The Empty Deck: How Sport Makes Nine-Million-Dollar Decisions on Data That Isn't There

March 2026, an hourly meeting room in Seaport, Boston. A Major League Soccer club is weighing a 9.2 million dollar outlay for a 24-year-old forward. The dossier runs 46 pages, full colour, plastic-bound. Page 12 is an xG-per-90 chart built on six matches. Page 19 covers salary structure. Page 31 is the medical report. Page 37 carries the heading 'Missing Data', and every cell beneath it reads N/A. The meeting lasted 71 minutes. The decision was taken at minute 68. They spent 9.2 million dollars after reading 45 pages and one blank one. I tell this story because it is familiar. In four years of reporting I have sat in at least eleven meetings with the same architecture, and in seven of them there was a page where the data had been left empty while the scaffolding stayed intact: right heading, right metric name, right table format, no content. That empty frame has a strange power. It makes a room feel that everything is under control, even when the only thing under control is the presentation. I call it the empty deck. In modern professional sport it is more common than people assume, and it is more dangerous than a wrong number. ONE SPREADSHEET AND ONE QUESTION To understand why blank pages exist, you have to look at how the industry's data flows changed over the past decade. In 2026 I was sixteen, a high school student in Boston, and I started a blog called MLS Moneyball on Medium. My only data source was the MLS Players Association salary release, published twice a season, listing base salary and guaranteed compensation for every player in the league. No tracking data. No xG. No forecasting model. Just a spreadsheet, one question and a great deal of time. I took apart the New England Revolution wage bill. The result: 71 per cent of the club's salary budget sat with five players, against a league average of 55 per cent. The post, titled 'New England Is Betting in the Wrong Place', drew 12,000 reads in a week and was shared by a local journalist. That led to a freelance contract with an independent sports outlet. Looking back, what I learned from that first piece was not the 71 per cent. It was that the same dataset can produce two opposite conclusions. I read 71 per cent as evidence of a mispriced bet. Someone else could read it as the rational outcome of a star-concentration strategy, and they would be right if the club won. Data does not lie, but it needs someone who knows how to listen. In 2026, aged seventeen, I watched the France-Uruguay quarter-final in Nizhny Novgorod on 6 July from behind a screen. France won 2-0, Raphael Varane headed the opener, Antoine Griezmann doubled the lead after a handling error by goalkeeper Fernando Muslera. What I recorded was not the goals. I counted 27 pressing actions from France, above the tournament average of 19, and a transition time from defence to attack 0.8 seconds faster than Uruguay's. I filed 'How Deschamps Digitised the Press' two hours after the final whistle. It was shared more than 3,000 times. From then on I built a working habit: every analysis opens with a striking number, sits on a transition metric in the middle, and closes with a firm conclusion. I keep a template file open on my machine so I can publish within 90 minutes of full time. In 2026 the pandemic swept through. MLS stopped. I was twenty, a second-year student interning at Boston Sport Analytics, handed a scenario model for FC Cincinnati. If the club played 12 matches behind closed doors, it would lose 14.2 million dollars in ticketing and 2.8 million dollars in food and beverage. I presented to the board, recommending a 20 per cent cut to academy costs and a delay on signing a foreign striker. My report was passed up to the league as a reference document. Empty stadiums did not kill football, they exposed who was living off it. My model had three variables: matches, seats, days. Three. Yet those three were enough to change a club's personnel decision. I began to believe that what sport lacks was never data, but people who know which three variables to keep. In 2026, aged twenty-one, still a student, I was freelancing for a transfer outlet. Through a scout I learned Arsenal were prepared to pay 7.5 million dollars for New England Revolution goalkeeper Matt Turner, with a 15 per cent sell-on clause. The selling club denied it outright. I held my ground and published an exclusive headlined 'Sources Confirm Turner to Arsenal'. Three days later Arsenal made it official, and the fee matched to the last digit. I also remember what nearly happened. My source could have misheard, or read a document that had been edited. I called two more sources before publishing. One confirmed, one stayed silent. Since then I run a three-step verification routine: check the source, cross-check both sides, and state the confidence level inside the copy. THE REVENUE STRUCTURE CHANGED On 14 June 2026, MLS announced a global media rights deal with Apple worth 2.5 billion dollars over ten years, starting with the 2026 season. The MLS Season Pass package turned the entire league into a single paid product on one platform. That same year, the Leagues Cup expanded into a competition involving every MLS and Liga MX club, reaching 47 teams. In the summer of 2026, the 