An Empty Spreadsheet at 3 A.M.: The Silent Discipline of F1 Analysis

Three in the morning in London, late February. The third-floor flat still smells of the coffee that went cold at dinner. I open the old laptop, click the file named "2026_pre_season_v7", and look at nine blank tabs. Not a number. Not a name. Not a timestamp.

The night before, a document arrived for analysis. It came as a pre-processed summary: empty title, empty source, unclassified article type, a domain label reading "f1" with no content behind it. The most important field — the list of information points — was an empty cell. Every other field in that document depended on the empty cell, including the entity list, which explicitly instructed the reader to "identify from the information points above," where there was nothing above.

I sat still for a while. Outside, the first night bus of the number 24 route rolled past and threw orange light across the ceiling. In this trade people say a race begins when the red lights go out. For an analyst it begins when you sit in front of a gap and decide what to do with it. There are two options. Fill the gap with confident prose, or publish the gap itself.

I chose the second, and this article is about why the second is many times harder than the first.

Context: two stages, and a door locked from the inside

My method has two stages. Stage one is decomposition: read a source, extract discrete information points, number them, label them. Stage two is analysis: run each point against a nine-dimension framework, where every conclusion must trace back to a specific numbered point.

That architecture has a fatal weakness I have known about for years and rarely said out loud. If stage one fails, stage two cannot rescue anything. It can only produce a skeleton of the correct shape — full headings, full tables, full "conclusions" sections — and absolutely nothing inside.

A complete analytical framework is not evidence of analysis. It is only evidence of a framework.

The nine dimensions are nine doors every race weekend opens to some degree: car and technical; race strategy; team and driver; competitive landscape; regulation and governance; driver market and talent ecosystem; risk profile; public narrative and expectation; and industry transmission.

Nine doors. That night, all nine were locked from the inside, and the key sat in one empty cell.

Where F1 data actually comes from

Timing loops record every car to the thousandth of a second, split into three sectors with intermediate points inside each. A 58-lap race generates hundreds of thousands of individual time values.

Then onboard electronics. Since the standard ECU FIA mandated from 2026, every team runs the same hardware platform, and each car transmits hundreds of channels: brake temperatures, tyre temperatures and pressures, fuel consumption, energy deployment mode, steering angle, brake force, speed at every point on the circuit.

Then the physical layer: track temperature, air temperature, wind speed and direction, humidity, minute-by-minute rain probability. The tyre layer: allocated compounds, tyre age by lap, estimated degradation. The visual layer: onboard cameras, trackside cameras, pit lane photography. The textual layer: race control notices, stewards' decisions, technical directives, weighing reports.

And a final layer, the least discussed and often the most decisive: paddock sourcing. No data channel records that a technical director left a meeting looking different than usual, or that a team suddenly hired three aerodynamicists.

That night I had nothing for any door.

Door 1: technical, and the gap between paper and tarmac

In the ground-effect era that began in 2026, the floor became the main battlefield. Air is accelerated under the floor to generate downforce, and in doing so it generates the vertical bouncing the paddock calls porpoising. The 2026 season was a race to raise floors, lower floor edges, add slots, cut slots, and try again.

An upgrade that has never run on track is only a drawing. It has not been tested by three things no wind tunnel reproduces: track temperature that changes continuously, dirty air from the car ahead, and the abrasion of a track surface after dozens of laps.

Two numbers determine an upgrade's real value. First, the lap time gained, measured on the same compound, the same fuel load, the same track conditions. Second, the stability of that gain across consecutive laps. An upgrade worth 0.05 seconds on lap three but that kills the tyre by lap twelve is a bad upgrade.

Alongside that sits the aerodynamic testing restriction system. Since 2026, teams higher in the standings receive less wind tunnel time and less computational allowance. The leading team gets roughly seventy percent of the baseline allocation; the last-placed team gets around one hundred and fifteen percent. That mechanism erodes on-track advantage over time, and it is part of why modern eras of dominance are far shorter than those of the 2000s.

But to write about it I need to know which team, which circuit, which upgrade, and how much it gained. That night I had none of it.

Door 2: strategy, and the biggest silence on the circuit

Transition is not the running. It is the silence between two intentions that few people know how to read.

In F1 that silence has three forms. The first sits between two pit stops, when the old tyre is past its cliff and the new one is not yet warm. The second sits between the braking point and the apex, where a driver trades entry speed for exit speed. The third sits between the engineer's answer on the radio and the hands on the wheel.

None of the three appears in a timing sheet. They appear only when you place the timing sheet beside the pictures and beside the radio transcript.

In the first form, the deciding number is pit loss, which varies enormously between circuits depending on pit lane length, pit speed limit, and garage position relative to the entry. Where pit loss is low, the undercut becomes a cheap weapon. Where pit loss is high, track position wins and teams accept longer stints.

