F1 Azerbaijan GP FP1: Russell Fastest, Antonelli's W17 Smokes — Reading the Baku Timing Sheet with a Skeptic's Eye

The W17 stopped near the exit of the southern service road, around the thirty-fourth minute of the first practice session. White smoke crept up from behind the engine cover, slowly, almost politely, before anyone in the Mercedes garage could send a single command over the radio. Andrea Kimi Antonelli switched off, undid his belts, and climbed out. Race control raised the yellow flag. On the timing screens along the pit lane, his name was still sitting fifth.

That is the image I kept longest after rewatching the entire FP1 session at Baku. Not George Russell's 1:45.387 lap. Not the four-thousandths-of-a-second gap between Max Verstappen and Charles Leclerc — a coincidence so neat it makes people want to believe it means something. It was the smoke.

Every tactical diagram begins with a shaky hand-drawn line on PowerPoint. And a plume of white smoke from an engine cover, in a session where every team hides fuel loads, hides engine modes, hides even its run plan, is more reliable data than any number on the timing sheet.

That is why I am writing this piece against the grain of how session reports are usually read. The FP1 timing sheet at Baku is a beautiful document. It is simply not honest.

Context: a circuit that lives once a year

Baku City Circuit has a feature few circuits on the F1 calendar possess at once. It is a street circuit. It is used exactly once a year. And so, when the cars roll out for FP1, the asphalt is close to virgin — dusted with a fine layer of sand and old rubber accumulated over twelve months of nobody driving on it.

Teams call this track evolution. I call it the geometry of unfairness.

Picture a surface whose grip coefficient changes by the minute, by the lap, by the set of tyres the car ahead just dropped. By the twentieth minute, the track is still slippery. By the fortieth, it is a different place. A driver on softs at minute twenty-five is driving a different circuit from the one on softs at minute fifty. Same corner, same layout, same name on the timing sheet — but two entirely different tests.

At Baku, this effect is amplified by track shape. The opening sector is a series of slow right-angle corners where the car must rotate at low speed and rear traction decides everything. The middle is a chain of tight corners joined by concrete walls, where a twenty-centimetre error is a ruined afternoon. The final sector is a straight of nearly two kilometres, where power unit output and aero efficiency in low-drag trim tell almost the entire story of the gaps between cars.

These three sectors demand three different parameter sets. No setup is optimal for all three. So at Baku, every team enters FP1 with a long list of compromises, and every lap time they set is the product of a compromise we never see.

That is the foundation for reading a timing sheet. A timing sheet with no fuel load, no engine mode, no tyre age, no consecutive-lap count, no wind direction, no track temperature — is a timing sheet that says almost nothing about competitive performance.

It only says something in another dimension: it tells us about reliability, upgrade direction, and the silent gaps.

Why FP1 is the least reliable session

In my analytical framework, every session carries a weight. Not a weight I feel, but one the structure of the session itself dictates.

FP1 carries the lowest weight, for structural rather than emotional reasons.

First, FP1 runs on a track not yet rubbered in. Every data point is polluted by a surface changing faster than engineers can recalibrate their models.

Second, FP1 is a systems-check session, not a performance-optimisation one. Fuel pumps, sensors, brake cooling, start procedures, telemetry links — all sit on the list before anyone thinks about pushing a car to its limit.

Third, and most importantly, FP1 is a session where each team's run plan differs in kind. Some run a hard tyre through the whole opening phase to gather baseline data. Some split the session into two distinct phases. Some treat FP1 as an aero test with sensors plastered over the bodywork, and their times are dragged down by the very sensors.

Comparing two cars on different run plans is comparing two things that do not share a unit. That is the most basic methodological error in reading FP1 news, and it is the error most session summaries commit.

I have made that error. In 2026 I wrote an analysis of a first practice session and drew the wrong conclusion about the pecking order. The piece performed well. The conclusion was wrong. I still keep that file on my machine, named bai_hoc_01.

