The Basketball Analytics Industry Is Fooling Itself With Empty Data Tables

On the night of March 14, 2026, on the most-watched NBA commentary show in the United States, the host held up a stat sheet: a guard with 28 points, 9 assists, and a true shooting percentage (TS%) of 61.4%. The whole studio nodded as if they had witnessed a perfect performance. No one asked where that sheet came from, over how many games it was sampled, or whether it reflected anything real. I have followed professional basketball for 48 years, and I have never seen the analytics industry sink into such blatant self-delusion. The next day, I spent four hours tracing the source of that sheet. The result: it was built from a three-game sample, with no cross-checking, no notes on opponents, and no context on physical condition. An empty sheet presented as scientific evidence. Over the past decade, professional basketball entered the era of motion-tracking data. Each arena installed dozens of cameras recording every step, every burst of speed, every shooting angle. Teams spend millions of dollars a year hiring analysts, buying data tables, and building predictive models. The media is no bystander either: every broadcast now comes with a flood of advanced metrics, from true shooting percentage (TS%), usage rate (USG%), to per-half plus-minus (+/-). Fans have been taught that to understand basketball, they must understand the numbers. The problem is this: when data becomes the standard of evaluation, the pressure to produce data rises too. Stat platforms race to release new metrics, agents use data to negotiate contracts, and coaching staffs use data to justify decisions. In that churn, data quality is left behind. I once sat in a meeting room of a team in the western United States, where an assistant coach presented a 40-page scouting report, but not a single page stated where the data sample came from. When I asked, he laughed: "People just need to see a pretty number, they don't ask where it comes from." The story of the empty sheet on March 14 is not isolated. It is a symptom of a disease spreading through the basketball analytics industry. I call it the "three-game sample syndrome" — the habit of taking a few games to represent an entire season, then presenting it as universal truth. Take the metrics of distance traveled and speed bursts. These metrics are packaged as measures of effort. The player who runs the most is praised as a "warrior," the one who runs little is criticized as lazy. But a guard who runs 4.5 km per game while his team controls the ball 70% of the time covers most of that distance running for nothing. He runs to hold position, not to create value. Effort metrics, detached from tactical context, become pretty numbers to sell to the audience. Distance traveled and speed bursts are measures of fatigue, not of effectiveness. Then comes plus-minus (+/-). A player enters the game, his team outscores the opponent, and his plus-minus looks good. But if he enters at the same time as four of the opponent's bench players, that number reflects a weak opponent, not his own quality. Not to mention the teammate factor: a good three-point shooter will inflate the assist numbers of the player who passes to him, even if that pass was a simple play. Every advanced metric, in the end, is a function of other variables, and when we forget that chain of dependency, we read data as if reading a verdict. I have spent many years interviewing low-tier league coaches — men with no budget for an analytics department, no team of experts, but who must make personnel decisions every week. They taught me something the big television shows forget: data does not score, but data is quietly rewriting history. A coach in the second tier of American basketball once told me he only trusted three data sources: game film, his assistant's eyes, and the player's own feel. Those three sources had to match. If the metric said one thing and the film said another, he dropped the metric. Meanwhile, at the top level of the industry, people do the opposite. Commentary shows invite "number gurus" on air to read data tables, and audiences trust them because they seem knowledgeable. But many of those gurus have never watched enough games to understand context. They read numbers the way one reads a stock ticker. And when data is misread, it becomes a tool to deceive fans, even if the reader does not intend it. The more serious problem lies in the transfer market. Player agents understand that a pretty stat sheet can raise a contract's value by tens of millions of dollars. They select the most favorable metrics, omit the unfavorable ones, and present them to team leadership as a flawless résumé. In more than thirty years of following this market, I have seen no small number of fat contracts signed on data samples so small they are meaningless. Agents are the largest hidden cost of the transfer market; the noise they create distorts a player's true value. A player valued at 30 million dollars a season may rest only on six weeks of breakout play, while the rest of his career is quite ordinary. Then there is the injury factor. Medical confidentiality leaves fans and media completely blind. Teams only announce injury status when it benefits their image or commercial value. A star may be carrying a serious injury but is still listed as "minor injury, game-time decision," and any analysis based on his playing data becomes meaningless because we do not know his true condition. Medical confidentiality blinds fans and media; clubs only announce injuries that benefit the stock price. I once wrote about a player said to be "finding his form again," while my insider source confirmed he had to take painkilling injections before every game for two months. I remember a playoff game in the 2026 season. A team led by 20 points after three quarters, with dominant possession and superior advanced metrics. Then in the fourth quarter, they lost the lead. The post-game box score still had them winning in most categories. That is the lesson about the limits of data: it measures process, not result. Possession is an illusion; winning is the bare truth. I know I am going against the current, and I may be wrong. There is a legitimate argument for using data: it helps us remove bias, measure what the eye misses, and provide a common standard for comparison. A metric like true shooting percentage (TS%) is genuinely useful when used correctly, in a large enough sample, with full context. The problem is not the data, but the way people use it as a shield to avoid thinking. But if I am wrong, then my error lies in admitting that the analytics industry has a fatal blind spot: it evaluates the process of producing data without evaluating the quality of the data. A metric computed from a three-game sample and one computed from an entire season can look identical on screen, but their information value differs by a world. When a platform releases a new number without publishing its method, without stating the sample, without stating the source, that is empty data. And an industry built on empty data is fooling itself. I also do not believe all stat sheets are worthless. My faith lies in the three-source rule: any conclusion about a player must be cross-checked through at least three independent sources — game film, a large-sample dataset, and the voices of those inside the locker room. If those three do not match, the conclusion must be postponed. In an industry racing for speed, the slowness of verification is a luxury, but it is also the only thing that preserves credibility. There is another angle I want to raise: fans also bear part of the responsibility. They like shocking numbers, like hot takes, like the feeling of being handed a decisive conclusion. That demand creates the incentive for shows to produce empty data. If audiences stopped rewarding claims without evidence, the market would correct itself. But that requires a patience that the age of fifteen-second clips does not encourage. In this regular season, when dozens of games and thousands of metrics update every night, I propose a small experiment. Next time you hear someone read a number on television, ask yourself three questions: how large is this data sample, who is the source, and does it match what I have seen with my own eyes? If the answer to all three is vague, you are listening to an empty sheet. The basketball analytics industry will not die from a lack of data. It will die from too much data and too little truth. And in that race, the ultimate winner will not be the one with the most data, but the one who knows which data is trustworthy. Once we learn to reject empty numbers, this industry will have a chance to truly mature.