In the second form, the gap is not created at the latest braking point but at the earliest exit. A driver braking two metres later but exiting three kilometres per hour slower loses the entire straight that follows.

In the third form, decisions are made in about seven seconds and never recorded in any official document.

The summer of 2026 taught me that a gap is never empty; it is only waiting for the right reader. Six months without crowds, I rewatched dozens of matches and realised what I lacked was not data but a way of reading it. I moved to F1 with the same conviction: if the timing sheet cannot tell the story, the storyteller has not read carefully enough.

That night I had no timing sheet to read.

Door 3: two drivers in one garage

This is the easiest dimension to be fooled by and the easiest to get wrong.

Comparing teammates requires at least four data layers: qualifying, race pace on the same compound, stint-to-stint consistency, and behaviour in abnormal situations such as safety cars or rain.

Qualifying is the clearest layer and the most misleading. A driver blocked on a fast lap can lose three tenths that never appear in the results. A driver running two preparation laps instead of one may arrive with the tyre at a much better temperature.

Race pace on the same compound is the most honest layer, but only if you isolate fuel load. A car carrying ten kilograms less will be roughly three to four tenths per lap quicker, and across a fifteen-lap stint that accumulates into an apparently impressive gap that is entirely artificial.

Consistency is the hardest layer. I once spent three weeks rewatching footage to count twenty-seven attacks exploiting the same gap on a football pitch. Moving to F1, I kept the habit: counting every tyre lock, every time a driver ran wide at the same corner.

A misplaced pass is not a mistake. It is data the system is trying to send you.

In F1 the equivalent is a slow pit stop. It is rarely one person's error. It is usually the signal of a process that drifted out of rhythm several laps earlier.

Door 4: the competitive landscape and the standings trap

The standings are among the most misleading datasets in sport. They accumulate results from circuits with entirely different characteristics. A team strong at high-downforce tracks may be weak where top speed matters, and points do not distinguish between the two.

So I build the landscape from track families first, then place each team's results onto those families, and look for the most stable pattern.

Meanwhile, the cost cap has changed how teams allocate resources. The cap arrived in 2026 at one hundred and forty-five million dollars for the first season, falling to one hundred and forty million in 2026 and one hundred and thirty-five million in 2026, before being indexed upward again for later seasons. In that environment, a midfield team can no longer burn money to close the gap. It can only choose the right place to burn it.

From 2026 the landscape resets again: a new power unit formula with a more even split between combustion and electrical energy, sustainable fuels, and active aerodynamics replacing the old drag reduction system. The door opens; but to write about it I need to know which circuit, which team, and what result.

I have none.

Door 5: regulation, stewards, and the scars of the sport

This is the dimension I consider most important and the one most neglected in daily coverage.

On 28 October 2026, the FIA published its findings on a team exceeding the 2026 cost cap. The penalty combined a seven million dollar fine with a ten percent reduction in aerodynamic testing over twelve months. It was a double penalty: one blow to the wallet, one to the rate of development. In a rule set where development is the only thing that determines long-term order, the second part is far heavier than the first.

But a penalty table says nothing about actual impact. Measuring it requires comparing that team's development curve across the twelve months before and after, on the same track families. That is the kind of analysis the media almost never does, because it needs long-horizon data and generates no attractive headline.

Another scar is how late-race situations are handled. After 2026, FIA's officiating structure was adjusted to split roles between race control and the stewards' panel, reducing pressure on a single individual. That is the kind of governance change you only see by reading official documents.

A third scar is track limits. At some circuits the number of laps deleted for exceeding limits can run into the dozens, and adjudication happens after the race ends. In that window, the classification on the screen is not yet the classification.

With no document in hand, I cannot write a line for this door.

Door 6: the driver market and the price of noise

This is the noisiest of the nine dimensions.

The driver market runs on contract cycles, and those cycles are dominated by a group I always view with suspicion: driver managers. They are the system's largest hidden cost. A leak timed correctly can push a contract's value up by millions, and the price is the distortion of the entire market.

I read the market in three layers. Confirmed contract data announced by the teams themselves. Verifiable statements, such as a team formally announcing a driver's departure at season's end. And paddock rumour, which has value only when it comes with a clear motive for the leaker.

In the 2026 season the market is more complex still, with an eleventh team carrying a North American brand, and a German manufacturer taking over a Swiss-based operation. Each ownership change triggers a hiring wave among engineers, and each hiring wave creates mandatory notice periods before an engineer may work for a new team.

That is the geometry of the gap at the personnel level: the period a person is forced to stand outside, while their knowledge retains full value.

This door needs names, dates, clauses. I have nothing.

