Baku FP1: Russell Fastest, but Six Mechanical Failures Are the Real Data

The Mercedes of Andrea Kimi Antonelli came to a halt just beyond the Turn 7 run-off. The wheels were intact, the engine was still running, but the car would not move another metre. A Virtual Safety Car was deployed seconds later, and on the Mercedes pit wall nobody looked at anybody. They looked at the data screens.

Fifteen minutes before that moment, Carlos Sainz had brought a heavily updated Williams back to the garage with its rear brakes on fire. Half an hour earlier, Nico Hülkenberg had parked his Audi at the exit of Turn 4. Through the session, Lando Norris barely completed a meaningful run, and Oscar Piastri reported vibrations over the radio.

The timing sheet still read cleanly: George Russell fastest on 1m45.387s, Max Verstappen second, 0.400s adrift, Charles Leclerc third, Lewis Hamilton fourth, Andrea Kimi Antonelli fifth. That is enough to write a tidy headline: Mercedes are back.

I once believed in the timing sheet, until the timing sheet was torn apart by an overtake. In Baku, where the circuit wraps around an old castle and then dives to the Caspian shore, every number has to pass through two posts: concrete and tyre temperature. First practice here is the dirtiest data a race weekend can produce.

Why Azerbaijan FP1 cannot be read at face value

A first practice session lasts sixty minutes. In those sixty minutes, teams must validate systems, confirm wind-tunnel and CFD correlation against the track, test two or three tyre compounds and measure fuel consumption. Nobody runs representative fuel loads, nobody publishes power-unit deployment modes, and nobody runs the same tyre programme as their rivals.

So when you read "Russell is 0.400s ahead of Verstappen", you are comparing two different programmes, at two unknown fuel loads, on two different compounds, on a surface whose grip is changing minute by minute. The error bar on that comparison is larger than the number itself.

At Baku this is truer than usual, because the lap is over six kilometres long and most of it is straight. A straight that long means the tow is worth more than at any other circuit on the calendar. And that is the single most important detail of this session.

Four tenths, and the shadow in front

Russell's fastest lap was set on the soft compound, and it was set with a tow down the main straight. This is not my speculation. It is information attached to the lap itself.

On Baku's main straight, where cars reach top speed for nearly two kilometres, the difference between having a tow and not having one is more than enough to account for most of that four-tenths margin. That gain does not reflect the speed of the Mercedes. It reflects the fact that somebody went out at the right moment on a busy track.

None of this means Russell is slow. It means that particular number does not measure what people want it to measure. A soft-tyre qualifying lap with a tow is a beautiful piece of graphics: it tops the sheet, it looks tidy, and it says nothing about the race.

Had Mercedes' programme stopped there, the whole session would have been a hollow investment. But something larger happened afterwards: Mercedes did not complete a single clean soft-tyre qualifying simulation with either car. Russell's car set the headline lap, but with a tow attached. Antonelli's car was on its way out of the pit lane on softs when it died.

Read that way, Mercedes topped the session without producing one comparable qualifying data point — the biggest paradox of Azerbaijan FP1.

Ferrari ran the medium, and that is the story

Leclerc and Hamilton finished third and fourth, close to Verstappen and Russell. They did it on the medium tyre, while Russell and Verstappen used the soft.

On a cold, dusty, rubbering-in street circuit, the medium takes longer to reach peak grip. The performance gap between medium and soft over a single qualifying lap in Baku is meaningful. When both Ferraris sit that close to the soft-tyre benchmark of Mercedes and Red Bull, there are two readings.

The first: Ferrari has a stronger car than the order suggests, and is deliberately saving softs for sessions where the data is worth more. The second: Leclerc and Hamilton are over-driving the medium to post a pretty number that will collapse when everyone runs softs.

Nothing in this session can separate those two readings. But one detail leans towards the first: the two Ferraris sat side by side with a tiny gap between them. A team-mate pairing that closely matched on the same compound is the cleanest data point the session produced.

While every other team ran different tyre programmes to hide its hand, Ferrari was the only team to put a directly comparable pair of data points on the board — and that pair sat in the leading group.

Antonelli and the hard-tyre peak

Antonelli finished fifth, which reads like an ordinary result. It is not.

Around the twenty-five-minute mark he was the fastest man on the hard compound — the hardest of the three nominated for the weekend, the slowest to warm up, and on a cold Baku surface close to the worst possible choice for a qualifying lap. His best time, 1m46.265s, was set in those conditions.

He then pitted, took softs, went out, and the car failed with a hydraulic problem on the out-lap. His real qualifying run never existed.

His fifth place is therefore built on the slowest compound, and was never upgraded to the number he might actually have produced. Informationally, it is a truncated sample — but the part that was measured was very strong.

Six mechanical failures across five teams

If I had to pick one number to describe this session, it would not be 1m45.387s. It would be six.