32-team FIFA Club World Cup was staged in the United States. In June 2026, a 48-team World Cup kicks off across three North American countries with 104 matches. On the esports side, the money followed a different trajectory through the same logic. Game publishers own the match data, the event rights and the ability to rewrite the rules with a single patch. FaZe Clan listed via a SPAC in July 2026 at a valuation around 725 million dollars, then slid repeatedly, received a Nasdaq delisting notice, and was eventually absorbed by GameSquare in an all-stock deal completed in 2026 at a fraction of its peak. 100 Thieves once raised capital at a reported valuation near 460 million dollars in 2026. That contrast is where every empty deck begins. Data grows exponentially, expectations grow exponentially, and the capacity to read data grows linearly. HOW AN EMPTY DECK GETS MADE An empty deck is rarely the work of one careless person. It is a production line. That line has six stations: raw data, vendor, dashboard, analyst, slide, board. A club buys its data package per seat. The dashboard is designed to look good on a big screen. The analyst has three days to file. The slide must fit one page. The board has 70 minutes. Break one station and the content vanishes while the frame still gets printed. And because the frame is printed, it automatically becomes part of the meeting. In American soccer, positional tracking data is now standard in most stadiums, measuring sprint speed, distance covered, and the spacing between lines. In esports the data is finer still: every champion pick and ban, every position, every second of a teamfight. Both platforms share one fatal weakness: sample size. A balance patch lands on a Wednesday. By Saturday there are six professional matches on the new build. A champion win-rate table built on those six matches is printed, framed, and becomes the basis for a ban. Six matches. I have watched this happen more than ten times in three years covering esports. In football the variant is per-minute metrics. A player logs 412 minutes in a season, scores three goals, and is sold for 9.2 million dollars. The analyst writes 'xG per 90: 0.41' without noting the 412 minutes, because that footnote would ruin the story. The data was not falsified. The data was presented without its denominator. That is the most sophisticated form of lying in this industry: telling the truth while hiding the sample. THE MOST DECEPTIVE NUMBER ON THE PITCH Two examples are enough to show where the problem sits. On 1 July 2026, Spain held over 75 per cent possession against Russia in the round of 16, took more than 20 shots, and went out on penalties after a 1-1 draw. On 6 December 2026, Spain held 77 per cent possession against Morocco, scored none, and went out on penalties, 0-3. Two matches, two tournaments, two generations of players, one outcome. Possession share measures a team's patience, not its capacity to win. A side that plays 600 sideways passes in its own half reaches 63 per cent possession without creating two clear chances. Meanwhile a side with 37 per cent of the ball but 1.8 xG has generated more value. I tested this against public MLS data during my internship: the correlation between average seasonal possession and final points was notably weaker than the correlation between expected goal difference and final points. Possession is a flattering mirror. xG is a mirror that hurts. And coaches still put possession on the screen in press conferences, because it is the only metric the audience understands instantly without explanation. In a 70-minute meeting, the easily understood always beats the correct. WHEN VAR MOVES THE ARGUMENT INDOORS Refereeing technology is another data pipeline, and it behaves exactly as described above. VAR was deployed fully at the 2026 World Cup and immediately produced a record number of penalties awarded in a single tournament. The promise was explicit: technology would reduce controversy. Reality diverged. The controversy did not disappear, it relocated. It left the grass and moved into a room full of monitors, where only a referee and a technician sit and the crowd sees nothing but a machine-drawn line. The result is a new category of decision: one that looks evidence-based but cannot be verified by anyone outside the room. The frame is captured at 50 frames per second, the moment of contact is pinned to within a few hundredths of a second, and the shirt-line is drawn by an algorithm that is not public. Combine those three and you get a firm conclusion delivered in technical language, standing on a grey area. A goal disallowed for a toe. A penalty given for a handball at 40 centimetres. The same incident, two referees, two conclusions. Fans are not reacting to technology. They are reacting to technology that promised certainty and delivered ambiguity, except this time the ambiguity is wearing a blazer. As a writer I took one lesson from those nights: when a system presents its conclusion in the format of precision, readers tend to skip the verification step. The frame has done the hardest work for them. THREE VARIABLES AND THREE HUNDRED VARIABLES There is a paradox in how clubs make decisions. When I modelled for FC Cincinnati in 2026 I had three variables. Those three produced one clear, actionable conclusion: