chant, but intuition is the candle — I light both whenever I read a match. This approach holds for badminton too, where a pressing metric is rarely recorded yet decides who controls the tempo after each long rally. Back to the empty report on that Sunday. I decided to fill nothing. Instead, I added a line for readers: this tournament lacks sensor data, so any metric comparison should be read as a hypothesis, not a conclusion. For a consultant, writing that line is harder than writing a long analysis table. It forces me to admit my tools have limits, and those limits are not the model's fault. I believe in models, but I pray before every match — because football, and badminton too, is not an equation. Some days the tables are full and say nothing. Some days they are empty and that is the most honest moment. What I want to carry into next week is not a new metric but a habit: before asking what data says, ask whether it is qualified to speak at all. If the answer is no, the kindest act is silence, rewinding the footage, and letting the eyes do their work.