Greatest Driver of all time statistical analysis
Discussion
Adam Ansel said:
hairyben said:
As for hamilton, he's one of the best examples for being a multiple WDC whose been in as close as you'll get to equal terms/machinery with all his team mates, as a rookie he beat one of the most prolific drivers in his prime and then 3 years continuously beat a very quick and accomplished driver in rosberg also in his prime, to rank them both way over him makes a mockery of the instrument. hypothesise what you like about language and perceived fanboy allegiances I don't see the relevance, I'd certainly say this paragraph is more representative of facts than that table...
Hamilton was outscored by Button at McLaren.Adam Ansel said:
hairyben said:
As for hamilton, he's one of the best examples for being a multiple WDC whose been in as close as you'll get to equal terms/machinery with all his team mates, as a rookie he beat one of the most prolific drivers in his prime and then 3 years continuously beat a very quick and accomplished driver in rosberg also in his prime, to rank them both way over him makes a mockery of the instrument. hypothesise what you like about language and perceived fanboy allegiances I don't see the relevance, I'd certainly say this paragraph is more representative of facts than that table...
Hamilton was outscored by Button at McLaren.For example does it take into account that LH was ahead of JB at the 2012 Singapore and Abu Dhabi GPs? Or how about the 2010 Japanese GP when he was ahead of Button before losing that advantage after succumbing gearbox problems?
While the model looks at performance relative to team mates and tries to negate team effect/politics. I think a far more important measure of a drivers skill is how good they are in the wet, after all we are often told wet conditions are a great leveller regarding car performance, and also those drivers with great car control tend to excel.
Therefore there should be some formula applied to a drivers results achieved in wet conditions. Looking at the top 10's for example, Prost tends to crop up a lot, but he was a dire wet weather driver, no doubt ever since he was rear ended by Pironi in 82 at Hockenheim, an accident which ended Pironi's F1 career. On the other hand you have Gilles Villeneuve, as soon as it rained you knew he would get the car into positions it should never have been in (Ferrari T5, Monaco 1980 springs to mind). In fact if every race in 1979 had been wet I have no doubt in my mind he would have been champion that year.
A great driver should be quick in all conditions and capable of dragging whatever car beyond its mechanical capability.
There are so many other variables that could be considered such as tactical skill that it makes the use of a mathematical model difficult to use on its own.
Therefore there should be some formula applied to a drivers results achieved in wet conditions. Looking at the top 10's for example, Prost tends to crop up a lot, but he was a dire wet weather driver, no doubt ever since he was rear ended by Pironi in 82 at Hockenheim, an accident which ended Pironi's F1 career. On the other hand you have Gilles Villeneuve, as soon as it rained you knew he would get the car into positions it should never have been in (Ferrari T5, Monaco 1980 springs to mind). In fact if every race in 1979 had been wet I have no doubt in my mind he would have been champion that year.
A great driver should be quick in all conditions and capable of dragging whatever car beyond its mechanical capability.
There are so many other variables that could be considered such as tactical skill that it makes the use of a mathematical model difficult to use on its own.
hairyben said:
You can't apply a logical or mathematical process to reinterpret statistics this way without accounting for any inaccuracy (bias, team orders) first though, not afterwards, as the innacuracy could be amplified by whatever calculation takes place, exponentially flawing the results.
As for hamilton, he's one of the best examples for being a multiple WDC whose been in as close as you'll get to equal terms/machinery with all his team mates, as a rookie he beat one of the most prolific drivers in his prime and then 3 years continuously beat a very quick and accomplished driver in rosberg also in his prime, to rank them both way over him makes a mockery of the instrument. hypothesise what you like about language and perceived fanboy allegiances I don't see the relevance, I'd certainly say this paragraph is more representative of facts than that table...
I really don't think you understand what the model does, to be honest. The poster above you (glazbagun) gets it. I'm no mathematician, so I'm out of my depth in actually critiquing the maths but, the principles are sound. Moreover, the methodology has been published in a peer-reviewed journal, so I assume the relevant experts don't have a problem with the process employed to arrive at the results. I wish I had access to the paper where something interesting also caught my eye:As for hamilton, he's one of the best examples for being a multiple WDC whose been in as close as you'll get to equal terms/machinery with all his team mates, as a rookie he beat one of the most prolific drivers in his prime and then 3 years continuously beat a very quick and accomplished driver in rosberg also in his prime, to rank them both way over him makes a mockery of the instrument. hypothesise what you like about language and perceived fanboy allegiances I don't see the relevance, I'd certainly say this paragraph is more representative of facts than that table...
Dr. Phillips said:
Subjective ratings of the best drivers in the history of Formula One are common, but objective analyses are hampered by the difficulties involved in comparing drivers who raced for different teams and in different eras. Here, we present a new method for comparing performances within and between eras. Using a statistical model, we estimate driver and team contributions to performance, as well as the effects of competition with other drivers. By adjusting for team and competition effects, underlying driver performances are revealed. Using this method, we compute adjusted scoring rates for 1950–2013. Driver performances are then compared using: (i) peak performances for 1-year, 3-year, and 5-year intervals; and (ii) number of championships. Overall, these comparisons rank Clark, Stewart, Fangio, Alonso, and Schumacher as the five greatest drivers. We confirm the model’s accuracy by comparing its performance predictions to 2010–2013 lap-time data. The results of the analysis are generally in good agreement with expert opinions regarding driver performances. However, the model also identifies several undervalued and overvalued driver performances, which are discussed. This is the first objective method for comparing Formula One drivers that has yielded sensible results. The model adds a valuable perspective to previous subjective analyses.
