When F1 Data Lies: The Lesson of Numbers Nobody Ever Verified

On the evening of March 5, 2026, in the data office on the second floor of the western stand at San Siro, I reopened the footage of an AC Milan home match in Serie A. On the screen, my team generated 1.85 expected goals per home game, nearly double the 1.02 figure on the road. But when I counted actual goals, the two numbers came out equal. A team supposedly devastating at San Siro scored exactly as often as it did away from home. The contradiction was not in the striker, nor in the coach. It lived in a sensor placed in the south-west corner, lagging two-tenths of a second every time the goalkeeper played the ball out. I wrote a fourteen-page internal report and recommended recalibrating the whole system. Coach Vincenzo Montella used that finding to shift the ball circulation toward the right flank, Milan won five of the last eight matches and secured a Europa League place. I retell that old story not to praise myself. I retell it because those fourteen pages shaped the way I have looked at every number on an F1 track for the nine years since. A sensor off by two-tenths of a second was enough to erase an entire season. And in a sport where each car streams hundreds of channels of data every lap, that two-tenths error is not the exception. It is the rule. When I began covering F1 in 2026, an entire race weekend's data fit into a single notebook. Lap times were hand-timed with a mechanical stopwatch, top speed was an estimate printed on a notice board, and the only measurement taken seriously was the finishing order. Thirty-eight years later, a modern F1 car sends roughly a million data points to the control tower during a single fast lap, equivalent to around 1.5 gigabytes per team over a weekend. Thousands of sensors measure wheel speed, throttle position, brake pressure, steering angle, the temperature of every tyre layer, the current flowing through the energy recovery system, three-axis acceleration, and GPS positioning accurate to the centimetre. That technology turned F1 from an endurance contest into a discipline where victories are often decided before the car reaches the pit lane. But it also built a dangerous faith: that whatever is measured must be true. Over nine years of working weekends at circuits, I have watched that faith break more times than I can count. Not because the engineering crews are incompetent, but because they place on an altar numbers that belong on an operating table. At one European round I followed, a midfield team lost its place in the third segment of qualifying simply because the GPS signal on the car lagged half a second behind the start line. Nobody in the team understood what was happening until a data analyst realised the acceleration peak did not match the moment the driver released the brake in the radio transcript. The number was still there, as round and confident as any other number on the screen. It just happened to be lying. The biggest lesson about data in modern sport does not come from software departments or prediction models, but from measurement conditions. A movement number only means something when we know exactly the circumstances in which it was born. The San Siro sensor lagged two-tenths of a second because it sat too close to the power line of the floodlights, picking up electromagnetic noise every time the lighting system switched on. No software could have detected that on its own, if the humans had not been willing to spend time cross-checking data against real footage. On an F1 track, the same trap plays out far more subtly. A team can spend an entire night reading aero data from the wind tunnel, convinced it has optimised the pressure field under the floor, and then go out the next day and lose a full second per lap. The cause is not the design itself, but the difference in track temperature between the night session and the afternoon race. The data is not technically wrong. It is only wrong in context. I remember a chief engineer at a top team telling me over the phone, his voice exhausted after three sleepless nights: 'We do not lack data. We lack the time to understand what our data means.' That sentence has haunted me ever since. Data speaks only a part; the rest lies in whether people know how to listen. In 2026, as a specialist commentator for Sky Sport Italia at the World Cup in Russia, I once posted an analysis on social media in the seventieth minute of Germany against South Korea. I wrote that the German back line was pushing an average of sixty-eight metres high, that pressing had failed seventeen times, and that South Korea had already produced twelve counter-attacks. I concluded that if the block did not drop deeper, the goal would come from an aerial situation. In the ninety-third minute, Kim Young-gwon scored exactly as the script I had described. Thousands of accounts piled in to mock me for turning the emotion of a match into arithmetic. But what I remember is not the mockery. What I remember is a major Italian newspaper reprinting my analysis with a trapezoid diagram showing how the German defensive block had been twisted out of shape. I realised that a number only carries weight when it is translated into spatial imagery. From then on I abandoned the phrase 'the back line pushes sixty-eight metres high' and switched to 'the drawstring has snapped open all the way to the valve box'. The same truth, but the reader remembers the track, the gap, the way a defensive line tears itself apart. Back to F1. In recent years, the accuracy of position-tracking data has reached levels so astonishing that teams can reconstruct every steering input of a driver on the final lap and compare it with his best lap two hours