Tag Archives: statistics

Do the best drivers get the best F1 cars?

Which team has the best driver line-up? Although different fans will come up with different answers, it would take a bold person to pick a team other than Ferrari, Mercedes or Red Bull. These also happen to be the top 3 teams in F1 right now. The-race.com also deemed these 3 teams to have the best 2022 driver line-up, and their results showed a strong correlation between car performance and perceived driver line-up strength.

Of course there could be various different explanations for this, including both coincidence and a bias towards better results that were fundamentally caused by the car. Here, I’ll look at the recent history of Formula One, to see if the data supports the hypothesis that better drivers will indeed end up in better F1 cars.

This was a difficult analysis to produce, because the the data can be interpreted in different ways and the conclusions are somewhat nuanced. The short version is that there is indeed a correlation between good drivers and good teams, but it’s not a very strong correlation.

  1. Comparing Driver and Team Rankings
  2. Which team is the best driver likely to be in?
  3. Does the best team have strong drivers?
  4. Midfield and back of the grid team
  5. Conclusions

Comparing Driver and Team Rankings

Using the mathematical model, I have compared the driver rankings with the rankings of the car performance. Overall there’s a clear trend for these years for better drivers getting into better teams. I’ve shown below the year that shows this trend most clearly over the past 10 years (2016). Although the variables used here are not the best for doing, they do show the results in a way that should be relatively simple to interpret.

Positive correlation between driver and team rankings during 2016 F1 season.

There’s a clear positive correlation here. There are no examples of poor drivers in great cars (bottom right) and almost no examples (depending on where you draw the line) of great drivers in poor cars (top left).

Even so, there is considerable variation, with strong performances from Alonso and Sainz in middling cars (middle left of graph). That this year is the one that shows the positive correlation most obviously gives some context for how weak the correlation generally is. For this data, I also chose the car a driver was in the most. Picking the car a driver started the season in (i.e. before the Verstappen/Kvyat swap) would muddy the waters further.

Out of the last 10 years, 2015 was the year with the least correlation between driver and team performances:

A lack of correlation between driver and team rankings during 2015 F1 season.

What’s the trend here? Overall, there isn’t one. Looking at the top 6 ranked drivers, 3 of them were in good cars (Hamilton, Vettel and Rosberg, bottom left) and 3 were in below average cars (Verstappen, Alonso and Button, top left).

For those wondering, there wasn’t enough comparative data to get a representative ranking on the Marussia drivers. Logically they would be somewhere towards the back of the driver rankings (with all due respect, does anyone really remember Stevens and Mehri?), which would help to restore a slight positive trend.

A final consideration is whether the model is significantly inherently biased towards one type of result here. If a driver were overrated by the model, their individual performance would be inflated, which would mean that their car would be underrated (as the model is trying to explain the results in terms of both car and driver performance). A driver underrated by the model would be assigned an exaggerated car performance. This means that the correlation between driver and team performance is possible slightly stronger than shown here.

Which team is the best driver likely to be in?

As you can see from my history of F1 article, the model believes that one driver often dominates an era in terms of performance. Here I’ve looked at data the modern era (post 1994) to see how likely it is the best driver of a given year is in a competitive car. The results are as follows.

Graph showing the top driver is likely to get a competitive car.

The top ranked driver has been in a car in the top half of the grid over 90% of the time, and in a top 3 team (which roughly equates to a team capable of winning races) almost 80% of the time. It’s therefore extremely likely that a generational talent will get a competitive car.

Surprisingly though, the top ranked driver was only deemed to have the absolute best car 3 times in 29 years, which is barely better than random chance!

One thing to remember is that although 29 years is a long time, the period only covers 3 different drivers that the model considers to be the best of their generation (Schumacher, Alonso and Verstappen), and Schumacher and Verstappen both spent almost all of their career peak in just one team.

Does the best team have strong drivers?

So top-line drivers are likely to have good (but not necessarily the absolute best) cars. Does that mean that the best car is likely to have good drivers too?

