Austin: the hard tyre we wrote off won the race, and Antonelli won it from the front
Antonelli led from pole and won by 13 s on medium → hard with the stop on lap 24 – Pirelli's Sunday plan and Ruth Buscombe's reading of Mercedes, not ours. Verstappen, with the two new softs we had just written about, turned them into a two-stop and third place behind Leclerc, who ran soft → medium → hard. The hard that Friday said was the worst tyre ran 40 laps in both McLarens and wore at nothing measurable.
Antonelli won the United States Grand Prix from pole with one stop, medium → hard on lap 24, 13.1 s ahead of Leclerc, with Verstappen 4.4 s further back and Hamilton one second behind Verstappen. Norris, Hadjar, Piastri and Russell completed the top eight. The leader was never under pressure after the stop, and the race at the front was decided by the choice of tyre for the second stint and by how many stops each car needed.
We planned soft → medium with the stop around lap 27–29, with a two-stop soft–medium–soft as the drawer plan if the first stint wore at Friday's rate. Nobody in the top eight ran our plan. Mercedes and McLaren ran medium → hard, Hadjar soft → hard, Ferrari soft → medium → hard, Verstappen soft → soft → medium. The winning plan was the one Pirelli's Sunday guide put in its window 21–27 and the one Ruth Buscombe expected from Mercedes; the stop on lap 24 sat in the middle of both.
What decided it was the hard tyre. Our Friday fit had it wearing at 0.23 s per lap, the worst of the three, and Pirelli's own note gave it "a limited role". In the race the field's joint fit gives 0.035 s per lap raw and 0.025 after the fuel correction, over 444 clean laps in 18 runs. Piastri ran 41 laps on one set with a slope of −0.004; Norris 40 laps at +0.033; Antonelli 32 laps at −0.012; Hadjar 31 laps at −0.040. A tyre that does not wear makes the one-stop with an early switch to it the fastest race, and that is the race Mercedes ran.
Verstappen's race is the other story. Red Bull used his two new softs exactly as we described on Saturday: a new soft for the start, the second new soft on lap 13, and the medium on lap 31. It gave him the fastest race pace in the field, 99.12 s per lap on our fuel- and age-corrected measure against Leclerc's 99.45 and Antonelli's 99.97, and it cost him a second pit stop, 21 s at Austin, against a leader who needed one. He finished 17.5 s behind. On pace alone the one-stop soft → hard that Hadjar ran would have put Verstappen in the fight with Antonelli; whether he would have passed him on this track is the question the overtaking numbers answer below, and the answer is probably not.
The race wear, from the library's joint fit on the clean laps: soft 0.038 s per lap of age raw over 352 laps in 22 runs, medium 0.056 over 174 laps in 10 runs, hard 0.035 over 444 laps in 18 runs; after the fuel correction 0.033, 0.036 and 0.025. Friday had said 0.152, 0.163 and 0.232. The overstatement is four times on the soft, three on the medium and seven on the hard, and the hard is the one we did not discount. We wrote in the briefing that Fridays had overstated the race by two to eight times this season and then planned the hard as if Friday were right about it. The reason is probably simple: the Friday hard runs were short, on a green track, in the hottest part of the day, and a compound that takes long to switch on looks worst in exactly those conditions.
Pace by compound in race trim, from the long-run levels: hard 101.04 s per lap, medium 101.35, soft 102.63. The soft figure carries the opening stints with full tanks and traffic, so the gap is not all the tyre, but there was nothing left of the half-second soft advantage we had from Friday. The medium was three tenths off the hard and wore faster. On this track, this Sunday, at 36–40 °C, the hard was the race tyre.
The question we asked ourselves on Saturday afternoon – what a soft from qualifying costs – got its first measurement. Hamilton started on a used soft, the timing data shows the set at five laps of age on his third racing lap, which is a Q1 set with three laps on it. Leclerc started on a new soft. Their first stints: Hamilton 14 laps, slope 0.058 s per lap, mean 101.35; Leclerc 15 laps, slope 0.066, mean 101.15. Verstappen on a new soft: 13 laps, 0.071. The used set wore no faster, and the two tenths of level between the Ferraris is about what separates them in race pace over the whole afternoon, 99.54 against 99.45. One pair of cars is not a law, but the first reading is that a scrubbed Q1 soft costs a tenth a lap at most and nothing in stint length. Our allocation model ages a used set by four laps; on this evidence that is about right for the laps and generous on the pace.
The back-straight measurements from Friday said Mercedes arrived at Turn 12 with the biggest speed drop and Red Bull the smallest, and we read that as Mercedes saving energy for Turn 1 and Red Bull defending into 12. The race repeated the pattern, with 44–47 laps per driver behind the medians: Verstappen's speed fell 8 km/h between its peak and the braking point for Turn 12, Hadjar's 11.5, Hamilton's 11, Piastri's 13, Norris's 14, Russell's 19, Antonelli's 20, Leclerc's 20. Peak speeds were 317–331 km/h, 10 km/h below qualifying, as the race maps always are. The order from Friday held: Red Bull arrives at Turn 12 with the most speed in hand, Mercedes with the least, and the race did not test what that was for, because Antonelli never had a car within a second behind him and the passing at the front happened in the pit lane. The one new reading is the Ferrari split: Hamilton 11, Leclerc 20, on the same car and the same tyres. Full table in the Reddit follow-up on the back straight, same method as Friday.