Russia 2026 did not only warn about transition. It warned about how we read a match. And with circuits, that warning bites harder, because a circuit does not allow you to rewind the tape and correct yourself.

The 2026 regulation cycle and the cost-cap trap

To understand why the upgrade packages appearing at Baku matter more than the times, they need to be placed in the context of the regulation cycle.

2026 is a cycle in which the technical regulations change at a foundational level: the power split between the internal combustion engine and the electrical system changes, car mass changes, and the entire aero philosophy is rewritten. Cycles like this create a governance paradox: every team knows that resource spent on the current season is resource not spent on the next one.

Here, the cost cap and the aerodynamic testing restriction create a double constraint. Teams lower in the standings are allocated more aero testing time than the leaders — a deliberate balancing mechanism. But the spending cap applies equally to all. The result is an optimisation problem each team solves differently.

Leading teams tend to shift resource to next season earlier, accepting that the current season is good enough. Midfield teams tend to keep bringing upgrades to the track, because one or two championship positions are worth more financially than the development cost.

That is why McLaren's and Williams' packages at Baku deserve closer analysis than the timing sheet. They reveal where those two teams sit in the optimisation problem, not merely how fast they were in one session.

Reading the timing sheet: Russell, Verstappen, Leclerc, and four thousandths

George Russell led FP1 with a 1:45.387. Max Verstappen followed at a 0.400s gap on 1:45.787. Charles Leclerc at 0.404s.

The gap between Verstappen and Leclerc is four thousandths of a second. In a race, such a gap signals a duel measured in centimetres. In an FP1 session at Baku, it signals something else: noise.

A simple calculation. With Baku's surface changing continuously through the session, a driver entering a fast run merely thirty seconds earlier than another can generate several hundredths from the grip effect alone. Add tyre variance, track temperature, wind direction on the long straight, and four thousandths becomes a product of timing order rather than capability.

The 0.400s gap between Russell and Verstappen sits entirely within the noise generated by track rubber build-up and soft-tyre timing, so it cannot be read as a genuine performance delta.

I want to stress the word genuine. It does not mean Russell is slower than Verstappen. It does not mean he is faster. It means the data cannot answer the question. In technical analysis, the ability to say "I don't know" is a skill, not a weakness.

There is one thing the timing sheet cannot hide, and it is the cleanest signal of the whole session.

The cleanest signal: Mercedes' intra-team gap

Russell first. Antonelli fifth. Same car. Same team. Same engineering department. Same pre-session strategy briefing.

This is the only comparison in an FP1 session where at least half the variables are controlled. Same aero package, same power unit, same setup philosophy. What remains — driving style, experience, confidence on a slippery street circuit — are human variables.

I must be explicit: Antonelli did not complete the full session. His car stopped mid-session with a power unit problem. So the gap between first and fifth cannot be read as a pure capability measure. He ran fewer laps, in a phase when the track had not reached peak grip.

But the signal still exists, and my habit of self-correction obliges me to state both sides.

Comparing two teammates in the same session carries higher diagnostic value than any cross-team comparison, because it removes the car-design variable and exposes the human part of the decision loop.

For Mercedes, that signal matters more than Russell's headline time. It suggests the W17 has a fairly narrow operating window, and that finding the sweet spot depends more on the driver's ability to set the car up than on the car's fundamental quality. Cars with narrow windows are usually fast in ideal conditions and lost in adverse ones. At Baku, where conditions shift by the minute, that is a risk characteristic.

I note the judgement down, and I will test it in FP2 and in qualifying.

McLaren's package: reading the pain point instead of copying a concept

In FP1, McLaren brought three new components to the track: new sidepods, a new floor edge, and a revised diffuser. They were introduced with an explicitly stated aim — to address the team's low-speed weakness.

In technical analysis, such a statement matters more than the components themselves. It shows the engineering group has identified the pain point, understood the mechanism causing it, and designed a solution aimed directly at that mechanism.

How do these three components work together?

