Loading drivers…
Loading drivers…
Three different measures, kept apart. Vs teammate compares two drivers in the same car, the cleanest comparison with the car taken out. The driver rating is the estimated pace vs the mean driver (model): learned mostly from teammate gaps, lower is faster, with its ± sd. Gain vs model expectation is how a driver finished against what the model expected before the race (which already includes the driver rating): a surprise, not a measure against the car.
Power rankings shows total pace (car and driver together) and the car's pace by round; this page separates the driver. · Compare two drivers →
Round 18 · Singapore Grand Prix · 19 eligible drivers.
Reading the ranks. Rows are ordered by the race rating, fastest first. Each driver's rank varies across the 10,000 draws, so the median rank can repeat or skip a number; the 80% range shows the spread.
Rating vs results. The rank is lap-time pace, separate from the car, and moves slowly; the gain is how the driver's finishes compared with the model's expectation, which already includes the rating. So a mid-table pace rank can sit beside a large gain: places won through starts, strategy and tyres, or a rating still catching up. The gain column covers all races R1–R18, retirements included; this includes round 18's race, which came after the rating snapshot the rank is taken from.
| Median rank (of 10,000 draws) | Driver | 80% rank range | Rating ± sd, − = fasterRace rating: estimated pace vs the mean driver (model); lower = faster. | Gain vs model R1–R18, + = better | P(top 1) | P(top 3) | P(top 5) | P(ahead of next)Chance this driver's rating is faster than the next driver listed. |
|---|---|---|---|---|---|---|---|---|
| 4 | VERM. Verstappen | 1–13 | −0.288% ± 0.250 | 0.02 places better than the model expected | 27.7% | 47.7% | 60.8% | 53.1% vs LEC |
| 4 | LECC. Leclerc | 1–12 | −0.269% ± 0.188 | 0.72 places better than the model expected | 17.4% | 43.1% | 60.2% | 51.5% vs GAS |
| 5 | GASP. Gasly | 1–10 | −0.263% ± 0.150 | 0.56 places worse than the model expected | 11.0% | 37.8% | 60.7% | 51.2% vs NOR |
| 5 | NORL. Norris | 1–11 | −0.255% ± 0.154 | 0.33 places better than the model expected | 12.4% | 37.3% | 57.7% | 51.5% vs ALO |
| 5 | ALOF. Alonso | 2–11 | −0.244% ± 0.158 | 0.28 places worse than the model expected | 9.8% | 34.0% | 54.9% | 66.8% vs RIC |
| 8 | RICD. Ricciardo | 3–13 | −0.161% ± 0.137 | 0.59 places worse than the model expected | 2.1% | 12.6% | 29.4% | 51.2% vs SAI |
| 8 | SAIC. Sainz | 3–15 | −0.151% ± 0.188 | 0.78 places better than the model expected | 1.7% | 17.2% | 32.2% | 51.3% vs BOT |
| 8 | BOTV. Bottas | 1–16 | −0.134% ± 0.250 | 0.36 places worse than the model expected | 10.5% | 25.1% | 37.0% | 58.1% vs OCO |
| 10 | OCOE. Ocon | 5–16 | −0.084% ± 0.150 | 0.70 places better than the model expected | 0.2% | 3.8% | 12.4% | 49.6% vs ALB |
| 10 | ALBA. Albon | 4–16 | −0.080% ± 0.142 | 0.78 places worse than the model expected | 1.3% | 6.7% | 16.5% | 53.4% vs PIA |
| 11 | PIAO. Piastri | 5–16 | −0.065% ± 0.154 | 1.46 places better than the model expected | 0.2% | 4.2% | 12.2% | 50.8% vs STR |
| 11 | STRL. Stroll | 5–16 | −0.059% ± 0.159 | 1.06 places worse than the model expected | 0.1% | 3.4% | 11.1% | 53.3% vs TSU |
| 12 | TSUY. Tsunoda | 6–16 | −0.043% ± 0.137 | 0.92 places worse than the model expected | 0.1% | 1.3% | 6.1% | 52.9% vs RUS |
| 12 | RUSG. Russell | 3–18 | −0.020% ± 0.250 | 1.11 places worse than the model expected | 3.5% | 11.8% | 19.9% | 63.0% vs HAM |
| 13 | HAML. Hamilton | 4–18 | +0.014% ± 0.250 | 0.75 places better than the model expected | 1.8% | 8.3% | 15.3% | 57.7% vs HUL |
| 15 | HULN. Hülkenberg | 9–18 | +0.068% ± 0.151 | 1.10 places better than the model expected | <0.1% | 0.7% | 2.5% | 60.9% vs ZHO |
| 16 | ZHOG. Zhou | 8–19 | +0.151% ± 0.250 | 0.39 places better than the model expected | 0% | 2.4% | 5.4% | 50.4% vs PER |
| 16 | PERS. Pérez | 8–19 | +0.158% ± 0.250 | 1.20 places worse than the model expected | 0% | 2.6% | 5.8% | 59.9% vs MAG |
| 18 | MAGK. Magnussen | 14–19 | +0.228% ± 0.152 | 0.60 places better than the model expected | 0% | <0.1% | <0.1% | — |
Range bar: the track has one slot per eligible driver (rank 1 at the left); the coloured bar spans the 80% rank range and the dark tick marks the median. A P(ahead of next) near 50% means the model cannot tell those two drivers apart; a gap within about 1 sd is not a clear difference. Eligible: rated in the round's snapshot and at least 5 grand prix starts in our data (2022 on) before the round.
† Fewer than 5 races in the gain window: too noisy to read.
Measure: race rating only. 10,000 draws per round, seed 20261117; test run 20261116-234250-driver-ranking-test.