earlier. But the more precise it becomes, the more easily data is trusted blindly. Every tracking number needs to be placed on an operating table, not an altar. A clear example lies in how teams read tyre degradation. The surface-temperature graphic on the screen looks convincing, with bands shifting from blue to red through every corner. But surface temperature is only part of the story. It does not reveal core temperature, it does not reveal graining, it does not reveal which driver is being forced to cool the tyres by lifting earlier in a given sector. There are weekends where I watched a driver dive into the pit lane three laps earlier than planned because the surface data reported overheating, only for the crew to discover the tyre still had performance left and that they had just thrown away a top-ten finish. The most subtle trap is not in wrong numbers, but in empty data treated as real. When a sensor fails and stops transmitting, the software often fills a default value to stop the screen going blank. An engineer reading the screen at three in the morning can see that value without knowing it is a blank cell dressed up. In elite sport, the most expensive decisions are usually made on those blank cells. I once watched a team decide to keep a driver out on track one more lap based on a tyre-temperature forecast from a sensor that had been dead since the previous lap. The driver lost position in the pit stop, the team lost points, and the entire control room stayed silent for twenty minutes after the race ended. Nobody wanted to admit they had believed in a ghost. This is why I always question the source of data before discussing conclusions. In any analysis I write, I devote a line to measurement conditions: which session the data came from, how hot the track was, which tyre was fitted, whether the driver was in fuel-saving mode. Those details are not glamorous, but they are the difference between a correct analysis and an expensive mistake. Every collapse has a precondition; few people bother to look ahead. The DNF on the results sheet does not appear at the moment the car stops by the side of the track. It begins three rounds earlier, in a skipped data check, in a technical meeting where everyone was too tired to ask the hard question. The San Siro sensor lagging two-tenths of a second had been crying out for half a season before anyone opened the footage. Dependence on data also produces another illusion: the feeling that every variable can be controlled. Modern F1 teams can simulate thousands of strategy combinations in seconds, calculate the optimal pit window to a tenth, and forecast the probability of a safety car. But no model predicts a driver choosing to stay out half a lap longer because he feels the tyre still has grip. Human instinct, which appears in no dataset, remains the largest and most mysterious variable in this sport. Data speaks only a part; the rest lies in whether people know how to listen. And listening, in this context, is not only hearing engineers report figures. Listening also means hearing a driver's breathing over the radio, hearing the hesitation in a strategist's voice as he proposes a fallback, hearing the unusually long silence after a pit stop. Those signals never appear in tables, yet they carry what tables leave out. The grandstand is also a data channel. Emptiness does not kill a race, but it takes away something no metric can measure. A grandstand full of noise creates a very different psychological pressure from a silent one. Drivers make final-lap mistakes in front of a roaring crowd in a way that differs sharply from how they err in silence. Models cannot measure this, so they usually ignore it, even though it can be the deciding variable between two evenly matched drivers. Transfers are another field where data is easily sanctified. In recent years I have watched teams build spectacular deals on simulation metrics, believing that simply fitting a high-scoring driver into a fast car automatically produces results. A contract looks beautiful on paper only until someone tries to fit it into a running system. A driver can be perfect in the old team's data yet incompatible with how a new car distributes braking force or how a new crew communicates. No spreadsheet measures system compatibility before it is assembled and run. Over more than five hundred grands prix I have followed live, I have learned that the most valuable information usually comes from the smallest details few people notice. The difference between a championship team and a runner-up is sometimes just the ability to spot that their sensor is lagging half a second, or the ability to realise the driver no longer trusts the car even though every number looks perfect. There is another lesson I want to carve deep: the best data is cross-verified data. Never quote a number that has not been checked against at least two independent sources. When the official timing system says one thing and the engineer on the pit wall says another, I always spend the time to find where the gap comes from. In many cases, that gap is the real story. The number on the board is only the casing, and the story lives where two sources fail to agree. I once received a complete analysis dossier with full headings, tables, lists, and even professional assessments that sounded highly convincing. But on close reading, every content field was empty. No event, no team, no driver, no grand prix. Only one label survived at the top of the document: the F1 domain. Everything else was a pre-built frame waiting for someone to fill it, and nobody did. What is frightening is not the empty dossier. What is frightening is that the empty dossier could still be read aloud as professional