When the Data Goes Silent: The Craft of Reading Numbers and the Trap of Empty Cells

On the evening of March 20, 2026, I sat in a video meeting room of a club in Hai Phong, reviewing footage of the match we had just lost. On the screen, the statistical table appeared complete: points, rebounds, assists, shooting percentages. Every cell had a number. But when I asked the data department to export a metric on the space players create when they do not hold the ball, the system returned a single line: "no data." That action had not disappeared at all — it happens hundreds of times per game. It is just that no one had installed a camera and written an algorithm to record it.

That moment reminded me of something that has followed me throughout my years in this craft: what is not recorded is often more important than what is. It also opened a far more dangerous question: what happens when the data goes silent, and do people tend to fill that gap with conjecture?

Over eighteen years observing the basketball industry, from evenings in local arenas to professional analytics rooms, I have realized that the craft of reading numbers is not about reading the numbers that exist. It is about discovering the cells that remain empty — and knowing clearly that an empty cell is not a zero, but a reminder that we do not yet have enough data to judge.

At the end of this article, I will not offer a closed conclusion. I will leave an open question, because that is the only way to keep the craft of reading numbers honest.

The explosion of basketball data began in the 2026-2026 season, when the NBA installed the SportVU tracking system across all arenas. For the first time, the position of every player and the ball was recorded 25 times per second. By the 2026-2026 season, Second Spectrum took over and increased the sampling rate, turning each game into millions of coordinate data points. People joyfully declared that basketball had become the most thoroughly measured sport.

But "thorough" is a dangerous word. Every measurement system has its edges. SportVU records position, but not intent. Second Spectrum counts touches, but cannot count the hesitation of a thousandth of a second before a player releases a pass. And in leagues without the budget for tracking systems — including most basketball leagues in Southeast Asia, VBA among them — that edge is far wider.

When I began working as a data consultant for clubs, I noticed a paradox: the less data there is, the more people trust it. In a league with only a few basic metrics, each number becomes weightier, and people readily assign it an explanatory power it does not have. That is why in smaller leagues a player's three-point percentage can decide an entire contract, while in the NBA a professional analytics department will immediately ask: how many games did he shoot well, against which opponents, and with whom setting screens?

I once tracked a player in a regional league who shot 44% from three in a single season. The number was so beautiful that a club in a higher league nearly signed him after just two film sessions. When I checked again, I found three things: first, he attempted only 2.1 threes per game on average — too small a sample to conclude anything; second, most of those came from the right corner, where the opponent's defense was weakest; third, no one had recorded how poorly he shot when defended tightly. Three important cells were empty, and all three lay outside the official box score.

That is the essence of my work. Numbers do not lie, but the people who choose them do. And the person who chooses the numbers is not only the one deciding which metrics go on the board — he is also the one deciding which metrics stay blank.

Let us start with the largest empty cell in every basketball box score: defense.

In a game, the box score records everything a player does with the ball. But defense — the work that occupies nearly half the time on court — leaves almost no trace. There is no column for "correct rotations," "closing out drives," or "applying pressure that forces a sideways pass." Those actions exist only in the eyes of the viewer and vanish the moment the game ends.

That is why I always tell clubs: if you sign a defensive player based only on the box score, you are buying a name, not a skill. Modern defensive metrics such as Defensive Rating or Defensive Win Shares are inferential — they are calculated from collective outcomes, not individual actions. A good defender in a bad system will have a bad number, and vice versa. The empty cell here is not empty because the action does not exist, but because the measurement method has not reached it.

I once watched a VBA club lose seven straight games, and the coaching staff concluded the problem was their offense. The box score showed they shot poorly. But when I sat down to review the footage, I counted that on 62% of their possessions, their ball handler received the ball more than nine meters from the rim, because no one created space behind him. The box score recorded the missed shot — it did not record why that shot was so difficult. The empty cell was in spacing, but people went to fix the finishing.

The second empty cell is shot quality. In soccer, this problem was solved long ago with expected goals (xG). A shot from close range at a wide angle is worth far more than a long-range shot, even though both are recorded in the same "shots" column. Basketball adopted that model later, and adopted it incompletely.

Today the NBA has an expected-points metric based on shot location, defender distance, and play type. But that metric still leaves out a massive variable: the shooter's physical state. A three at minute 44, after four consecutive screens, has a very different success probability from a three at minute 6 when the player is fresh. No system is tracking that variable fully at the level of each individual shot.