Sidepods affect how airflow travels along the bodywork and how it is directed rearward. The floor edge affects how air is sealed at the outer lip of the floor, and therefore directly affects low-speed downforce generation. The diffuser affects how air is extracted from under the car and how downforce is distributed rearward.

All three attack a single mechanism simultaneously. This is the architecture of a solution with a clear correlation model, not a list of parts brought out to try.

McLaren targeting low-speed weakness directly suggests its relative 2026 advantage lies in high- and medium-speed aero efficiency, while it loses time in slow corners.

That claim has a concrete, testable implication. If correct, McLaren will be relatively strong at tracks with fast corners and relatively weak at tracks with slow ones. The season will confirm or refute it.

And Baku, in one specific respect, is a brutal test of that hypothesis.

Baku's slow castle section: a merciless test

Baku has a stretch engineers call the castle section, running through the old town with corners so tight two cars cannot go side by side. This section, around turns eight to fifteen, demands the car rotate at very low speed while keeping the front planted and the rear providing traction.

This is exactly the condition where a low-speed weakness shows itself most clearly.

A car short on low-speed downforce tends to understeer on entry. The driver must slow more than ideal. The time lost at one corner is a few hundredths. Multiply by ten corners in the castle section, and the total loss is large enough to change a qualifying result.

Worse, the problem does not stop at entry time. A car that understeers pushes front tyre temperatures higher than optimal. Higher temperature reduces grip. Reduced grip forces the driver to slow further. This loop produces a form of performance degradation I call geometric degradation, distinct from tyre degradation.

So when McLaren brings three new components to a circuit with a castle section, it is taking a calculated risk. Success validates its correlation model. Failure burns budget and precious aero testing time.

One FP1 session is not enough to conclude. It is enough to pose the question.

Williams: the most telling detail is weight

While McLaren drew attention, Williams brought a package with the most narrative-rich components of the session.

The team introduced a redesigned floor alongside a lighter chassis.

Each component carries its own meaning. But their appearance together in one package is what deserves analysis.

A lighter chassis delivers a benefit different in kind from a new floor.

A new floor acts on the aerodynamic field. It creates gains in certain operating regimes, dependent on speed, on body rake, on ride height. Some circuits suit it. Some are neutral. Some may even suffer if the aero balance shifts the wrong way.

A lighter chassis acts on mass. It delivers gains at every corner, every circuit, every temperature condition. Less mass means less inertia, allowing faster direction changes, later braking, less loading on the tyres.

Williams bringing a lighter chassis alongside a new floor suggests the team has carried a weight handicap for a long stretch, and that handicap has suppressed performance in every type of corner.

In recent years, Williams' story has been told through a familiar pattern. It is the story of a small team, with limited resources, trying to compete with bigger ones. That pattern is easy to hear, easy to sell, and easy to become an excuse for every poor result without looking at the technical data.

That romantic narrative hides a simpler operational reality. A car ten kilograms heavier than its rivals will be slower everywhere, regardless of whether the team has a big budget.

If Williams' package genuinely addresses the weight problem, its impact will far exceed a single aero component. It will shift the baseline of the whole season.

But here I must stop and self-correct.

Data limitations: what I have not measured

A lighter chassis does not automatically generate better performance. It changes weight distribution, and therefore changes the aerodynamic field and the thermal load on the brakes. An upgrade package combining new aero with new mass typically carries thermal side-issues before engineers have time to tune them out.

I have a specific suspicion, and I state it so I can test it later.

In FP1, there was data on brake overheating linked to the brake assembly of a car within the group developing a new package. That could be a direct side-effect of mass reduction. Changing car mass changes the thermal load the brake system must handle, and changes how airflow is routed to the brake cooling area.

This is a hypothesis, not a conclusion. I mark its confidence as low.

The lesson from summer 2026 still holds value for me here. Summer 2026 taught me that a gap is never empty; it is only waiting for the right reader. But that same summer taught me that the right reader must know what they have not yet read.

