Miami was decided on an out-lap

After Melbourne, Shanghai and Suzuka there was a pattern the teams could start to trust: the 2026 tyres degraded little and stood very long stints, especially the hard. But Suzuka had been run on C1–C3. Miami went the other way: C3 as hard, C4 as medium, C5 as soft – the three softest in the range. The surface is smooth and Pirelli knew from previous years that degradation is normally low, so one stop was the main line before the weekend had started, according to F1.com. Two uncertainties remained: the heat and the weather.

The sprint weekend this time gave a 90-minute practice session, and it proved valuable. The hard was used a lot, partly because teams were testing upgrades and power-unit work at the same time. The gap between compounds was around half a second, close to the simulations, and degradation was low – mainly thermal, driven by very high track temperatures, rather than wear, according to F1.com. Now there was 2026 data behind Pirelli's Miami model.

McLaren arrived with a major upgrade package and finished 1–2 in the sprint. Mercedes said after the weekend that their rivals' development steps had been considerably bigger than expected, according to F1.com. That matters for strategy: against a slower car you can wait; against an equal one you start thinking about undercuts and windows. Which is exactly what Mercedes would do.

After qualifying Pirelli were unusually clear: one stop was around ten seconds faster than two. The main strategy was medium to hard stopping around laps 22–28; soft to hard stopping on 16–22 was the alternative; medium to soft worked but required stretching the first stint to laps 32–38, according to F1.com.

Then came the weather forecast. Heavy storms were expected later on Sunday and the start was moved three hours earlier; even so there was about a 40 percent chance of rain during the race, according to F1.com. That changed the sums. If rain arrives around lap 25, nobody wants to have made a slick stop on lap 22 and then another for intermediates on lap 27. Being able to extend the first stint had a value of its own. That explains the starting tyres: although soft to hard was fully viable, practically the whole field chose the medium. The only exception was Hadjar, from the pit lane on the hard, according to press.pirelli.com. The medium gave start performance, a long stint and the ability to react to both rain and a Safety Car. Flexibility was worth more than the soft's extra grip.

Antonelli had pole, Verstappen P2. Both locked up into Turn 1; Verstappen hit Leclerc and spun, and was 8.8 seconds behind the leader after the first lap. Leclerc led, Antonelli ran second, and on lap 6 Norris got past Antonelli. The same lap Hadjar crashed, and Gasly after contact with Lawson. Safety Car, laps 6–11, according to F1.com.

Red Bull brought Verstappen in immediately: lap 6, medium to hard, sixteen laps before Pirelli's window, with 51 laps left to run on the same set. In isolation that is bold. With the information Red Bull had it was something else: Verstappen had already lost track position, the Safety Car made the stop cheap – 7.8 seconds to the field instead of 19.6 under green – and Shanghai had shown the hard could go almost any distance. For the leader the risk would have been large. For a car running eighth the upside outweighed the downside.

The simulator says the stop was almost neutral in time: had Verstappen stayed out and made a normal stop on lap 25 he would have finished P4 instead of P5, 0.8 seconds quicker – but with eleven laps stuck behind Colapinto. The stop bought clean air. A second stop for mediums on lap 35 would have cost a place.

This is where Miami differs from the first three races, and it shows in the numbers. Field degradation on the C3 hard was 0.046 seconds per lap – more than on the C4 medium, 0.027. The hard was 0.45 seconds quicker from new, but that advantage was used up after about 24 laps. Verstappen's 51 laps in ten-lap averages: 94.1 – 93.5 – 93.6 – 93.5 – 94.5. Pirelli noted afterwards that he needed no real pace management to get the tyres to the flag, according to F1.com, and that is true – but the last ten laps were a second slower than the middle of the stint. It was just enough: 0.9 seconds behind Russell at the flag. Hamilton had the steepest hard slope at the front, 0.087 per lap, and finished P6.

After the restart came the radio messages: rain in twenty minutes, rain in a few laps, drops on the track, according to F1.com. At the same time Pirelli's window was approaching. Stop now and it rains five laps later, and slick–slick–intermediate is a catastrophically expensive sequence. So why not wait?

Mercedes brought Russell in on lap 20, medium to hard – slightly before the window, as an undercut on Leclerc who was 2.1 seconds ahead. Ferrari responded on lap 21, but the stop took 3.7 seconds stationary and Leclerc came out behind Russell. His radio message says everything about Ferrari's dilemma: "Why did we stop? When is the rain?", according to F1.com. The pit wall protected the position against an immediate risk; the driver saw the bigger risk further ahead. Both were rational. The rain never got heavy enough, and with hindsight the stop was right.