cut academy costs by 20 per cent, delay the foreign striker. The board understood it in four minutes. But when a club hires an analytics firm with 300 variables, the output is usually a report nobody dares act on, because every conclusion arrives with a qualifier attached. The report is thicker, the decision takes longer, and eventually everyone falls back on the head coach's instinct. So I believe the value of analysis lies not in the number of variables but in the ability to compress. A good model must answer one question: if you could keep only three numbers, which three. Whoever can answer that is the person who gets a seat in the room. SELL-ON CLAUSES AND THE ART OF COSMETICS Finance has its own empty decks, and they cost far more. In January 2026, Enzo Fernandez moved from Benfica to Chelsea for what was then a British record fee, around 106.8 million pounds. In August 2026, Moises Caicedo moved from Brighton to Chelsea for around 115 million pounds, breaking the record set months earlier. Chelsea signed long contracts, in some cases running to eight years, and amortised the transfer fee across each season. The practice sharply reduced the annual book cost while the paper value of the squad stayed intact. UEFA later closed the loophole by capping amortisation at five years. But before that, a large number of decisions rested on an accounting version of the truth in which a 115 million pound outlay looked as light as a 14 million pound one. A speaking number beats a dressed-up contract. But a cherry-picked number lies better than a contract with no numbers at all. In esports the equivalent is valuing an organisation on social media following. FaZe Clan reached a valuation in the hundreds of millions largely on audience size rather than cash flow from core operations. When the advertising cycle softened, that valuation evaporated faster than any other investment in the sector. Modern football does not win on the pitch, it wins in the meeting room. And the meeting room wins with numbers that nobody in it has personally checked. GREY ZONES AND THE PRICE OF CERTAINTY One risk category is systematically underpriced across sport, and it sits directly on this theme. When a club makes a decision on a six-match sample, the risk is not that the conclusion is wrong. The risk is that the conclusion is coincidentally right, and is then used as precedent for the next ten decisions. Mistakes can be corrected. Precedents cannot. At governance level the same thing happens with new rules. A regulation is written to handle one specific incident, then applied to entirely different incidents, and each application generates a fresh grey zone. Rules do not shrink grey zones. They move them somewhere fewer people can see. At financial level the biggest risk is unrecognised losses: delayed wages, suspended sponsorship payments, an owner stopping the funding. Those signals almost always appear before a club dissolves, and they are almost always ignored because they do not appear on the league table. That is why I learned to read financial statements before reading the standings. THE COUNTERINTUITIVE ANGLE What should worry us is not the blank data pages. What should worry us is how safe they make us feel. A formally correct frame creates the sensation of work done. That sensation is strong enough to replace the work itself. When a board sees an xG chart with axes, labels and colours, nobody asks about the sample. They move to the next item. Which is why I argue the industry's problem is neither too much bad data nor too little good data. The problem is that we have not learned to tell the shape of analysis apart from analysis itself. Another counterintuitive point: data does not reduce uncertainty. It relocates it. Once, the risk in a transfer lived in a scout's eye, where anyone could argue with it. Now it lives inside a model's parameters, where nobody can argue because nobody has access. Uncertainty has not gone away. It has become less visible, and that is a step backwards for governance. Tactics are what you see, the market is what you must guess. The most dangerous moment is when both are presented on the same slide, in the same typeface, so the reader assumes both can be seen. Fans leave the stands, but the money never stops moving. When a club sells a player and nobody understands why, the answer usually sits in a spreadsheet no reporter is allowed to open. That is why I always ask for the number, even when the answer is a blank space. WHAT I CARRY FORWARD From MLS salary sheets to World Cup tactical maps, the journey of an observer does not end in finding the right figure. It ends in knowing when the figure does not exist. Tomorrow, another club will meet. Another analyst will file a report at 2am. Another board will look at a page with the right heading, the right formatting, and empty cells underneath. And if nobody in that room asks for the denominator, the decision will still be made, still be signed, still be announced in a single short line. A gap in a club's data file may be nothing more than a technical error. But a gap in a meeting is a choice. And that choice is being repeated more often than we think, from leagues with billion-dollar budgets to stadiums with four thousand seats. I started with a spreadsheet, and I still end with questions.