When there was no football, I drew football. And it turned out drawing is also a way of understanding. When the data is not good enough, I draw what is missing, and name it so I do not fool myself.

With Williams, what I have not measured is: the exact mass they removed, where they placed that mass, and how much their brake system had to change to adapt. Those three unknowns decide whether this package is a step forward or a temporary step back.

Audi: an upgrade in a different race

Audi also brought an upgrade to Baku. Details of that package are less specific than McLaren's and Williams', and I will not over-interpret what is not stated.

But one contextual point matters.

A new works team entering a regulation cycle must typically solve two problems at once: building operational processes and building regulatory understanding. Bringing an upgrade to Baku, at this stage of the cycle, signals the team has stabilised its processes enough to manufacture and validate new components within a permitted timeframe.

That is an organisational signal, not a performance signal. In long-horizon analysis, organisational signals usually matter more.

Reliability: the real engineering story of the session

This is the part I consider most important from the whole FP1, and the part session summaries usually place at the bottom as a footnote.

Three reliability-related events appeared in the same session.

First, Antonelli's W17 stopped with white smoke from the engine area.

Second, Isack Hadjar's car suffered power unit problems.

Third, Verstappen's RB22 drew a brake complaint.

Three events, two different teams, two different 2026 power unit suppliers.

The first two belong to the power unit failure category. The third to the brake system.

On the Mercedes event, white smoke from the engine area on a sudden stop typically points to a power unit or energy recovery system issue involving oil, coolant, or circulating fluid. That is logical inference from the location and nature of the symptom, and I mark confidence as medium, since I have no remote data to confirm it.

The important point lies in the governance consequence.

A power unit failure leading to a component change carries a component-quota consequence, and that consequence can surface as a grid penalty at some point in the season.

That is why a plume of smoke at FP1 deserves longer tracking than a four-thousandths gap on the timing sheet.

On the second event, a car from another team, using another power unit supplier, also hitting power unit trouble in the same session opens a broader hypothesis.

It may be that 2026-spec power unit components are still in an early phase of their development life, and durability has not reached stable levels. This is a sector-wide risk, not one team's risk.

I mark confidence for this hypothesis as low. Two events do not make a pattern. But two events in the same session at two different teams is a reason to watch, not a reason to conclude.

On the third event, the brake issue on the RB22 could belong to several causal groups: suboptimal cooling, brake pad material not matched to track temperature, or simply an incomplete setup in the first session. In FP1, brake complaints are common, because teams lack the data to pick an exact cooling configuration.

I do not read this event as a serious warning. I read it as data on how much work remains.

Two tyre phases and the trap of the time window

Now to the technical core of the session itself, the part I need to explain why the timing sheet cannot be read straight.

FP1 at Baku ran in two distinct phases.

The first was the hard tyre phase. Verstappen set benchmark times around 1:46.6. This was the baseline-building phase on a dusty surface, and its purpose was not fastest lap.

The second was the soft tyre phase, when teams switched to softs for the second half. This was when the track reached peak rubbering, and when the fast laps appeared.

The session's fastest laps landed in the final twenty minutes. This is the key information.

A run at minute twenty on softs has a different value from a run at minute fifty on softs. Same compound, same car, but the track has changed.

So when reading Russell's 1:45.387, one must know it was set in the session's highest-grip window. It was flattered by timing, and flattered systematically.

This is no team's error. It is the nature of an FP1 on a street circuit used once a year.

The difference between a session on a familiar circuit and one at Baku lies in how fast the surface changes. On a pre-rubbered circuit, the effect is small and correctable. At Baku, it is large and dominates the timing order.

That is why I keep the rule: do not read the FP1 order at Baku as a pecking order.

Yellow flags, virtual safety cars, and uncompleted runs

One more factor leaves this session's data incomplete.

The session was interrupted by two virtual safety cars for stoppages. These interruptions were no team's tactical choice. They were external noise.

But their consequences are concrete.

Late fast runs were disrupted by Arvid Lindblad losing control at Turn 7, and by the second VSC.

