The Empty Stadium, the Empty Spreadsheet: When Tennis Analytics Forgets the Human Being

On the screen is a dense spreadsheet. The left column carries two players' names. In the middle, dozens of metrics stand in neat rows like soldiers at attention: first-serve percentage, points won on first serve, points won on second serve, net points won, winners, unforced errors, points won in rallies longer than nine shots. Every cell is filled. Not a single dash. And yet I sat there for two hours, rewinding point after point, and by the end I still knew nothing more about the man who had won that match. I have felt this before. It is not the feeling of a rookie. It is the feeling of someone who has been in this trade for more than twenty years, who once ran on an NCAA track, who once sat in the third row of a suburban Challenger court, and who once filed a story at three in the morning with coffee that had long gone cold. It is the feeling of realizing that your strongest tool has become the thing blocking your view. The stadium is empty, but I can hear the heartbeat of a whole generation. The problem is that heartbeat is not in the spreadsheet. The 2026 season has closed and left behind a men's Grand Slam year that is rare in tennis history. Jannik Sinner won the Australian Open, Carlos Alcaraz reversed the tide at Roland Garros after saving three championship points, Sinner answered at Wimbledon, and Alcaraz took the US Open. The four biggest titles of the year were split evenly between two men. This is the first time since the Federer – Nadal – Djokovic era that two players born after 2026 have divided all four Grand Slams of a single season, a sign that the generational handover is complete. On the women's side, the picture is more scattered. Madison Keys won the Australian Open. Coco Gauff took Roland Garros. Iga Swiatek claimed Wimbledon. Aryna Sabalenka closed the year with the US Open title. Four different champions at four majors, something the women's game has not seen often in the past decade. I bring up these results not to summarize the season. I bring them up to pose a harder question: with all that data, all those metrics, all those forecast models, why do we still routinely fail to predict who will win, and more importantly, why do we still fail to understand why they won? In recent weeks, I had the chance to look again at the analysis pipeline that some sports newsrooms are using. The pipeline has several layers: one layer that deconstructs text to extract core information, and a deeper analytical layer built on what was extracted. It sounds scientific. It sounds modern. But one day, when the first extraction layer returned an empty result, the entire analytical system behind it also returned zeros. Nine dimensions of analysis, from technical-tactical to data and form, from tournament systems to the wider tour landscape, from rules compliance to team management, from risk to media narrative, all carried the same line: insufficient information to assess. I read that result three times. And what made me pause was not the technical failure. It was its structure. Imagine an analytical system designed to answer dozens of questions about a player. It comes with a framework to assess surface adaptability, a framework to measure nerve at decisive points, a framework to analyze point structure, a framework to screen injury risk, a framework to read media cycles. But when the input is empty, all those frameworks remain standing, beautiful, hollow. That moment looks exactly like a stadium without spectators: the stands are still there, the seats are still there, the lines are still there, only the people are missing. Amid endless data, I always look for a human being still breathing. But if there is no human being inside the data, there is nothing to find. That is the first lesson, and perhaps the biggest one the tennis analytics trade needs to learn. Because we live in an age when tennis data has become richer than ever. Hawk-Eye is no longer just for judging whether a ball is in or out. It records a player's return position on every point, contact height, spin rate, ball trajectory, distance covered. Independent analytics platforms such as Tennis Abstract, and the data services of the ATP and WTA, give fans metrics that twenty years ago only a team's coach could access. You can look up a player's win rate when serving to the outer corner on break point, or their net-approach rate in a deciding set. The problem is not a shortage of data. The problem is that there is so much of it that people can be lulled into the feeling that they already understand. I remember an evening in Melbourne a few years back. After a five-set match in heat above forty degrees Celsius, I went down to the backstage area. There, a player who had just lost sat curled on a chair, head down, hands wrapped in towels. The match statistics showed he had won more points, created more break chances, and hit more aces. On paper, he won. In reality, he lost. And the only thing that explained the difference was not in any spreadsheet — it was that he had lost a sliver of nerve on three key points, each lasting only a few seconds, which no model can measure. That is why I always tell younger colleagues: a number only means something when you know who it was born from, in what circumstances, and what that person had gone through before. Take a concrete example from the past season. The Roland Garros final between Alcaraz and Sinner is one of the most analyzed matches in recent history. If you read the statistic sheet, you will see Sinner dominating at times, leading in the early sets, and holding three championship points. But the sheet does not tell you anything about how Alcaraz changed his return position in the fourth set, or how he started hitting deeper into Sinner's backhand to extend rallies, or how he accepted higher risk at moments when reason said to play safe. Those decisions were the product of years of accumulation, of the memory of similar defeats, of knowing an opponent down to his smallest habits. A model can simulate thousands of times the situation where Alcaraz faces championship point down. But only one man actually stood there, and he had