F1 2026: The Empty Dataset and the Test of Nerve for the Racing Industry

02:14 London time, 12 November. On the third monitor of a small flat in the east of the city, a spreadsheet opened with 1,247 rows. The driver-name column was empty. The races-completed column was empty. The performance-index column was empty. A dataset perfect in structure and entirely hollow in content.

Across forty-four years of watching this sport — from editing at Motoring News in 2026, through 406 consecutive Grands Prix covered trackside, to running transfer-market work in London today — I had never received an empty dataset. Every race weekend a single team generates terabytes of information: on-car GPS, infrared tyre-surface temperatures, CFD flow fields, brake pressure sampled at every centimetre of circuit. What I had here was a skeleton with no bones.

That moment kept me at my desk longer than any championship argument. Data is never in a hurry, but people always are. And an empty dataset, in its own way, says more than ten beautiful charts.

Some context is needed to explain why a blank frame deserves attention.

The 2026 season opens the largest regulatory cycle since 2026. The new hybrid power unit splits output roughly 50/50 between the combustion engine and the electrical system, runs on 100 percent sustainable fuel, and replaces DRS with an active aerodynamic system operating in X-mode and Z-mode. Cars are smaller and around thirty kilograms lighter than the previous generation.

At the same time, the team structure has shifted at the root. Audi puts its own brand on the grid as a power unit manufacturer. Ford partners with Red Bull Powertrains. Honda moves to Aston Martin. Alpine steps away from its role as Renault's works team to become a Mercedes customer. And for the first time since 2026, the grid welcomes an eleventh team: Cadillac.

Eleven teams, five power unit manufacturers, a cost cap adjusted year by year, and an aerodynamic testing restriction system that sets allowances in reverse order of the previous season's standings. That is an equation with too many variables, and every variable needs data to solve.

In the economics of modern motorsport, data has become a second currency. The cost cap limits the money, so the real constraint sits in decision quality. A team can spend its full allowance and still be three tenths slower per lap if it uses poor numbers. In the other direction, a lower-spending team can compensate by valuing things more accurately.

The transfer market is a contest in which whoever prices correctly wins. That sounds paradoxical in a sport where speed decides everything, yet it holds in both directions: pricing drivers, and pricing technical upgrade packages.

THE COST CAP TURNS DATA INTO AN ASSET

Before 2026, a leading team could spend three times what a backmarker spent. The cost cap erased that gap on paper. It did not erase the gap in the ability to read numbers, which is built over decades.

When budgets hit a ceiling, the advantage shifts from whoever has the most money to whoever makes fewer wrong assumptions. This is the point most commentary misses. Pundits look at the spending table and conclude the game is now level. But equality of money is not equality of decision quality.

I spent three months in 2026 tracking Brentford, a Championship club then famous for signing cheap players on data. I analysed 1,247 players across fifteen European leagues and filtered out 38 potential targets using xG, PPDA and chance-creation counts. When Brentford signed Ollie Watkins from Exeter for 1.8 million pounds and later sold him to Aston Villa for 28 million, I understood something directly transferable to F1: data is not an auxiliary tool, it is a strategic weapon.

Brentford does not read the future, it simply reads data more carefully than everyone else. In F1 2026 that principle translates into a concrete question: on an identical budget cap, which team allocates resource to the area that actually produces lap time?

The answer is not buying more equipment. It is eliminating projects that produce no delta. A floor upgrade consuming six wind tunnel weeks might deliver two hundredths of a second. A small radiator change might deliver nothing and still eat budget. The winning team under a cost cap is the one that says no to more projects than its rivals do.

ATR: THE COUNTER-INTUITIVE TRADE-OFF

The ATR mechanism — aerodynamic testing restrictions — allocates wind tunnel runs and CFD allowance in reverse order of the previous season's constructors' standings. The champion receives the least. The last-placed team receives the most.

It is a cleverly designed levelling tool, but it creates a paradox I have rarely seen analysed properly. More testing time does not automatically produce a bigger step, if the team cannot ask the right question.

I once sat with a former aerodynamicist near Brackley. He said something I wrote down verbatim: we do not lack runs, we lack good hypotheses. A midfield team can have half again as many runs as the leader, but if it spends them confirming what it already believes, the gap will not close.

Conversely, the leading team with fewer runs must be more selective. Scarcity of resource forces prioritisation. In many cases that discipline is itself an advantage.

This is why I always cross-check at least three independent data sources before drawing a conclusion about an upgrade package. Wind tunnel data, CFD data and on-track data frequently tell three different stories. When they align, I believe them. When they diverge, I know something has not been modelled.

In the 2026 cycle that divergence will be larger than usual. A car with active aerodynamics, dependent on battery deployment and an even split of combustion output, will behave very differently between X-mode and Z-mode. Simulation-to-track correlation will be more fragile than at any point since 2026.

THE PRICE OF EXPERIENCE: A LESSON FROM THE NEW GRID SLOT

When the eleventh team announced its driver pairing, I reacted differently from the consensus. Both drivers are at an age the media routinely describes as the wrong side of the hill. Combined, their race starts exceed six hundred. Both are champions.

The default crowd reaction was scepticism. I read it as a valuation decision.

A brand-new team has three problems to solve simultaneously. First, it needs baseline data to calibrate its models — something a rookie cannot provide, because a rookie does not know what a correct car should feel like. Second, it needs rapid car development, which demands accurate technical feedback. Third, it needs reliability, because a new team cannot afford to lose itself in a run of crashes.

All three problems lean toward experience. But the real investment sits in a fourth problem few mention: the learning curve of the internal data model.