Six distinct mechanical events, across five different teams, inside sixty minutes: Antonelli's hydraulic failure, a safety-critical system that controls gearshifts, clutch and differential. Isack Hadjar's power-unit issue, which restricted his running and left him twentieth. Sainz's rear-brake fire, fifteen minutes in. Hülkenberg's Audi stopping at the Turn 4 exit. Norris losing most of the session to an issue, ending fourteenth. Piastri's vibrations, which cost him repair time and left him sixth.

Five teams. Six faults. Sixty minutes.

If the grid composition reads as I believe it does — a works Audi with Hülkenberg and Bortoleto, a Cadillac with Pérez and Bottas, Lindblad at Racing Bulls, Antonelli at Mercedes and Hamilton at Ferrari — then this is the first year of the new power-unit and active-aero cycle. In the first year of a new regulation set, this density of faults is close to inevitable: teams must simultaneously master a new energy system, a new cooling architecture and a new aerodynamic philosophy.

Here is what I want readers to hold onto. At this stage of a cycle, reliability predicts a weekend better than pace does. A fast car can take pole. A car that does not break can win.

The Williams upgrade is unvalidated, and possibly worse

Williams brought a significant upgrade to Baku. That is why this session mattered more to them than usual.

Fifteen minutes in, Sainz returned with his rear brakes on fire — a brake thermal event, where friction material overheats and ignites. It is not an isolated failure. It is a signal that the brake-cooling airflow is not behaving as designed.

On a car that has just undergone heavy aerodynamic change, that means the correlation between wind tunnel, simulation and on-track behaviour is unconfirmed. Worse, when an aero change affects brake cooling, the engineers face two variables at once: aerodynamics and thermal management.

Albon finished thirteenth, in the lower midfield. Neither car produced a trustworthy long run.

Two Virtual Safety Cars and the erosion of data quality

Two VSCs fragmented the session, the first for Antonelli's Mercedes at Turn 7 and the second for an incident involving Lindblad and Hülkenberg in the same area.

On a normal weekend, losing two long-run windows is a problem. In Baku, where teams need sustained running to rubber in the surface and understand degradation, it is worse. Every VSC forces every car to slow and hold tyre temperature in unfavourable conditions, and degradation data gathered in those windows is close to unusable.

The consequence is that the information value of the next two sessions rises sharply. Whoever recovers a clean long run in FP2 gains a disproportionate edge. And in Baku, an information edge on tyres usually converts straight into a strategic edge on Sunday.

A quick scan of the order

Mercedes led with Russell and Antonelli first and fifth, but both numbers carry asterisks — one for the tow, one for the truncated sample. Red Bull were second with Verstappen while Hadjar sat twentieth with a power-unit fault: Verstappen carrying the team alone in a session where nobody knows anyone's fuel load.

Ferrari produced the best-matched pairing of the session on the harder compound. McLaren had the period's clear negative outlier: Norris fourteenth after losing most of the session, Piastri sixth with vibrations.

Behind them, Racing Bulls put rookie Lindblad in the top ten — a driver who also reversed out of the Turn 7 run-off, two data points pulling in opposite directions. Haas ninth. Audi tenth and eleventh despite one car stopping at Turn 4. Williams twelfth and thirteenth. Alpine sixteenth and eighteenth. Cadillac seventeenth and twenty-second. Aston Martin nineteenth and twenty-first.

None of that is a verdict. It is a snapshot taken through a low-grip, wall-adjacent, VSC-fragmented hour.

Where I could be wrong

There is a simpler reading. On that reading, Mercedes really are quick, Russell really is quick, and the margin is the natural consequence of going out early and finding a clean lap. The tow is part of racing; every driver tries to use it, and nobody loses points for it.

That reading has real strength. In Baku, a tow is not cheating, it is traffic management. A driver who goes out at the right moment with traffic ahead understands the circuit. If Russell repeats it in qualifying, he will deserve the top of the sheet.

I may also be wrong about the regulation cycle. I read this grid composition as a marker of the new rules, but that is my inference, not a published fact. If I am wrong, my entire "first-year fault density" argument loses a leg.

And I may be underrating Ferrari. They could be saving softs deliberately and will separate from the pack in FP2 and FP3. They could equally be running mediums to mask a soft-tyre temperature problem they have not solved.

I have not tipped my hat to anyone this session. I have only placed a bet.

Three falsifiable bets for the next two sessions

First: when everyone runs softs in FP3, the Russell-Verstappen gap will be smaller than 0.400s, and at least one Ferrari will be within 0.150s of the benchmark on equal rubber.

Second: at least one of this session's six mechanical faults will recur in FP2 or FP3, and if it recurs in qualifying it becomes a grid penalty or an aborted session.

Third: Williams will not complete a clean long run in FP2. An unvalidated upgrade on a street circuit usually needs another session to answer its first question.

Baku is a city where the circuit passes through a castle that is only a few dozen centimetres wider than the cars. There is no room for error. And in a session where every team is guessing, the winner is whoever does not break — not whoever has the prettiest timing sheet.

The timing sheet will sleep soundly tonight. The fault list will not.