One Saudi Arabia, Two Datasets: When Oil and Tennis Share the Same Signature

On Thursday morning, I opened the spreadsheet tracking the pressing index of male players on the ATP Tour and found a column of data that did not belong where it sat. Twenty-six rows. Brent settled at $105.64 a barrel, down 19 cents from the previous session. WTI at $102.10, down 33 cents. Not a single player name. Not a single tie-break. Only the Strait of Hormuz, which once carried one-fifth of the world's oil supply, Oman's Sohar port, and two damaged pumping stations on the East-West pipeline with an unclear repair timeline. I nearly wrote a complete tactical breakdown for a match that does not exist. The column I thought was the number of net approaches turned out to be air strikes. The word damaged I took for an injury turned out to be infrastructure. Pipeline I read as a well-drilled combination turned out to be an oil conduit. In nine years of following tennis, I have never met a dataset so confident yet so lost. But on closer inspection, that dataset was not lost at all. It sat exactly where it belonged. The thing that placed it wrongly was the label stuck on it. And the one who nearly misread it was me. Every day, sports newsrooms in Australia, where I work, process thousands of items through an automated classification pipeline. Each article enters the system with a field called the domain label. Tennis. Cricket. Rugby. Football. That label decides which editorial desk receives it, who verifies it, and which bulletin it appears in the next morning. When that label field is filled in wrongly or left at a default value, a whole chain of consequences unfolds in silence. An article about oil prices is pushed to the tennis desk. A writer like me opens it, sees numbers everywhere, sees professional jargon everywhere, and if he is not alert, he will use it to build a story. The most worrying part is that the terminology in that article sounds very much like tennis language. Attack. Damaged. Pipeline. Flows. Spike. Five words, five times I could have assigned them to a rally if I had only skimmed. A damaged pipeline sounds like an injured player. A blocked flow sounds like a broken run of points. A price spike sounds like a lob. This problem goes beyond the mechanical layer. It lives at the human layer, in the way we read data and then assign it the meaning we want to see. And in modern tennis, when the same name, Saudi Arabia, appears at both ends of the data world, that boundary becomes more fragile than ever. To understand why, I return to the very name that made me stop mid-morning: Saudi Arabia. In that oil dataset, Saudi Arabia is an exporting state, seeking to move cargo through Oman to soothe supply fears. In my tennis dataset, Saudi Arabia is an investor. The same entity, two roles, two entirely different sets of metrics. If a careless analyst merges those two sheets, the result is an unsalvageable mess. Saudi Arabia is present in professional tennis in a way without precedent. Its Public Investment Fund has become the naming partner of the ATP rankings, turning a pure points system into a branded property. The Next Gen ATP Finals, the tournament for the next generation with entirely different experimental rules, is held in Jeddah. The WTA Finals moved to Riyadh. A country that has never produced a male player inside the world's top 500 now holds the most important nodes of the tour's bloodstream. For a data analyst, this is the most compelling problem I have encountered in years. Not long ago, I spent three weeks building a tracker for money flowing into ATP 250 and 500 events, wanting to understand one thing: when Gulf money pours in, how does tournament structure change? I found something that forced me to rewrite my entire approach. The money is not distributed evenly. It concentrates in events that can serve as brand gateways, not in events with long traditions. Capital does not buy tradition; it buys geography and broadcast windows. That was when I realized the oil dataset and the tennis dataset share one operating logic. Both revolve around chokepoints. In oil, it is the Strait of Hormuz, where one-fifth of world supply passes and whoever controls it holds leverage. In tennis, it is tournament weeks and weekend prime-time slots, where a large share of the whole season's commercial value concentrates. When a pumping station on the East-West pipeline is damaged, global oil prices react within hours. When a Grand Slam changes its playing window, the broadcast value of an entire system shifts. Both are systems dependent on chokepoints, and both are only worth as much as the reliability of the information flowing through them. Everyone knows what a wild card at a major is worth to a young player. It is the smallest chokepoint in the system, and therefore the easiest place to manipulate. It is like a pumping station. Damage one station, and the whole pipeline weakens. This is where that oil dataset genuinely taught me something, not about the strait, but about how it measures risk. The energy analysts at DBS Bank lay out two scenarios. A base case with Brent between $85 and $95. A bear case with prices spiking to $120 before cooling back to $100. The spread between the two is so wide that it confesses the future is highly uncertain. I used to think my tennis models lacked that humility. We tend to give a single number, say a player's title probability at 23.4 percent, and present it like a prophecy. The energy industry, with all its numerical rigour, is forced to admit it does not know. It can only measure the dispersion