Science proves that Fangio was the greatest F1 driver of all time

[Olivier]

Oh no! Another thread about who's the best F1 driver ever? ... I'm afraid so. This time it's supposedly backed by science. I haven't looked at the paper, but this is their top-20 list:

1. Fangio
2. Prost
3. Clark
4. Senna
5. Alonso
6. Piquet
7. Stewart
8. Schumacher (Michael)
9. Fittipaldi
10. Vettel
11. Fitipaldi (Christian)
12. Hamilton
13. Rosberg
14. Hill (Graham)
15. Dan Gurney
16. Scheckter
17. Button
18. Surer
19. Hill (Damon)
20. Louis Rosier

As a bonus, here's their top-10 list of teams:

1. Ferrari
2. McLaren
3. Mercedes
4. Red Bull
5. Benetton
6. Lotus (2010s)
7. Williams
8. Daimler Benz AG
9. Brawn
10. Mattra Intl

Here's the link:
Science proves that Fangio was the greatest F1 driver of all time

Here's the paper:
Formula for success: Multilevel modelling of Formula One Driver and Constructor performance, 1950-2014

 

[Titch]

Until science proves the model wrong, then it will be someone else.

 

[Incubus]

Can I enrol? It's gotta be a nice life spending all that time debating who are the greatest drivers in history and getting funded for it...

 

[TR]

Interesting sidenote to the above ranking: (from the paper)

Michael Schumacher, who holds the record for the most championships and race victories of any driver in Formula 1, comes in a relatively modest eighth place. This is in part because those victories were won in an excellent car, but also because his ranking is dragged down by his more recent post-retirement performances (2010-2012) when he performed less well than in the main part of his career and crucially was generally outperformed by his Mercedes teammate Nico Rosberg. Thus, we re-ran the analysis with the latter section of Schumacher’s career treated as a separate driver. In this formulation, pre-2006 Schumacher’s ranking rises to 3rd and Nico Rosberg’s ranking falls from 13th to 49th. This is because Schumacher’s high standing as a driver in the model effectively deflated 2010-2012 Mercedes’ team ranking in the first model, meaning Rosberg’s performances appeared more impressive. When treated as separate drivers, post-retirement Schumacher performed less well, the Mercedes team effect appears greater, and so Rosberg’s performances no longer stand out compared to his team.

 

[Greenlantern101]

Pointless comparison, the difference in the cars and drivers between Fangio's era and todays era are just too great. Fangio also swapped teams mid season to get in the best car. Just not possible in this day and age of contracts.

The only way to know for sure would be to invent a time machine and take Hamilton back to 1950 and Fangio to 2016 and let them have a go.

 

[TR]

Pointless comparison, the difference in the cars and drivers between Fangio's era and todays era are just too great. Fangio also swapped teams mid season to get in the best car. Just not possible in this day and age of contracts.

The only way to know for sure would be to invent a time machine and take Hamilton back to 1950 and Fangio to 2016 and let them have a go.

I assume that the point of the paper is not actually who is the greatest driver, but showing off their statistical method for comparing "performance" measured under different conditions.

 

[Legs]

I'm not sure whether Marc Surer or Christian Fittipaldi is the bigger head-scratcher on that list!

 

[teabagyokel]

Lewis Hamilton a worse driver than Christian Fittipaldi, proved by science. Shit!

 

[The Pits]

well, science IS the only way to prove stuff.

 

[cider_and_toast]

I Don't understand how they calculated that and don't get me wrong, Jim Clark was an outstanding driver but the cars he drove in 63 and 65 were also the class of the field.

 

[Il_leone]

Does anyone think Christian Fittipaldi was a typo error

 

[Greenlantern101]

Well I have skimmed through their school boy paper and frankly its crap. The whole thing is full of assumptions and view points and constructed equations. A completely different result could be had from tweeking variables and making alternative assumptions. Science my arse.

 

[Galahad]

Fangio at no.1 and Ascari not in the top 20 is baffling. Although nice to see some credit for one of my favourite underrated drivers Marc Surer.

I'll have to look at the paper to understand it properly. The issue I have is that drivers' performances vary from year to year, and across their career, even when other variables such as team, team mate, engine and tyre are held constant. So there's an awful lot of noise in the model even accounting for the big stuff.

 

[The Artist.....]

I work in this area, (I know one of the authors) - however, I'm always dubious by the underlying assumptions of multi-level modelling (which the paper relies upon). In terms of nested models, it's probably the only game in town, but the underlying assumptions that are needed outweigh the actual benefit of the models...

Edit

Having had a closer look at the paper, it has become apparent that the ranking exercise is more or less pointless; since these are statistical estimates, they are estimated with a certain level of uncertainty (and hence all have confidence intervals around them). The implications of this is that based on the results, you cannot, for instance, claim that Fangio was any better than Rosier.

This same problem also existed for contextualised value added in School League tables - it gave a reasonable measure of school performance, but because there was a measure of uncertainty associated with it, using it as a ranking exercise was pointless. See, for instance, http://www.bristol.ac.uk/media-library/sites/cmpo/migrated/documents/wp176.pdf

 

[Titch]

So new model and new greatest driver. How many pints per model?
Who's funding this research anyway.

 

[Spinodontosaurus]

It appears as though this model is treating all DNFs as low finishing positions regardless of cause, which is probably the main source of some of the crazier results.

 

[gethinceri]

I have spent less money than the authors and my findings are as valid as theirs (based upon the horse and cart that I can drive through their "science"), I have utilised the equally valid method of pulling the results from the depths of my own arsehole and established that Derek Warwick is the greatest F1 driver of all time, ever, FACT.

 

[The Artist.....]

gethinceri - oddly, Derek Warwick will be one of the underrated drivers in this model- thanks to his year at Renault. The model tries to quantify team performance (some of which is treated as persistent year-on-year). However, Warwick's Renault was a pale imitation of Prost's the previous season. Their model tries to account for some of this sort of variation, but obviously fails.

There are other, clear, flaws in their model- such as that it suggested that in 1992, Nigel Mansell should have only finished 3rd in the championship, with Patrese second, and Senna first; a model that makes out that Patrese was better than Mansell was fundamentally flawed!