Formula One has an incentive problem called Qualifying

Sandy Rogers
5 min readSep 12, 2017


“There’s no point in old boys like me going round, saying ‘Oh, it was so much better in the old days’, because they don’t have the old days, they’ve got it as it is now.”
— Patrick Head, in ‘Lunches with Legends’ by Maurice Hamilton

There are lots of complaints about lots of areas in modern Formula One. But really, F1 is very simple. It should be the world’s best drivers having an all-out race on a Sunday. And here we have a problem: the racing doesn’t really happen on race-day Sundays, it happens in qualifying.

Not worth trying.

How F1 teams are incentivised

It’s expensive to run an F1 team: championship-competitive teams spend around €500m per season, and that money needs to be recouped by racing. (Back-of-the-grid teams have a budget about a fifth of that.) The sport as a whole makes getting on for €1bn each year and gives about 2/3 of that to the teams. The allocation is not entirely fair: Ferrari and Williams have long-standing sweetheart deals, and other teams reach their own agreements with FOM, meaning that Ferrari were awarded the biggest share of the pot this year despite coming third in the constructors’ championship last year.

But roughly, teams are incentivised to:

  • create themselves a heritage status in the sport,
  • score lots of championship points.

And here’s the problem: championship points aren’t won on Sundays, they’re won on Saturdays.

What the data show

I looked at data from every historical F1 race and championship. Points systems, qualifying setups, engine formulae, aero restrictions have all changed in that time. So, I calculated a simple metric:

for each race, what fraction of points awarded after the race were not “already won” by cars’ qualifying positions?

For example: if a car qualifies 4th in a 2017 weekend, they’ve “already won” 12 points on Saturday. If they finish the race in 3rd, they’ve “newly won” an extra 3 on race day. If that’s the only position change in the Top 10, that’s 3 out of 101 available points that are newly won on Sunday — 3%.

This race-day points fraction seems to me a good proxy for a team boss asking:

“how much incentive is there for me to make a good racing car this season, vs. a good qualifying car?”.

This metric has been decreasing for 35 years, through five different engine eras. There has never been less incentive for F1 teams to make an overtake-capable car.

There has never been less incentive for F1 teams to make an overtake-capable car.

How did this happen?

Well imagine you were running an F1 team in the mid 1980s. You were busy catching up with the revolution of turbo-chargers taking over from the dominant Ford DFV. The sport was allowing you to produce qualifying-spec engines with >1300 bhp. You probably looked at some statistics and thought:

“Hmm, 60% of the points people are winning are based on their grid position, and we can go nuts with power and ground effect traction and everything, so sod race day”.

You (and everybody else) would tune your cars increasingly to maximise grid positions, and therefore increasingly sacrifice race-day ability, and so Sundays would become ever less of a points-winning opportunity; and so the cycle would continue. If that’s how it happened, then we’re still in that cycle. And now in 2016 and 2017, 74% of the points allocated were determined by the qualifying grids.

Things aren’t all bad

The sport progresses in technology, and reliability and consistency are high in this era. In fact, in a way the competitiveness of the field is better than ever. I calculated the average race-time separation between cars through F1 history. In the modern era, cars usually finish races just a few seconds apart from one another.

The solid line is the race duration split for the whole field, the dashed line for the podium

But parading a few seconds apart is a very different spectacle to overtaking one another for points.

What to do

F1 has tried a lot of ideas to make racing better: different engine formulae, cost-restrictions, aero limits, single tyre manufacturers, DRS, the overtake button… But maybe we just need to look at the incentive structure, and flip the importance ratio of qualifying : race.

We need to make it more probable that a good race-day performance will turn into a good haul of championship points. That’s the basics. But the sport has never really attempted to achieve that by making it less probable that a good qualifying performance will lead to points.

Formula 2 has a go at this by having two points-paying races, with a grid reversal for the second race. The maximum points haul involves qualifying with the fastest lap, winning a race from pole with a fastest lap thrown in, and then overtaking your way from 8th on the grid to lead again in the sprint race, while setting another fastest lap.

British Touring Cars have a slightly-randomised grid reversal for the third race in each weekend, and a ballast allocation to handicap leaders.

These are of course very artificial systems, and I’m not sure F1 should or would adopt them.

There are (perhaps?) more palatable options though. For example we could have no qualifying, and each race grid would be set by championship order. The only incentive therefore is to do well in races. Empty Saturdays could be filled with:

  • points-paying fastest-lap sessions which don’t so directly impact the race grid.
  • a double-bill with Formula E races (once they can race on F1 circuits, or on street circuits in the nearby host city).
  • a rookie championship, where F1 cars are shared with a separately-scored rookie championship for drivers with less than 2 years’ F1 experience. Who knows?

Or we could just have randomly allocated grids, fresh each race.

However it is done, there is a simple enough message: the sport could choose to make (all) teams race for all of their points on Sundays. Presumably they’d soon figure out how to make the cars follow closely and overtake once their pay-cheques depend on it 100% (instead of 26%). The teams sit on the Sporting and Technical Working Groups, so it’s not like the rules would get in their way for very long either.

I used data from this excellent resource

The analysis code is available on Github,, to do with as you wish.



Sandy Rogers

Reformed astrophysics researcher, recovering marathon runner, and recalcitrant data wrangler @SaberrUK.