of what it does not know. If I applied the same logic to a Grand Slam quarter-final between two top-ranked players, I would not say who wins. I would say: the base case is a match with fewer than three breaks of serve. The bear case is a player double-faulting in the second tie-break, and every calculation collapsing within seven minutes. The risk is not in who is stronger. The risk is that we slap a certain label on a chain of variables we do not control. Looking at how Jannik Sinner and Carlos Alcaraz have split the majors over the past two years, a hasty analyst would conclude a new era has begun. But the data curve on Novak Djokovic's minutes played at nearly forty is far flatter than the stereotype of age suggests. The same dataset, three ways of labelling, three opposite conclusions. I once tried something smaller. I took the return points won figure of a young player in the top 30, saw it higher than the leading group, and hastily placed him among title contenders. Three months later, that figure sank to average. The problem was not the metric. The problem was that I had labelled a trend onto a sample of only seven matches, three of them against clearly weaker opponents. In tennis data, a small sample is the best liar in the room. That is why I began to notice a side effect few people mention. When every serve, every net approach, every break is recorded as data, that data also flows straight into betting companies. Live data supplied to bookmakers is the darkest side effect of the digitisation of sport, and it is a subject I will never stop digging into. A dry table of average first-serve speed becomes an input for an odds line. A mislabelled qualification round becomes a false signal for a model. And when the money wagered is large enough, the very operators of the system have an incentive to keep the wrong label in place, because ambiguity can be profitable. I learned my trade at Sports Illustrated as a fact-checker. Back then I thought my job was to verify the number. Now I understand the real job is to verify the context of that number. This is where I have to hold myself back, because my temperament dislikes ambiguity. When I see a domain label applied wrongly, the first reaction of an order-loving person is to hunt for the error, find who filled it in wrongly, fix the field, and get on with the work. Clean. Tidy. Exactly the way an executive operates. But if I treat that wrong label as a one-off exception to be fixed, I will miss the bigger point: we do not only mislabel other people's data; we mislabel our own. Every time a player wins three straight matches on hard courts and then falls in the fourth round, and I hastily conclude the form is rising, I am doing exactly what that wrong label did. I am assigning an unverified meaning to a short run of numbers. Data does not lie; it is the reader of data who makes excuses. In 2026 I learned that a 95 percent probability still has 5 percent that knows how to laugh. My model ranked the team I was certain would win the title in fourth place, and when the fourth-ranked team actually lifted the trophy, I understood that the number itself does not create truth, the reader places expectation upon it. A dataset entirely accurate within its own field can still do terrible damage when placed in the wrong one. Transfers are where people pay hundreds of millions to buy a single row in a spreadsheet. I still remember a newsroom argument about allowing five substitutions in football. A colleague argued it helps squad depth. True. But the data I collected from matches after the rule took effect pointed to another conclusion: the final twenty minutes became a war of attrition, as teams took turns sending on fresh legs to hold their pressing rhythm. The rule changed the surface, and the old data no longer labelled the new match correctly. Tennis is the same, every time the serve clock rule or off-court coaching right changes, the entire old metric system has to be read again from scratch. So if I dismiss the wrong label as a mere technical glitch, I am missing the biggest opportunity it offers. That wrong label taught me that every analytical system has a structural weakness: it is only as good as its ability to classify inputs. A perfect prediction model built on a mislabelled catalogue will produce worse results than the intuition of an ordinary fan. There is another temptation I must avoid. On discovering a rare error like this, it is easy to turn it into a grand revelation, a systemic con worth a front page. But that is not the case. There is no conspiracy. There is no saboteur. Only a data field left blank, defaulting to the wrong value. The most frightening errors in data are usually not loud. They are as quiet as an empty cell in a spreadsheet. What I will track in the coming rounds is far simpler than building another model. I will track the shifting flow of capital into tournaments, and alongside it the transparency of the labels attached to the data. A correct label does not guarantee a correct conclusion, but a wrong label almost always leads to a wrong one. I still keep that oil dataset on my machine, undeleted. It sits in my System Errors folder, next to other spreadsheets that once nearly led me to a hasty conclusion. Sometimes the best way to understand a big number is to look at a small number that was placed in the wrong spot. And perhaps, in a season where Gulf money pours into every prime-time slot, fans should ask themselves: when one country owns both the oil pipeline and the tennis rankings, who is writing the label for the match we are watching?