- 27 Apr 09, 21:38#112795
Multi linear regression model utilising historical performances from drivers and cars to estimate performances of drivers standardised for car performance. Of course only as good as the evidence provided and sometimes drivers can improve dramatically compared with their historical data.
This model seeks to provide an estimate of the number of points a driver would have scored at their peak had all cars been identical, had 100% relaibilty and the same level of pit support based over 25 races and where all drivers faced the same level of competition. i.e. the average standard achieved since 1950. Scores are based on 9 - 6 - 4 - 3 - 2 - 1.
I know it's not perfect but its a basic stab at providing a value added statistical assesment of performance excluding human opinion.
I guess I just like playing with numbers and thought someone might be interested.
Tim
Personally I would have to go for the following order using a statistical model I produced to evaluate GP Drivers.
1 Jackie Stewart 183pts
2 Jim Clark 159pts
3 Damon Hill 153pts
4 Graham Hill 141pts
5 Nigel Mansell 140pts
6. Stirling Moss 132pts
7= Lewis Hamilton 117pts
7= John Surtees 117pts
9= David Coulthard 110pts
9= Mike Hawthorn 110pts
I fully realise that Stirling Moss is rated lower on this list than in the eyes of many experts as Damon Hill is rated higher. Of course Lewis' peak rating may eventually become higher but at the moment he is 7 points below his 117 peak at 110 pts.
Only Michael Schumacher and Juan Manuel Fangio have exceeded 200 rating points based on this methodology. Ayrton Senna came close with 197pts.
Tim
Uh, and what IS your statistical model?
Multi linear regression model utilising historical performances from drivers and cars to estimate performances of drivers standardised for car performance. Of course only as good as the evidence provided and sometimes drivers can improve dramatically compared with their historical data.
This model seeks to provide an estimate of the number of points a driver would have scored at their peak had all cars been identical, had 100% relaibilty and the same level of pit support based over 25 races and where all drivers faced the same level of competition. i.e. the average standard achieved since 1950. Scores are based on 9 - 6 - 4 - 3 - 2 - 1.
I know it's not perfect but its a basic stab at providing a value added statistical assesment of performance excluding human opinion.
I guess I just like playing with numbers and thought someone might be interested.
Tim