When a run is disrupted, the driver does not complete a clean lap on softs, and their time does not reflect their capability.

The clearest case is Lando Norris, who ended the session fourteenth without a representative soft-tyre lap.

That is important data on the completeness of the dataset.

With at least one front-running driver failing to complete a clean soft-tyre lap, the FP1 timing sheet at Baku cannot be treated as a complete dataset, even at the level of relative comparison.

In my analytical work, an incomplete dataset is a dataset I must flag as incomplete. I record it in the data limitations section at the end of every piece, a habit built since 2026.

For this session, the limitations list is longer than the conclusions list.

A counter-intuitive angle: the real signal is not speed

Here I want to give the angle I consider most important of the whole piece, and it runs against how most fans read an FP1.

The conventional read is: fastest is strongest, slowest is weakest. That read is convenient. It is also wrong in this case.

My read is: the informational value of an FP1 session lies not in the timing order, but in three data groups — upgrade direction, system reliability, and intra-team gaps.

These three groups share a trait: they do not depend on track conditions.

Upgrade direction is a design choice. It exists whether the track is slippery or gripped.

System reliability is a property of the car. It is unaffected by whether the surface has been rubbered in.

Intra-team gaps are measured under the same track conditions at the same moment, and therefore remove a variable that cross-team comparisons cannot.

These three groups determine much of what will happen over the rest of the season. FP1 timing determines very little.

There is a further implication I want to state clearly, because it is part of my professional outlook.

In F1, as in any sport with a complex economic structure, a permanent pressure pushes stakeholders to amplify stories that sell easily. A story about a small team overcoming odds always spreads faster than a spreadsheet on cost gaps. A story about a revolutionary upgrade always spreads faster than an explanation that the upgrade addresses one pre-identified pain point.

Romantic narrative hides financial gaps and sustainable operational reality. With Williams, that gap sits in weight and component manufacturing resource. With McLaren, it sits in having built a correlation model good enough to target the right weakness.

Those are verifiable data. Stories are not.

Execution blind spots: what an upgrade cannot fix

In technical analysis there is a permanent temptation: attributing every car problem to aerodynamics.

That is a blind spot.

A new aero package can generate more downforce. But more downforce only produces faster lap times if the driver can exploit it, and if the suspension and tyres can transmit it to the road without losing balance.

There are three blind spots I track separately when assessing an upgrade.

The first is compatibility with driving style. A car with more front downforce reacts faster on entry. For a driver with gentle entry style, that may create a sense of instability and reduce confidence. Confidence is a variable no sensor measures, and it directly affects lap time.

The second is compatibility with the tyres. A new aero configuration shifts tyre temperature distribution. Shifting tyre temperature shifts the tyre's operating window. A tyre may be faster over one lap and degrade faster over ten. That is a trade-off an FP1 struggles to detect, because teams run few consecutive laps.

The third is compatibility with the brake system. Mass changes, downforce changes, and brake cooling airflow changes. All three act on brake temperature. As noted earlier, this is a point I am tracking separately with Williams' package.

These three blind spots are why I do not write about an upgrade as an event, but as a process.

Transition: the real silence of a session

Transition is not a stretch of running. It is the silence between two intentions that few can read.

In a session, there are silences the timing sheet does not record, and that is where the real information sits.

The silence between two tyre phases. That is when a team decides to switch compound, and that decision reflects its model of how the track is changing.

The silence between two pit entries. That is when the data engineer talks on the radio, and when the driver answers. The exchange between analyst and driver is where performance is created or lost. It does not appear on the timing sheet.

The silence between braking point and turn-in. That is when the driver decides whether to trust the car's downforce or their own feel. With a new aero package, this silence carries the most risk, because feel has not been updated.

At Baku, there is a silence the circuit shape itself creates: the silence between the exit of a slow corner and the start of the long straight.

This is where the car must switch from low-speed rotation to maximum acceleration. Power unit output, low-drag aero efficiency, and rear traction control must coordinate in a very short window.

