Statistical Analysis of F1 Competition

[jwielage]

Sorry if I'm stating the obvious, but one solution to the reliability issue would be to exclude contests where a particular driver has a mechanical related DNF. Each driver's respective percentages would then be calculated based on the number of races less their mechanical DNFs.

Clearly one of the keys to making this approach work is differentiating between "driver induced" DNF's and "pure mechanical" DNF's. For example, if a driver commits and error which results in a mechanical problem on the car, then the real cause of the failure is the driver's error. I'm not talking about when a driver shears their suspension off after making contact with a wall, but lets say they have a minor impact which damages the ERS system resulting in a cascading series of problems culminating in a retirement. These retirements might be hard to tease out.

[Moxie]

The Williams drivers run so over the curb in turn 2 on every single lap and their suspension hold up for the entire event. One of the Tyrrell drivers takes the same line over the curb, but his suspension breaks. His teammate does not run over the curb, and finishes the race. Suspension reliability issue or driver issue?

Video and press may be analyzed to help provide information on the earlier races, we could probably account for cars taken out by the shunts of others. However, except for the obvious instances, the data just isn't there to determine if failures were necessarily driver induced. In more recent races I believe that such a determination can be made for most DNF's.

In any case, that is a lot of work. I may be a super nerd, but I have a 5yo child, and a wife that would kill me if I spent that much time on this hobby.

[Moxie]

Here is the chart for Team Win% for the top three teams, showing the standard deviations.
Team #1 Blue had a SD of 17
Team #2 Magenta had a SD of 13
Team #3 Green had a SD of 10

I am sorry for my lack of a title and legend, particularly my lack of references on the chart, but I had a hard time posting this chart without distortion. The original data came from the Official F1 site. http://www.formula1.com/results/team/1974/ I will revise this link with a chart that contains proper references when I figure it out.

https://drive.google.com/file/d/0BzuAgI ... sp=sharing

What I find so interesting about this chart is that the areas of overlap between the SD's Team #1 and Team #2 and at the same time the areas of overlap between the SD's Team #2 and Team #3 seem to come to an end at approx. 2014 or 2015. I am not sure what to think of this yet, but the timing is priceless, considering the problems F1 is having these days.

[Stradivarius]

Sorry if I'm stating the obvious, but one solution to the reliability issue would be to exclude contests where a particular driver has a mechanical related DNF. Each driver's respective percentages would then be calculated based on the number of races less their mechanical DNFs.

Clearly one of the keys to making this approach work is differentiating between "driver induced" DNF's and "pure mechanical" DNF's. For example, if a driver commits and error which results in a mechanical problem on the car, then the real cause of the failure is the driver's error. I'm not talking about when a driver shears their suspension off after making contact with a wall, but lets say they have a minor impact which damages the ERS system resulting in a cascading series of problems culminating in a retirement. These retirements might be hard to tease out.

Why do you want to distinguish between driver induced failures and pure mechanical failures? These statistics are in general not making that distinction, they are based on results which clearly depend on both car and driver performance both in terms of reliability and in terms of speed.

Considering how the drivers have become more professional with time, I expect that they generally make fewer mistakes than before, at least if we only consider the front runners. Of course, the cars and teams also help the drivers avoid mistakes to a greater extent than before, as the cars are easier to drive and so on, not to mention the tracks which now have run off areas instead of walls, gravel traps and grass.

It all depends on what you really want to study, but I think this information is very interesting without the distinction between driver error and mechanical failure as reson for DNF.

[Moxie]

Sorry if I'm stating the obvious, but one solution to the reliability issue would be to exclude contests where a particular driver has a mechanical related DNF. Each driver's respective percentages would then be calculated based on the number of races less their mechanical DNFs.

Clearly one of the keys to making this approach work is differentiating between "driver induced" DNF's and "pure mechanical" DNF's. For example, if a driver commits and error which results in a mechanical problem on the car, then the real cause of the failure is the driver's error. I'm not talking about when a driver shears their suspension off after making contact with a wall, but lets say they have a minor impact which damages the ERS system resulting in a cascading series of problems culminating in a retirement. These retirements might be hard to tease out.

Why do you want to distinguish between driver induced failures and pure mechanical failures? These statistics are in general not making that distinction, they are based on results which clearly depend on both car and driver performance both in terms of reliability and in terms of speed.

Considering how the drivers have become more professional with time, I expect that they generally make fewer mistakes than before, at least if we only consider the front runners. Of course, the cars and teams also help the drivers avoid mistakes to a greater extent than before, as the cars are easier to drive and so on, not to mention the tracks which now have run off areas instead of walls, gravel traps and grass.

It all depends on what you really want to study, but I think this information is very interesting without the distinction between driver error and mechanical failure as reson for DNF.

So to further understand what lead to this lopsided competition that is F1 today. It is not enough to simply assume that the the increased reliability allows the top cars to stay in the race, thereby allowing them to win more races. It is a good theory, but without data it is just that. As it turns out, the domination displayed by the 1st and 2nd place teams has been on a steady rise since 1974. Yet the second set of charts shows that the DNF rates remained fairly stable at around 37% until the mid 1990"s and the dramatic decline of DNF rate began. So while the theory is still a good one, it may not be the whole picture. Therefore it is logical to take a closer look at the DNF rates. Perhaps the mechanical reliability of the cars has not changed so much, but the number of driver induced DNF's has decreased. Perhaps it is the other way around. How does such data correlate with other data?