A veteran driver does not only bring a steering wheel; they bring a library of feel validated across multiple car generations. When engineers ask where downforce is missing, their answer converts straight into a measurement channel. A rookie will say the car is difficult. A former champion will say the car loses balance at entry to medium-speed corners, and the engineering group knows exactly which data channel to open.

That value is quantifiable. It is unglamorous, generates no headlines, and shortens the development cycle.

I have mispriced this before. In 2026 I undervalued a club because it signed older players instead of buying young talent. My model then measured resale value only. It did not measure calibration value. I rebuilt the framework afterwards, adding two indices for technical communication and feedback consistency.

That is why I regard an experienced pairing at a new team as sound in data terms, even when it is unsound in media terms.

THE PREVIOUS CYCLE IS ALWAYS A TRAP

A pattern repeats at every major rule change, and I have watched it unfold at least four times in my career.

Before each new cycle, teams pour resources into analysing the one just finished. They hunt for patterns in old data. They build models on what once worked. And nearly half of it turns out wrong.

Every cycle imitates the data of the cycle before it, and nobody learns. The reason is simple: when the rules change, variables interact in ways that never appeared in the historical record. The model is not wrong because the data was poor. It is wrong because old data no longer describes the new system.

In 2026, when the V6 hybrid arrived, many teams underestimated the value of electrical deployment on corner exit. They held data on naturally aspirated V8s, and that data was useless under the new rules.

In 2026, when ground effect returned, some teams reused older aerodynamic models and spent half a season stuck with porpoising.

In 2026 the new variable is the interdependence of energy management and active aerodynamics. A driver must decide when to deploy electrical power and when to switch wing state, and the two decisions affect each other in ways no existing dataset records.

The team that understands its historical data is depreciating lap by lap will adapt fastest.

I am not saying old data is worthless. It remains useful for calibrating method, auditing process and detecting systemic error. Using it to predict performance inside a new regulatory cycle is a methodological mistake, and that mistake costs more than having no data at all.

FROM BRENTFORD TO SILVERSTONE: A VALUATION FRAMEWORK

I built a twelve-index framework of my own, developed at Brentford and then ported to F1. It does not measure the fastest driver. It measures the mispriced driver.

The first four indices measure performance capability: relative lap time against a teammate, consistency across races, long-stint tyre management, and output under pressure.

The next four measure technical capability: quality of feedback to engineers, adaptability to changing car configuration, contribution to development direction, and how often the driver identifies a problem before the sensors do.

The final four measure organisational capability: influence inside the engineering room, cultural fit, willingness to take responsibility in failure, and commercial value delivered.

The twelfth index is the one I use most when assessing transfers, and the one most often ignored in coverage. It does not measure how good a driver is. It measures the gap between true value and market price.

When that gap is positive, there is opportunity. When it is negative, there is risk.

For the 2026 cycle I see three forms of mispricing forming. New teams undervalue continuity. Big teams undervalue switching costs when regulations change. And nearly every team undervalues the time required for a new model to reach reliability.

These mispricings do not appear in the standings. They appear in resource-allocation decisions, and they surface only twelve to eighteen months later.

I logged them in a tracking book in September and will reconcile them when the 2026 season closes. That is the only way an analyst can audit himself: state the prediction first, then let the results judge.

THE BLIND SPOT OF THE DATA READER

Here I must argue against myself.

My entire writing career rests on the belief that numbers are more trustworthy than sentiment. But that empty dataset taught me something else: faith in data can become a new superstition, if we forget that every figure is the product of a process.

An aerodynamic model can predict to a thousandth of a second and still be entirely wrong, if it was fed a faulty assumption about track temperature. A driver performance index can look immaculate and mean nothing, if it was computed on a superior car.

Correlation is not causation, and in F1 confusing the two costs more than any technical failure.

There are three blind spots I have seen repeat across almost every team and every analytics department I have worked with.

The first is the proxy metric. When you cannot measure what you want, you measure something easier and assume equivalence. Overtake counts become a proxy for attacking ability, despite depending mainly on grid position and strategy.

The second is sample size. Three strong races say nothing definitive about a driver or an upgrade. But in a sport with twenty-four rounds a year, three races is nearly an eighth of a season — enough for media to build a narrative, and not enough to conclude anything.

The third is the equipment filter. When evaluating a driver, people forget that their results depend more on the car than on themselves. Strip out that filter and the ranking changes substantially.

There was a period in 2026, when circuits stood empty, that forced me to review my whole method. Empty stadiums in 2026 exposed a truth: much of what we called composure was just noise. Without crowds and grandstand pressure, some drivers held form and others collapsed. Earlier data could not distinguish the two groups.

That is the lesson I carry into 2026.

At sixty, I no longer believe in luck, only in numbers that have not yet spoken. But I have also learned that before trusting a number, you must ask which process produced it.

SIGNALS FOR THE NEXT LAP

The first three months of 2026 will not tell us who wins the title. They will tell us which team built the right model.

Three signals I will track, all of them measurable.

One: the divergence between wind tunnel data and track data across the opening three rounds. The team that closes that divergence fastest has the best process.

Two: strategic decision speed in the first two races with complex weather. The cost cap makes every wrong call more expensive.

Three: how many drivers who changed teams reach stable form within eight rounds. This is a direct test of my valuation framework.

Forty-four years in this business taught me something simple. People argue about conclusions and rarely about method. But every conclusion begins with a dataset, and the best dataset is only as good as the question that produced it.

That night I did not delete the empty spreadsheet. I saved it and named the file "the test". In three months, when the season begins, I will open it again and ask myself whether the numbers I am about to collect truly answer the question I am posing — or merely fill a void with noise, beautifully formatted.