Below is a graph of the likelihood of a given driver in the best performance team being ranked in a given position. In slightly over half of all cases a driver for the best team will be ranked within a top 6. This is another piece of evidence that suggests the best team tends to hire high quality drivers.

Graph showing the top team is likely to have strong drivers.

However, the data becomes more nuanced when split between the lead and secondary driver in the top team (in this instance, “lead” means the best performing driver for that given year).

Graph showing a lead driver for a team is likely to be very strong, whereas the second driver is likely to be average or even below average.

So the secondary driver seems to show the opposite trend, with more instances of them being below average than above. Does that mean that top teams intentionally hire below average drivers? I believe that the solution is actually more nuanced than this. One important factor is the statistical quirk from the way the data is produced. Remember, the graph is considering the worst driver in the team. If you considered the worst driver in every team, you’d of course find that they are, as a whole, below average.

You’d actually expect a secondary driver to be occupying the bottom 4 rankings more than would happen by random chance, but in the graph above the opposite is true. Surprisingly then, the data still indicates that the secondary driver in the best team is actually still slightly better than expected by random chance, even if their ranking is often disappointing for a driver in a competitive car.

It appears as though the best team does not regularly attempt to fill the car with the 2 best drivers. The reasons for this are varied, including the considerations of top drivers (who may not want another high profile name as a teammate) and teams (who could also do without the hassle of their star drivers falling out).

Whilst there’s almost 30 years of data used here, it may also feel like just a few teammate match ups are skewing the data. After all, Hamilton/Bottas and Vettel/Webber account for nearly a third of the total years, and the model thinks of both Bottas and Webber as rather average F1 drivers. Bottas’ highest Mercedes ranking was 10th, for example, whilst Webber’s highest for Red Bull was 9th. There’s a reason those partnerships lasted so long though, and I’d argue that the gap in performance between the lead and “number 2” driver is by design. Almost all long running teammate combinations share a common theme: They’re with a top team and one driver has a noticeable edge (think Shumacher-Barrichello, Alonso-Massa, Vettel-Räikkönen and Häkkinen-Coulthard)

Meanwhile, “better” driver pairings at top teams such as Alonso-Hamilton, Senna-Prost and Hamilton-Rosberg are inherently more divisive and short lived. There are actually no examples since 1994 of the highest ranked team using 2 drivers ranked within the top 5 for that year, which would be pretty unlikely if teams were only focused on obtaining the best drivers possible.

Midfield and back of the grid team

We can also see how this data compares to other teams. Let’s start by comparing the lead drivers for the top and 5th best team:

Graph for comparing the lead driver for the 5th best team to the best top. The trend is similar, but the 5th best team generally has a worse lead driver.

You can see a clear difference. Whilst the 5th best team is will probably have at least one strong driver, the likelihood of having an absolute top tier driver is still quite low, and the likelihood of having two below average drivers is much higher than for the absolute best team. (You can see there’s a decent chance of having the top driver being below average, which would obviously mean that the second driver was too.)

Comparing secondary drivers is more nuanced:

There’s still an edge to the lead team (a 38% chance of their secondary driver being in the top 10 compared to just 22% for the fifth best team), but there are different ways to interpret the data. The indication is that a secondary driver for a top team is, on balance, typically weaker than a typical driver in a midfield team (given that this includes both the top and secondary drivers for the midfield team).

Finally, let’s have a quick look at how driver’s in the 10th best team (typically the back of the grid) have been ranked.

You can see there’s a further inching of the data to the right, with not a single ranking within the top 6 and plenty of poor rankings. Still though, the association between a poor car and below average drivers is quite weak.


  1. In general, there’s a positive correlation between team and driver performance, although a relatively weak one.
  2. Although the model rarely aligns the top driver with the top team, the lead driver for the top team is typically well above average.
  3. However, the secondary driver for the top team is, on average, mediocre.