A car short on low-speed downforce loses time on entry and may not recover it on the straight. A car strong at low speed but weak in low-drag trim does the opposite.

So that silence is what I will focus on in FP2, not the fast laps in the final twenty minutes.

The geometry of gaps on a circuit

I approach F1 from football. My way of reading a circuit starts there too.

In football, I measure the space between lines. I measure the radius a player needs to receive in a specific gap. I measure the exit angle of a pass.

Mapped onto a circuit, I measure three things.

I measure corner radius. Radius determines the maximum speed a car can carry, and therefore the downforce required.

I measure braking points. Braking points determine the cumulative distance a car can recover or lose, and at Baku, braking happens fifteen times in a lap.

I measure exit angles. Exit angles determine the speed a car carries out of a corner, and therefore its speed at the start of the next straight.

These three measurements combine into a map of the engineers' intentions.

At Baku, that map has a clear structure: a long straight dominates time, a slow section dominates risk, and a medium-speed section links them.

For McLaren's package, the map predicts uneven effectiveness: improvement in the slow section, possibly no improvement or a slight loss on the long straight if the package raises drag.

For Williams' package, the map predicts a more even gain, because reduced mass acts everywhere.

For Mercedes, the map predicts that setup precision will decide qualifying more than the car's underlying quality.

Three predictions. Three things I will test over the next two days.

What I will verify in FP2 and qualifying

Analysis has no value unless it produces falsifiable predictions. Here is what I will track.

One, McLaren's ability to complete consecutive long runs. If the package fixes the low-speed weakness, it must show in consistent performance through the slow section across a multi-lap run, not just on one fast lap.

Two, brake temperatures of the cars running the new packages. If my hypothesis about the side-effects of mass reduction is right, overheating will reappear across a long run.

Three, the gap between the two Mercedes drivers under identical conditions. If the W17's window is genuinely narrow, the Russell-Antonelli gap will not be stable across sessions.

Four, the state of power unit components. If a component is changed, a quota consequence appears, and I will fold it into strategy analysis for the rest of the season.

Five, Norris's position in an uninterrupted session. Fourteenth in a VSC-disrupted session is empty data.

These five are not rhetorical questions. They are tests that can come back right or wrong.

A misplaced pass is data, and so is a plume of smoke

A misplaced pass is not an error. It is data the system is trying to send you.

In the same spirit, a plume of white smoke from a racing car's engine cover is not an unfortunate incident. It is data on a system that has not reached the required stability.

And a four-thousandths gap between two drivers on the timing sheet of a first practice session on a street circuit is not a sign of balance. It is a sign of noise.

In sports analysis, distinguishing signal from noise is a foundational skill. In F1 analysis it is harder, because the car is a complex system with thousands of interacting variables. And in an FP1 at Baku it is hardest, because we have only a few dozen laps to understand a machine built for an entire season.

I always start with a blank sheet and one line. The first line is usually shaky. I do not fix it immediately. I leave it, keep drawing, and look back at it at the end.

The first shaky line in this analysis is the FP1 timing sheet.

It is shaky because it was drawn on a surface that keeps changing.

A forward-looking close: what will decide the Azerbaijan GP

The FP1 timing sheet at Baku establishes an order I consider temporary. Russell leads, and that may be right, may be wrong, and I do not have enough data to tell.

What I do have enough data to say is three other things.

The team whose upgrade targets the right pain point will improve gradually through the season, not immediately.

The team that solves its weight problem will improve at every circuit, and that is a different kind of improvement from aero gains.

The team that has not solved power unit reliability will have to manage risk rather than optimise performance, and how it manages that risk will shape the rest of the season.

For the Azerbaijan Grand Prix, what I will track is the moment the slow section delivers its verdict. If a car short on low-speed downforce loses time there and cannot recover it on the straight, we will know which upgrades are real and which are just a parts list.

I will redraw that map by hand once qualifying ends. The line may still be shaky. The data will force me to read it correctly.