analysis, because it was presented so beautifully. This is the biggest trap of the data age: tidy form can conceal empty content. A table with no real data looks exactly like a table with real data, if the reader does not bother to check. I drew three principles for myself after that episode. First, always check whether the data actually exists or is only an empty frame. Second, always question the origin and the conditions that produced the number. Third, always retain doubt toward conclusions that sound too smooth. No sports analysis is perfect, and excessive smoothness is usually a sign that something, somewhere, was skipped. Back to the current F1 track. This season is witnessing fierce competition among the leading teams, with drivers such as Max Verstappen, Lando Norris, Charles Leclerc, Oscar Piastri, George Russell and Lewis Hamilton often separated by a few hundredths on the same lap. At that margin, the difference no longer comes from raw speed, but from the ability to read data correctly and decide at the right moment. A team that misreads a tyre signal loses a position. A team that misreads a weather signal loses a whole race. I noticed a small detail at a recent round: over the final three laps, a leading team kept changing engine modes at an unusual rhythm, unlike any of their earlier runs that weekend. Looking at the public data, nothing seemed suspicious. But when I cross-checked against the moment the driver began lifting in the final corner, I realised they were deliberately cooling the tyres to set up an attack on the last lap. No table recorded that intent. It only surfaced when the number was placed beside the behaviour. This is what I want to stress in every analysis: numbers do not tell stories by themselves. The storytellers are the people who place numbers in the right context, compare them with observed behaviour, and doubt them when everything looks too perfect. In a sport where every team has access to the same enormous volume of data, the real competitive advantage lies not in owning more data, but in understanding your own data more deeply, and knowing when to trust and when to doubt. Modern teams now invest millions of dollars in analytics departments with hundreds of engineers, yet I still see basic mistakes repeating: misreading measurement conditions, trusting a dead sensor, ignoring human signals. Technology advances by leaps and bounds, but the nature of the trap has not changed since I reopened that footage at San Siro. A number without context is a number capable of deceiving. I remember a conversation with a veteran engineer at a team. He told me his hardest job was not analysing data, but convincing people that data can be wrong. An entire crew is built on faith in data, and pointing out that part of that faith is mistaken takes more courage than finding a technical fault. Because data, in many people's eyes, has become a religion. That is why I repeat this line to myself before every working weekend: every tracking number belongs on an operating table, not an altar. Placing it on the operating table means dissecting it, testing it, cross-checking it, finding where it came from and what it truly means to say. Placing it on an altar means worshipping it, trusting it absolutely, and ignoring what it cannot say. In a race, the gap between winner and loser is sometimes just a decision made on a misread number. A pit stop one lap early. Holding out on track a little longer. A small shift in braking distribution. Those decisions are all supported by data, but they are only correct when the decision-maker understands exactly where their data stands between truth and illusion. In the current season, with the title fight still open and every point precious, I believe the deciding factor will not be the fastest car but the team that reads data best. The team that can separate real signal from noise will lift the trophy. The team that worships its own spreadsheets will sit and watch others celebrate. I have no intention of advising teams to throw data away. Doing so would be suicide in this sport. What I want to stress is the need to treat data as a source of information that can be wrong, not as an inviolable truth. The need to keep people capable of saying 'wait, this number is not right' in a meeting room where everyone is rushing to decide. The need to build a healthy culture of doubt alongside a culture of faith in technology. In the rounds ahead, as teams enter the sprint phase of the season, I will pay special attention to three things. First, how they react when data and a driver's instinct conflict. Second, how they handle anomalous signals in practice sessions. Third, how they treat imperfect numbers, numbers sitting at the edge of certainty. Those moments will reveal more than any timing sheet about who truly understands their own race. Data speaks only a part; the rest lies in whether people know how to listen. The best listener in a paddock is not the one with the most screens, but the one who knows when to look away from the screen to meet the driver's eyes, hear an engineer's voice, and feel the pulse of a race changing. The F1 of the future will depend ever more on data, and will ever more need people who know how to doubt it. That is the core contradiction any championship team must solve before it solves the problem on track. When the grandstands fill with noise again and every lap is recorded in millions of data points, I will still be here, with fourteen pages from 2026 in memory, reminding myself that behind every beautiful number there may be a sensor lagging two-tenths of a second. And my job, like that of anyone doing this profession properly, is to find it before it takes a season away.