Door 7: the risk profile, and the only risk I can see

My risk framework has six branches: sporting, technical, personnel, regulatory and financial, reputational, and systemic.

Each branch needs a specific subject. Technical risk needs a suspected component. Regulatory risk needs an alleged infraction. Personnel risk needs an unfilled seat.

That night all six branches were empty. And here is the point: when a risk table is empty, the standard writer's reflex is to type "low". That is a serious error. Writing "low risk" into a cell that was never assessed fabricates a conclusion. It is like declaring a car has passed scrutineering when it never entered the weighing bay.

The only identifiable risk here sits on the reader's side: the risk that an empty document is presented as substantive analysis and consumed as such.

Door 8: public narrative and the expectation gap

Public narrative is a data type with its own cycle. It starts with a surprise result, heats up through bulletins, peaks at a controversial statement, then fades as the next race approaches. Its problem is that it does not correlate with actual performance.

A driver can be praised for two weeks over a wet-weather drive while the data shows the team's strategy was the deciding factor.

I measure the expectation gap in three steps: record the circulating expectation, build an independent assessment from timing and tyre data, then measure the distance between them. The wider the gap, the more likely one side is wrong. In most cases, the wrong side is the more numerous one.

I once received feedback that my analysis explained the result but not why the opponent generated so many dangerous counterattacks. The criticism was correct. I lacked transition data entirely. So I built a separate table to log every transition phase, and at the end of every piece I added a section called "Data limitations".

That section is now the one I am proudest of.

Door 9: industry transmission, the hardest layer

Upstream sits manufacturers and academy pipelines; midstream, teams and the commercial rights holder; downstream, broadcasting, sponsorship and derivative markets. These three tiers run on three different clocks: years upstream, seasons midstream, months downstream. Any analysis that mixes the three without saying so will reach a wrong conclusion.

With an empty data cell, building the chain is impossible. You need at least one name in each tier to begin.

The contrarian angle: this industry rewards volume, not verification

The sports media economy does not reward accuracy. It rewards speed and volume. An analysis published thirty minutes after the chequered flag, half wrong, will be read ten times more than one published three days later that is correct to the detail.

That asymmetry creates an inverted incentive: the less data there is, the more confidently people write, because confidence is the only thing that can fill a gap quickly.

I have seen this repeat across many seasons. After every race the news market floods with unverifiable claims: an upgrade delivered a big gain, a strategy call was a mistake, a driver is losing form. None come with data, and none are ever revisited.

Meanwhile, the serious analyst faces the opposite reality: the more carefully you read, the less you dare to assert. That is why real analysts say "I don't know" more often than anyone else.

There is another trap. When you have too much data, you tend to publish all of it. I once wrote twelve-page pieces with dozens of tables, and readers took nothing away. So I set a question before publishing: which number changed my conclusion? Numbers that fail that test get cut.

A third trap is turning a concept into a template. I am known for writing about transitions, and that is a risk. When a keyword builds your brand, you start seeing it everywhere, including where it does not exist. So I set a rule: use that concept only when at least three data points back it.

A fourth trap is dismissing the human factor because it cannot be measured. A spreadsheet cannot measure an engineer who has had a hard week at home. But a shaky line on a drawing comes from a person whose hand is shaking, and that affects what happens on track.

A fifth trap is mistaking correlation for causation. After finishing a conclusion, I force myself to imagine a data scenario that would falsify it. If I cannot imagine one, the conclusion is not tight enough.

And a sixth trap, the one I was staring at three in the morning: believing that a complete framework means a complete analysis.

When there was no football, I drew football. It turned out drawing was also a way of understanding. When there was no F1 data, I nearly drew F1. This time I stopped, because drawing something that does not exist is not understanding. It is colouring in.

What the next race will test

Every tactical diagram begins with a shaky hand-drawn line on PowerPoint.

I still keep that habit. Every race I open a blank page and draw by hand — crooked lines, misaligned arrows, hurried annotations. Those lines are proof that strategy is a process of self-interrogation, not a pre-packaged product. A perfect diagram usually signals that the author already knew the conclusion.

Based on my experience following races across many seasons, the next race weekend will test one very specific thing: whether a complete analytical framework can exist without data. My answer is no. And saying no, in an industry that rewards always saying yes, is part of the job.

That night I saved the file and named it "null_result". Then I did the necessary thing: I logged the date, the time, the failed source, and the conditions that produced the failure. Three possibilities went on the table: the source was behind a paywall, the page rendered only through JavaScript, or the response was truncated during fetching.

All three are testable. That is the difference between a gap and a mystery. A gap is something you can fill by returning to the right place. A mystery is something you can only fill by inventing.

Next race, when I reopen those nine tabs, I will start with the first and only question that matters: did the data arrive?