Ferrari's fourth floor package of the season went on both cars and Ferrari had the second- and third-fastest race pace behind Verstappen, 0.3 and 0.4 s per lap quicker than Antonelli on our measure. It did not produce a win, because the pace advantage was smaller than one pit stop and the track did not allow it to be used on the leader. Mercedes brought nothing new and won; the fourth-best race pace in the field was enough from pole on a one-stop that nobody could undercut.
Dry, track 36–40 °C through the race, within the 6 °C of Friday that our temperature rule asks for, which is why Friday's wear went into the plan. The rule held; the Friday numbers did not. The lesson belongs to the tyre engineer, not the weather: at matched temperature the Friday slopes were still three to seven times the race slopes.
Pit loss 21.2 s under green, the library's median of two Austin races, and the 2026 stops fell on it. The VSC on lap 51–52 for Bottas's stopped Cadillac came too late to matter for anyone who had not already stopped; Bortoleto had taken softs on lap 50 and Lindblad, Alonso and Stroll had all switched to soft between laps 40 and 43 for the finish.
Mercedes ran what their tyre count allowed: new medium, new hard, stops on laps 24 and 22. It won with Antonelli and finished eighth with Russell, whose race pace was 0.8 s per lap off his team-mate's. McLaren ran medium → hard with the earliest stops in the top ten, laps 15 and 16, and then 40 and 41 laps on the hard; the strategy was right, the pace was not, Norris 1.2 s per lap and Piastri 1.7 off Verstappen on our measure. Ferrari split nothing and ran the same three-compound race with both cars, stops within a lap of each other; it produced second and fourth and the fastest final stints of the race, Leclerc's 13 laps on the hard at a mean of 98.79. Red Bull split its cars by tyre count: Verstappen two stops on the two new softs, Hadjar one stop soft → hard on lap 25 for sixth from third on the grid. Racing Bulls put Lindblad on a late soft and gained nothing; Aston Martin tried the same with both cars for fifteenth and sixteenth.
Race pace, fuel- and age-corrected, laps in clean air or steady running: Verstappen 99.12 s, Leclerc 99.45, Hamilton 99.54, Hadjar 99.91, Antonelli 99.97, Norris 100.36, Lindblad 100.64, Lawson 100.77, Russell 100.81, Hülkenberg 100.81, Piastri 100.83. The winner was fifth on pace and first on the road all day, which is what pole and a one-stop on a non-wearing tyre buy at Austin. Hadjar's one-stop was the cleanest race in the field: 22 laps on the soft at +0.014 per lap, 31 on the hard at −0.040.
The start decided more than the tyres did. Nobody in the top eight changed position on lap 1, and the lap-1 telemetry says why: Antonelli covered 66 m in the first three seconds and reached 100 km/h in 2.16 s, the best launch in the field together with Piastri (67 m, 2.16) and Norris (67 m, 2.36) – all three on the medium. Verstappen on a new soft did 58 m and 2.60, Leclerc 59 m and 2.60, Hadjar 51 m and 2.88; Hamilton on his scrubbed Q1 soft 64 m and 2.36, quicker off the line than both cars ahead of him. Eight metres after three seconds is more than a car length, and it was the margin that gave Antonelli the inside of Turn 1 without a fight. The soft bought no traction at the start; the three best launches were made on the medium. The telemetry samples at 0.24 s, so the times are good to a tenth, and we do not measure wheelspin directly.
No championship was decided. Antonelli leaves Austin on 365 points against Russell's 242 and Hamilton's 241, with six races left; the first chance is Mexico City. Bearman retired on lap 24 after three stops; Colapinto stopped on lap 35 and Bottas on lap 51 with an engine problem.
The hard tyre, and with it the plan. We called it the weakest tyre on Friday's 0.23 s per lap and planned around the soft and medium; the race measured the hard at 0.025 and it carried the winner 32 laps and the McLarens 40. Pirelli had the stop window right, 21–27 for medium → hard, and we argued against it with Friday numbers we had ourselves said were unreliable. Our Saturday answer of soft → medium around lap 27–29 was a race that nobody in the top ten chose.
The tyre count. Verstappen's two new softs became a two-stop, Hamilton's scrubbed Q1 soft started his race without visible cost, and Mercedes ran medium → hard because it was what they had. Verstappen's long-run wear on Friday, 0.07 on the soft against Leclerc's 0.16, matched the race: his soft stints wore at 0.07 again. And the general warning that Fridays overstate the race held for the fourth weekend in a row – the error was not believing it about the hard.
Whether Verstappen on a one-stop would have beaten Antonelli. He had the pace for it, 0.85 s per lap on our measure; he would have needed to pass on track, and Austin's overtaking threshold this season sits at 0.82 s per lap of pace advantage, which makes it a coin toss, not a given. And whether the scrubbed-soft finding is general: one pair of Ferraris, one track. We will keep counting.
Simulate
Change one driver's pit stops and let everyone else run exactly as they did. The altered driver is modelled with his own pace per compound from the race, degradation, fuel, traffic and the VSC discount – cars that get stuck behind him lose time.