Knowing which car a driver joins predicts his season better than knowing who he is
F1 Legend Sim · Staff · 22 September 2026 · 8 min read
A transfer is the closest thing Formula 1 offers to an experiment: the driver stays the same and the car changes. We took the 94 team changes since 1988 and pitted two predictions against each other for that first season in the new place. One says the driver carries his level with him. The other says the car he joins decides. The car wins, and not narrowly.
Two predictions, one result
For each transfer we tried to guess the driver's average finishing position in his first new season, two ways: from his own average the year before, and from the average that car was doing the year before, when he was not in it yet. Whichever misses by less, wins.
| What we predict from | Average error | Cases it wins |
|---|---|---|
| His own level the year before | 2.81 places | 42 of 94 |
| The car he joins | 2.15 places | 52 of 94 |
Knowing which car a driver is getting into tells you more about how his year will end than knowing how the last one ended. The gap is 0.66 places of error, and the car wins 55.3% of the head-to-heads.
It is worth saying what the number does not say, because 55.3% is not 90%. The car wins, but it leaves room: in 44.7% of transfers the driver beats what that car was doing before he arrived. What the table kills is the idea that a driver performs roughly the same whatever he climbs into.
The twelve biggest jumps
Here it is in the flesh. The last column is what the new car had been doing the year before, the prediction that knew nothing about the driver.
| Year | Driver | Move | Before | After | Car said |
|---|---|---|---|---|---|
| 2017 | Esteban Ocon | Manor Marussia to Force India | P17.00 | P7.95 | P8.79 |
| 2011 | Vitantonio Liuzzi | Force India to HRT | P11.23 | P19.75 | P17.72 |
| 2014 | Daniel Ricciardo | Toro Rosso to Red Bull | P12.19 | P3.75 | P2.97 |
| 2022 | George Russell | Williams to Mercedes | P12.59 | P4.38 | P3.79 |
| 2012 | Daniel Ricciardo | HRT to Toro Rosso | P19.25 | P12.05 | P11.53 |
| 1992 | Martin Brundle | Brabham to Benetton | P10.44 | P3.64 | P5.90 |
| 2015 | Fernando Alonso | Ferrari to McLaren | P5.41 | P12.18 | P8.44 |
| 1991 | Thierry Boutsen | Williams to Ligier | P4.00 | P10.60 | P11.36 |
| 2025 | Carlos Sainz | Ferrari to Williams | P4.71 | P10.75 | P13.80 |
| 2019 | Charles Leclerc | Sauber to Ferrari | P10.13 | P4.21 | P3.19 |
| 2017 | Valtteri Bottas | Williams to Mercedes | P8.53 | P2.95 | P2.18 |
| 2022 | Valtteri Bottas | Mercedes to Alfa Romeo | P5.00 | P10.41 | P12.80 |
Read the last two columns together. In almost every row, where the driver lands looks far more like the car's number than like his own from the year before, and it makes no difference whether the jump is up or down. Esteban Ocon went from P17.00 to P7.95 on joining Force India, a car coming off P8.79. Vitantonio Liuzzi made the opposite trip, P11.23 to P19.75 at HRT, a car coming off P17.72.
The same driver can appear twice in this table in different years, going up and coming down, with nothing about him having changed. That is the cleanest demonstration of what all this measures.
What this does not mean
It does not mean the driver is irrelevant. It means the position you finish in, which is the only thing standings measure, is dominated by the machinery. Those are two different things and they get confused constantly: an excellent driver in a midfield car shows up in the statistics as a midfield driver, and there is no way to tell from the outside by looking at results.
The only way to separate them is to take the car out of the equation, which is exactly what reality cannot do and a simulation can. We have measured it from that side in how much the car weighs against the driver and in the 22 drivers of 2026 in all 11 cars. There is no simulator here: these are real results, and they reach the same place by a different road.
If you want to run the experiment yourself, What If? lets you put any driver in any car and run the season. And the duel against the one rival who shares the machinery, the team-mate, is counted with real data in the team-mate duels.