only one chance to choose. This is the boundary that I believe matters most in our trade today: the line between predicting and understanding. Predicting is the work of the model. Understanding is the work of the journalist. When we do only the predicting part, we produce something that can be very accurate but empty of meaning. When we do only the understanding part, we produce something emotionally rich but short on evidence. Sports journalism needs something in between. And that is when I think of another, less discussed problem: the fragility of a system when its input loses contact with reality. A tennis analytics system can be built with hundreds of indicators, dozens of evaluation dimensions, and a very tight concluding framework. But if it has no mechanism to recognize that it is analyzing a subject that does not exist, it will return a result that looks valid — and that is the most dangerous moment. Fortunately, in the case I witnessed, the system spoke the truth: insufficient information to assess. But I know there are systems that cannot do that. They fill the gaps with assumptions, with inference, with conclusions that sound convincing but are in fact only echoes of the writer. In tennis, we call that a double fault. In journalism, we should call it professional malpractice. The current period is the transfer window and the run-up to a new season. This is when the market overflows with rumors: this player changes coach, that player adjusts his schedule, a new fitness team comes in, a new sponsor appears. In that current, the most valuable thing a sports journalist can give readers is not a list of rumors but a filter. That filter needs three things: contract structure and agent moves, real injury status rather than official statements, and the structural logic of the squad — who plays with whom, in what position, in what system. It sounds dry. But behind every contract clause is a human being deciding where he will live for the next twelve months. I once sat in Moscow for several days just to interview assistant coaches, only because a question would not leave my head: what makes a person run more than twelve kilometers in a match and still stay lucid? When I found the answer, I realized it was not in the fitness metrics, but in something very human: someone on the staff had spent years teaching that young player how to save every step. That is the most beautiful paradox of this trade. The contrarian angle I want to raise here is this: the sports analytics industry is not weak because it lacks data. It is weak because it is sometimes too confident that it has data. An empty model can still look full if the user does not check the source. A conclusion built on a single metric can still look professional if the presenter knows how to arrange the slide. And the most dangerous thing is when professional form masks empty content. There is another silence I think about as I write these lines. It is the silence of players who have no personal analyst, no data team, no Hawk-Eye coach. They compete at Challengers, in Grand Slam qualifying, on courts where the stands hold only a scattering of people. For them, data is not a tool but a luxury. They learn by eye, by ear, by memory. They write in notebooks. They rewatch footage on an old phone. If the tennis analytics trade really wants to prove its worth, perhaps it should start there — with the people who appear in no model. That is also the kind of story I have always pursued: not rushing in when their names are already on the honors board, but seeking them out before the world knows their names. I have met many players over the years. Very few of them had a beautiful statistic sheet. But almost all of them had something no spreadsheet records: a reason to keep standing on court the next day, when no one else believed in them. That is why I write about emptiness. About a stadium at four in the morning when only the security guard and one player hitting serves remain. About a locker room emptied out after the season ends. About an empty seat in the stands belonging to a mother who could not come. And it is also why I tell you about an analysis session that returned all zeros. Because sometimes the most important thing data teaches us is not a number, but the truth that not everything can be quantified. Tennis is a sport of points. Every point is recorded. Every set has a result. But between those two columns of points, there are stretches of time the machine cannot measure: the moment a player decides not to give up, the moment a coach chooses silence instead of speech, the moment an athlete remembers his family ten thousand kilometers away. Those moments decide matches more than any metric. I am not calling for abandoning data. I am calling for using data as someone who opens a door, not someone who guards it. Data is memory, not a verdict. It helps us remember on which point Alcaraz hit a second serve, but it does not help us feel how his hand trembled before that point. A new season is coming. There will again be long statistic sheets, forecast models, carefully packaged conclusions. Many people will read them and believe. I only hope that, among those lines of data, someone remembers that behind every number is a person still breathing. In the failed analysis I mentioned at the start, there is one detail I have not told. After the system returned all zeros, the person in charge of the pipeline did exactly what should be done: he stopped, stated clearly that there was insufficient information, and proposed returning to the first step to check whether the source document actually existed. That was the bravest act I have seen in an analysis room. Because admitting you do not know — in an industry where everyone wants to appear as the one who knows first — is the hardest thing. The gold cup is not at the finish line, but at the turns we never planned. And perhaps the best answer the tennis analytics trade can give its audience is not a perfect forecast, but an honest question: who are we really looking at? When the stands are empty, the truest voice comes from an old phone. And when the spreadsheet is empty, the only thing left to do is go back and find the human being.