Skip to content
Reference Lap

Compare drivers

Two drivers over the grands prix they both started in a season. Only rounds in the same car compare the drivers with the car taken out; for the rest, the model's rating difference is the estimate separate from the car, with its uncertainty.

Swap

Changing the season keeps the drivers when both raced that season; otherwise it picks a pairing.

Head-to-head

ANTAndrea Kimi AntonellivsRUSGeorge Russell· 20 shared races, 20 as teammates

Teammates in every shared race: the same car, so this is the cleanest comparison.

ANT first.
MeasureANT–RUS
Weekends together20
Qualifying ahead11–9
Race ahead, both classified12–6
Race ahead, finish order (a DNF counts as behind)14–6
Sprint ahead, finish order2–4
Points (grand prix and shared sprints)390–262
DNFs0–2
Average finish vs the model's expected finish before the race, over the shared races, retirements included; gains in places per race, + = better than expected.
DriverRacesAvg finishAvg expectedVs model, shared racesSeason, all races ± SE
ANTA. Antonelli203.44.20.83 places better than the model expected0.83 places better than the model expected ± 1.01
RUSG. Russell206.74.62.11 places worse than the model expected2.11 places worse than the model expected ± 1.41

Measured against the model, which already includes each driver's rating and car: a surprise, not a comparison of the drivers. Avg finish is over the races with an expectation.

Rating difference by round

Andrea Kimi Antonelli's rating minus George Russell's, where each rating is the estimated pace vs the mean driver (model). Negative (higher on the chart) = ANT estimated faster; the band is ±1 sd. A difference within about 1 sd is not a clear difference.

In rounds as teammates the sd uses the model's covariance of the two ratings (teammates' estimates are linked).

Round 20: race −0.199% ± 0.106 (ANT estimated faster), qualifying +0.023% ± 0.098 (not a clear difference).

race rating: ANT − RUS
Data table
race rating difference ANT − RUS, % (± sd).
RoundANT − RUS
R10.220 ± 0.145
R20.136 ± 0.127
R30.082 ± 0.121
R4-0.028 ± 0.112
R5-0.085 ± 0.107
R6-0.146 ± 0.106
R7-0.204 ± 0.109
R8-0.181 ± 0.107
R9-0.244 ± 0.103
R10-0.283 ± 0.104
R11-0.267 ± 0.107
R12-0.242 ± 0.104
R13-0.229 ± 0.104
R14-0.277 ± 0.105
R15-0.208 ± 0.106
R16-0.150 ± 0.107
R17-0.112 ± 0.105
R18-0.209 ± 0.106
R19-0.213 ± 0.106
R20-0.199 ± 0.106
qualifying rating: ANT − RUS
Data table
qualifying rating difference ANT − RUS, % (± sd).
RoundANT − RUS
R10.327 ± 0.131
R20.247 ± 0.112
R30.148 ± 0.108
R4-0.016 ± 0.098
R50.036 ± 0.093
R6-0.051 ± 0.094
R7-0.019 ± 0.095
R80.047 ± 0.095
R9-0.061 ± 0.090
R10-0.121 ± 0.092
R11-0.147 ± 0.093
R12-0.082 ± 0.089
R13-0.043 ± 0.091
R14-0.090 ± 0.092
R150.033 ± 0.097
R160.028 ± 0.098
R170.070 ± 0.092
R18-0.016 ± 0.093
R190.016 ± 0.093
R200.023 ± 0.098
Differences by round, with the sd used
RoundCarsRace ± sdQuali ± sdsd
R1same car+0.220% ± 0.145+0.327% ± 0.131with covariance
R2same car+0.136% ± 0.127+0.247% ± 0.112with covariance
R3same car+0.082% ± 0.121+0.148% ± 0.108with covariance
R4same car−0.028% ± 0.112−0.016% ± 0.098with covariance
R5same car−0.085% ± 0.107+0.036% ± 0.093with covariance
R6same car−0.146% ± 0.106−0.051% ± 0.094with covariance
R7same car−0.204% ± 0.109−0.019% ± 0.095with covariance
R8same car−0.181% ± 0.107+0.047% ± 0.095with covariance
R9same car−0.244% ± 0.103−0.061% ± 0.090with covariance
R10same car−0.283% ± 0.104−0.121% ± 0.092with covariance
R11same car−0.267% ± 0.107−0.147% ± 0.093with covariance
R12same car−0.242% ± 0.104−0.082% ± 0.089with covariance
R13same car−0.229% ± 0.104−0.043% ± 0.091with covariance
R14same car−0.277% ± 0.105−0.090% ± 0.092with covariance
R15same car−0.208% ± 0.106+0.033% ± 0.097with covariance
R16same car−0.150% ± 0.107+0.028% ± 0.098with covariance
R17same car−0.112% ± 0.105+0.070% ± 0.092with covariance
R18same car−0.209% ± 0.106−0.016% ± 0.093with covariance
R19same car−0.213% ± 0.106+0.016% ± 0.093with covariance
R20same car−0.199% ± 0.106+0.023% ± 0.098with covariance

Shared races

Every grand prix both started. “Different cars” rows compare cars as much as drivers.
RoundCarsANT qualiRUS qualiANT finishRUS finishAheadANT vs modelRUS vs model
R1same car21P2P1RUS3.2 places better than the model expected2.4 places better than the model expected
R2same car12P1P2ANT2.6 places better than the model expected1.3 places better than the model expected
R3same car12P1P4ANT2.1 places better than the model expected0.8 places worse than the model expected
R4same car15P1P4ANT2.2 places better than the model expected0.2 places better than the model expected
R5same car21P1DNF · UnknownANT2.9 places better than the model expected16.0 places worse than the model expected
R6same car16P1P12ANT2.0 places better than the model expected6.4 places worse than the model expected
R7same car31P16P2RUS11.9 places worse than the model expected1.4 places better than the model expected
R8same car41P3P1RUS1.5 places better than the model expected2.3 places better than the model expected
R9same car14P15P2RUS11.9 places worse than the model expected2.6 places better than the model expected
R10same car14P1DNF · UnknownANT2.0 places better than the model expected17.7 places worse than the model expected
R11same car47P3P7ANT2.4 places better than the model expected0.9 places worse than the model expected
R12same car32P2P3ANT2.4 places better than the model expected1.8 places better than the model expected
R13same car72P1P2ANT5.7 places better than the model expected1.7 places better than the model expected
R14same car26P1P5ANT2.6 places better than the model expected1.1 places better than the model expected
R15same car161P5P1RUS1.8 places better than the model expected2.2 places better than the model expected
R16same car48P2P20ANT2.1 places better than the model expected13.8 places worse than the model expected
R17same car46P3P9ANT1.5 places better than the model expected2.1 places worse than the model expected
R18same car16P1P8ANT2.8 places better than the model expected2.0 places worse than the model expected
R19same car52P3P6ANT2.0 places better than the model expected1.2 places worse than the model expected
R20same car32P5P4RUS1.4 places worse than the model expected1.6 places better than the model expected

Ahead: by finishing order (a retirement counts as behind). Andrea Kimi Antonelli · George Russell