But the undercut held for only ten laps. Ferrari was quicker on the hard – Leclerc's base pace 93.56 against Russell's 93.87 – and on lap 31 Leclerc was back past. Had Russell waited until lap 26 he would have finished P4 anyway. Russell's stop won him nothing lasting; what gave him P4 came on the last lap, and we will return to it.

Norris led from lap 13, with Antonelli 2–3 seconds behind, and both had mediums that were working. Mercedes did not bring Antonelli in with Russell. They waited until lap 26 – inside Pirelli's window, 22.1 seconds in the pit lane, and a clear undercut attempt on the leader, according to F1.com. Norris stopped on lap 27. He left the pits just ahead of Antonelli. But Antonelli's hards had a lap in them; Norris's were cold. By the end of Norris's out-lap Antonelli was 0.2 seconds ahead. Mercedes afterwards described exactly that – the tyre temperature that let Antonelli attack straight after McLaren's stop – as the decisive factor, according to F1.com.

The simulator shows how wide the window was. Had Antonelli stopped on the same lap as Norris he would have finished P2, 4.5 seconds behind. Had he stopped as early as lap 22 he would still have won, with a smaller margin. Antonelli won with a stop anywhere between laps 20 and 26; lap 26 simply gave the biggest margin. What decided it was McLaren's answer.

Norris was leading. Rain still threatened. His mediums were working. Pirelli's window ran to lap 28, and a later stop meant fresher hards for the end. Waiting was rational. But had McLaren covered Antonelli and stopped Norris on lap 26 Norris would have won, according to the simulator – and the same with a stop on lap 24 or 25. The margins are two or three tenths either way; this was a race won and lost on an out-lap.

Piastri waited until lap 28 and paid for it: on laps 25–27 he ran 94.0–94.6 against Norris's 93.0, four seconds in three laps, before he had even stopped. After that he matched Antonelli's pace on the hard, and the fresher tyres gave him P3 on the penultimate lap. Verstappen, leading on lap 28 on 21-lap-old hards once everyone else had stopped, was passed by both Antonelli and Norris on lap 29.

Aston Martin kept Alonso out until lap 41 waiting for a Safety Car or rain, then fitted softs to use the offset, according to F1.com. The offset lasted one lap: 93.5, then 96–97. The C5 soft sloped at 0.051 seconds per lap and was slower than his worn mediums after four laps. A normal stop on lap 26 would have given P15 – exactly the same place. A strategy that is absurd for P1 can be rational for P15, because it costs nothing.

Ferrari's strategy should not be judged on P8. Leclerc had worked his way back to P3 on the road and held it from lap 50 to 55, ahead of Verstappen and Russell. On the penultimate lap Piastri caught him on fresher tyres and took the place. On the last lap, chasing Piastri, Leclerc lost the car in Turn 3, spun and hit the wall; he continued with a damaged car, left the track repeatedly, lost places to Russell and Verstappen, crossed the line P6 – and was then given a 20-second time penalty for gaining an advantage off track, which made it P8, according to F1.com. Leclerc took the blame himself. Strategically he was on the podium with two laps to go despite a slow stop and a stop he did not want to make; it was not the pit strategy that cost him.

Judged on the information available when the decision was made:

• Pirelli – M → H, one stop, two stops ten seconds dearer: very accurate.
• Red Bull/Verstappen – hard on lap 6 under the SC: a rational asymmetric risk, almost neutral in time but it bought clean air.
• Mercedes/Russell – undercut on Leclerc: worked for ten laps, then Ferrari was back past.
• Ferrari/Leclerc – covering Russell despite the rain threat: defensible, and back to P3 on the road.
• Mercedes/Antonelli – stop on lap 26 against the leader: the race's strongest tactical call, with a wide window behind it.
• McLaren/Norris – waiting one lap: rational under the rain threat, but the lap that lost the race.
• McLaren/Piastri – lap 28: too late, four seconds gone before the stop.
• Aston Martin/Alonso – 41 laps on the medium: a free gamble with no payout.

Everyone knew the strategy: medium to hard, one stop. The tactics – exactly when the stop is made relative to the rival – are where Mercedes won Miami. And after four races the teams had a new data point to weigh: even the C3 could be run for 51 laps, but unlike the C1 at Suzuka, it cost something. The hard was no longer free.

Sources: F1.com (preview, Pirelli comments, race report, team radio, stewards' decision on Leclerc), press.pirelli.com, FastF1 data for lap times, stint slopes, pit times and running order per lap. Simulator pit loss: 19.6 s under green, 7.8 s under the Safety Car (measured). Lap numbers refer to the lap on which the driver entered the pits.