pace_delta = fuel + compound + constructor + dirty_air + driver_skill. In the table and text below, positive means slower than the field baseline and negative means faster.
The problem with lap time as a single number
Verstappen’s lap times in the first laps of the final stint looked almost the same as Hamilton’s. Over laps 45 to 50 (lap 48 is not in the clean-lap table) Hamilton averaged 72.36 s and Verstappen 72.77 s. Commentary read that as matched machinery. The decomposition finds a gap of 0.41 s a lap built from four small pieces that partly cancel:
The pieces are not shown by the lap time alone.
Setup: two drivers, two tyre ages
Both drivers made two stops. The tyre ages in the warehouse reset after lap 43 for Hamilton and after lap 40 for Verstappen, so Hamilton’s final set was three laps younger at every lap of the last stint. The table shows selected laps fromfct_lap_residuals.
Compound is compound_component_s and Driver Skill is driver_skill_residual_s. The constructor term is constant in this race: Mercedes −1.51 s, Red Bull −1.36 s. The fuel term is the same for both cars on a given lap. The dirty-air tax takes only two values in this race, 0.00 s or +0.33 s a lap.Finding 1: the car was worth a little, not nothing
The constructor term puts Mercedes 0.15 s a lap ahead of Red Bull in this race. The earlier reading of identical cars does not hold. The driver isolation split of the same race gives a larger car effect. On 18 matched laps (same compound, tyre age within two laps) it puts the car at +0.67 s a lap in Hamilton’s favour, a pace gap of −0.52 s a lap (Verstappen’s isolated pace was faster), traffic of −0.18 s a lap against Hamilton, and relative pace of −0.041 s a lap (level). The two estimates agree on direction and disagree on size: 0.15 s from the season-level constructor term, 0.67 s from the per-race car term. The gap is not lap selection: on the pre-cliff laps the isolation uses, the raw Mercedes–Red Bull difference after fuel, compound and dirty air is 0.19 s a lap, close to the constructor term. The 0.67 s comes from the isolation’s fixed-effects fit, which sets the car term jointly with driver levels estimated across other races, so treat the 0.15–0.19 s range as the within-race evidence for the car.Finding 2: the tyre-age edge from the later stop is small
Hamilton’s final set was three laps younger. The compound term prices those three laps at 0.06 to 0.07 s a lap throughout the stint:
Summed over laps 59 to 71, the tyre-age difference is 0.86 s of an 11.0 s gap in total time over those 13 laps. The earlier 2.91 s figure for the strategic tyre advantage was an artefact of the seven-term model and is withdrawn. As a measured quantity, the later stop is worth under one second here.
Finding 3: lap 59 is the pass, and the residual carries it
Lap 59 is the lap on which Hamilton took the lead (his position goes from second to first). Verstappen’s lap was 1.69 s slower than Hamilton’s, 74.07 s against 72.38 s. The five terms split it as follows:
The residual is 1.80 s of the 1.69 s gap. It is large because Verstappen lost time on the lap he was passed. A residual that is produced by a pass is not independent evidence of better driving. Hamilton’s own residual that lap was −0.36 s, in line with his laps before the pass.
The dirty-air cost before the pass
Hamilton ran in Verstappen’s wake for 11 of the 13 laps from lap 45 to lap 58 that are in the table (lap 48 is missing). At 0.33 s a lap the tax totals 3.6 s over those laps. Over the whole race the tax comes to 9.6 s for Hamilton (29 laps) and 1.0 s for Verstappen (3 laps). The two clean-air laps in that window, laps 45 and 46 on fresh tyres, were his fastest of the stint at 72.14 s and 71.98 s. The earlier claim that clean-air laps 59 to 65 confirm a 72.0 to 72.1 s capability does not hold: those laps ran 72.38 s to 73.29 s.What the residual does across the stint
Before the pass, laps 45 to 58, the mean pace delta was −0.58 s for Hamilton and −0.41 s for Verstappen, a gap of 0.17 s a lap. The residuals were −0.32 s and −0.14 s, a gap of 0.18 s. This gap is about the size of the car term. After the pass, laps 59 to 71, the gap becomes 0.85 s a lap (Hamilton −0.09 s, Verstappen +0.75 s). Residuals are +0.61 s for Hamilton and +1.27 s for Verstappen. Both drivers end the stint slower than the model predicts. The compound term is a gentle curve (about 0.6 s at 28 to 31 laps), so it under-predicts late-stint wear for both cars, and Verstappen’s older set loses more than the model allows. The warehouse cannot say why. The residual absorbs everything the four modelled terms do not: car-specific tyre wear, tyre management, engine modes, a driver backing off after losing the lead, and driver skill. The earlier page read the residual as effort (“running 9 to 13 seconds faster than the model’s structural prediction”). Under the current identity the residual is positive at the end of the stint, and no such reading is available.Conclusion: where the 11.0 s came from
Over laps 59 to 71 Hamilton’s cumulative time was 950.9 s against Verstappen’s 961.9 s, a gap of 11.0 s, in the same sense as the 10.5 s winning margin. The five terms split it as:
About three quarters of the gap sits in the residual. The car and tyre age together explain about 2.8 s. Hamilton’s wake penalty cost him 0.3 s in this window.
Reconciliation with driver isolation
The race-level driver rating inmart_race_driver_summary gives Hamilton −0.306 s a lap (95% interval −0.45 to −0.16) and Verstappen −0.177 s (interval −0.38 to +0.02). The rating is the median gap to the teammate, so Hamilton was 0.31 s a lap ahead of Bottas and Verstappen 0.18 s ahead of Pérez. The intervals overlap, so the race rating does not separate the two drivers.
The driver isolation page gives the same picture from matched laps: relative pace of −0.041 s a lap, which is level, with Hamilton’s edge placed in the car and Verstappen’s isolated pace ahead. That sample covers 18 laps, mostly before the final stint, because the final-stint laps are three tyre ages apart and fall outside the matching window. This page covers the stint the matching excludes, and finds the gap there is large and mostly residual.
Taken together, the data supports these statements:
- Hamilton won through a large pace advantage in the last 13 laps, 0.85 s a lap.
- The car and the later stop account for about a quarter of that advantage. Dirty air cost him about 3.6 s before the pass.
- The remaining three quarters is residual. It is consistent with Hamilton driving better, with Verstappen’s tyres falling away faster than the model expects, or both. The warehouse does not separate them.
- The race rating and the matched-lap split do not show Hamilton as the faster driver in the race. “Brilliant driving” is not a measured result here.