Every number is a confession, if we are patient enough to listen. But a confession only rings out when we admit that something remains unrecorded. Stephen Curry's three-point rate once shattered every convention of traditional basketball, and what made it special was not the final number, but that the empty cell — "no one shoots threes from that distance at that frequency" — was filled by him before any analyst could record it.

The third empty cell is the lesson from the bubble. In 2026, when the NBA staged the remainder of its season in the Orlando isolation zone, my team and I analyzed data from 200 matches in the Portuguese and Danish leagues to understand what happens when soccer returns after a pandemic. The result startled us: central midfielders' running distance dropped 9.7% in the first month, but the number of long forward passes rose 13.2%.

That data said something no box score records: when the crowd disappears, players play differently. Pressure from the stands is an invisible but real variable, and its absence leaves a trace in every number. Basketball in the 2026 bubble was the same. Those games had markedly different pace, free-throw rates, and stress levels from a normal season — and anyone using bubble data to predict the following season is comparing two different worlds.

When the court is empty, only the data whispers the truth. But that truth can only be heard if we accept that we are measuring something that never existed before.

The fourth empty cell is the challenge of scouting from data-poor leagues. This is where I spend most of my working time, and also where the temptation to fill empty cells is strongest. A club with a limited budget cannot buy premium tracking data, so it must rely on the human eye and rough metrics. The most common mistake is not misreading the data — it is turning an empty cell into a favorable assumption.

I call this phenomenon "filling the cell with hope." When there is no data on a player's defensive ability, the scout will implicitly assume he is "probably decent." When there is no data on his locker-room character, they will assume "probably fine." Each such assumption is an empty cell sealed with something lacking evidence, and accumulating dozens of such assumptions, you build a contract around a player who never existed.

My job is not to fill the empty cells, but to point them out and say plainly: this is unknown. An honest scouting report must have clearly marked blank lines, not a table stuffed with numbers that looks complete.

The fifth empty cell, and the one that troubles me most, is the numbers that never appear. In a game, there are actions that have never entered any system: where a defender ran to block a pass the opponent was about to make, who was the last to leave the court after the game ended, who stood up to pull a teammate up after a collision. Those things build a team's culture, but there is no column to record them.

When the court is empty and the cameras are off, I often stay behind. That is where I learn the most — not from the box score, but from what remains after the box score has been printed. A team can win through numbers, but it builds its identity from the empty cells no one bothers to record.

Here I must argue against myself. If I say empty cells matter, it is easy to fall into the opposite trap: believing that every empty cell contains a hidden truth. That is not so. Some cells are empty simply because the action is not worth measuring. The problem is that we must distinguish which empty cells are worth digging into and which should be ignored — and that is judgment, not a formula.

Correlation is not causation. A player with high advanced metrics is not necessarily the one who makes his team win; perhaps he is just in the right system. A team that shoots threes well does not necessarily win because of threes; perhaps they win because the opponent's defense is too weak in that area, and the threes are merely a consequence. When we fill an empty cell with correlation, we turn a relationship into a cause — and that is the fatal error of the craft of reading numbers.

I once thought I was right. Qatar taught me I was wrong. In November 2026, based on a prediction model combining four years of qualifying data, I declared that Argentina would beat Saudi Arabia with 94% probability and a minimum score of 3-0. The result: Saudi Arabia won 2-1 through an offside trap ten times in the first half. My model was not wrong in mathematics — it was wrong in that I had filled an empty cell named "match conditions" with the assumption that temperature and altitude did not matter.

After that failure, I spent two weeks re-watching 47 matches in Gulf-region tournaments over ten years to understand what I had missed. The lesson was not "don't predict," but "don't seal an empty cell with confidence." Since then, every analysis of mine comes with a confidence interval, and every prediction is written in the language of probability rather than assertion.

Data is a mirror; do not get angry when it reflects an ugly truth. And the ugliest truth the craft of reading numbers must face is this: most of what we think we know is actually empty cells filled with conjecture, repeated often enough to look like truth.

So what is the signal for the next cycle? I do not expect a new set of metrics to appear and solve every problem — new metric sets are not born in offices, but in crises. What I hope for is that clubs begin to treat the marking of empty cells as a mandatory part of the process, rather than a weakness to hide. A report with three clearly noted empty cells will be more useful than a report full of numbers that no one dares admit is incomplete.

For fans, the signal is the same. When you read a statistical table so perfect that there is no room for doubt, ask yourself: who chose these numbers, and what empty cells do they want you not to see? Because in basketball, as in everything that is measured, the truth often lies in the blank — where no one wants you to linger too long.