(my emphasis)Paper: Uncovering Formula One driver performances from 1950 to 2013 by adjusting for team and competition effects
So, it has also been validated, in which case I believe the GIGO charge is a bit unjustified. Let's just agree to disagree on this.
Re: Hamilton. In the only season he competed against Alonso as team mate, he only equalled him on points and wins in one of the most messy F1 seasons in history in terms of team harmony. To then conclude that he dominated Alonso in that season...it is much shakier ground than that list of the greatest drivers. IMHO. Before you say, he had more poles. Yes, he's fast but is that enough to make one great? I think not.
entropy said:
You can't base on your arguments just on points scored. It's a very useful guide but does have its flaws and not the be all and end all - analysis can run deeper than that.
For example does it take into account that LH was ahead of JB at the 2012 Singapore and Abu Dhabi GPs? Or how about the 2010 Japanese GP when he was ahead of Button before losing that advantage after succumbing gearbox problems?
Points are everything. They decide both championships. Qualifying and fastest laps don't count. And there are no points for style.For example does it take into account that LH was ahead of JB at the 2012 Singapore and Abu Dhabi GPs? Or how about the 2010 Japanese GP when he was ahead of Button before losing that advantage after succumbing gearbox problems?
I think that Hamilton learned a huge amount from Button about how to run a race, he could see that he was outscored so he worked to fix the reasons why.
I really don't think there is a definitive answer to who is the greatest F1 driver.
You simply cannot compare eras and machinery. I think you could probably narrow it down to who the best driver of each decade was....maybe?
50s - Fangio
60s - Clark
70s - Lauda
80s - Prost/Senna
90s - Schumacher
00s - Schumacher/Alonso
10s - Hamilton/Vettel
Every F1 champ in history has been flawed in one way or another. They are all magnificent drivers of which I have immense respect for.
You simply cannot compare eras and machinery. I think you could probably narrow it down to who the best driver of each decade was....maybe?
50s - Fangio
60s - Clark
70s - Lauda
80s - Prost/Senna
90s - Schumacher
00s - Schumacher/Alonso
10s - Hamilton/Vettel
Every F1 champ in history has been flawed in one way or another. They are all magnificent drivers of which I have immense respect for.
Adam Ansel said:
Points are everything. They decide both championships. Qualifying and fastest laps don't count. And there are no points for style.
I think that Hamilton learned a huge amount from Button about how to run a race, he could see that he was outscored so he worked to fix the reasons why.
Points accumulated doesn't give the big picture or tell a fuller narrative.I think that Hamilton learned a huge amount from Button about how to run a race, he could see that he was outscored so he worked to fix the reasons why.
LH v JB is a good case example. In 2012 JB was struggling more with getting the tyres in the right operating window whereas LH never did; the former was supposedly going to be better with Pirelli's crap tyres because his driving style would be his advantage whereas the latter was often criticised for being too hard on his equipment but in never turned out that way in that season. In Spain LH came from the back of the grid and beat JB, in Canada LH won and JB was destroyed because he struggled with his tyres. More analysis than looking on paper at points accumlated reveals a greater depth to the narrative as a whole.
How about what I personally call [i]anomaly seasons[i/] eg. LH in 2011 saying he wasn't in the right "headspace" or Vettel struggling and/or trying to enforce a get out clause in 2014. How would one look at those years when analysing a whole career?
IMHO what can be analysed subjectively has equal weight to what can be quantified.
Another case in point is the 1979 season. Jody Schekter was WDC that year but at the Italian GP Gilles Villeneuve was riding shotgun. If Gilles was ruthless and selfish enough, had he disobeyed team orders it would have been he who would have been WDC so do that make him any less of a driver? It's a subjective matter because one has their opinions.
And regards to Button's influence on LH's driving? It's hard to say. You could argue the Pirelli era has influenced it? maturing his driving as he gets older? Lauda's influence - one of the very people in the paddock he genuinely listens to?
glazbagun said:
I'm actually a little sad I wasn't born in the 60's. I could have grown up with Clark in the background, witnessed JYS & Hunt, all the while listening to the Beatles, Hendrix and Led Zeppelin making music like I'd never heard before.
Same here. Although you can have The Beatles all to yourself, I'm happy listening to Led Zep 24/7 
Dr Z said:
I really don't think you understand what the model does, to be honest. The poster above you (glazbagun) gets it. I'm no mathematician, so I'm out of my depth in actually critiquing the maths but, the principles are sound. Moreover, the methodology has been published in a peer-reviewed journal, so I assume the relevant experts don't have a problem with the process employed to arrive at the results. I wish I had access to the paper where something interesting also caught my eye:
I appreciate what the model is aiming to do, I just place far more emphasis on how flawed the input data is. And if it wasn't flawed enough already, he then "flawifys" it some more by converting modern scoring to the old 10-6-4 system; any tr00 fan will have no trouble explaining the problem here. 
wink smiley included to indicate my use of tr00 was a light-hearted way of making a serious point and shouldn't detract from it.Re interpreting F1 results using the same data is like deciding to knock a house down and rebuild it on the basis the wood it's made from is all rotten, then rebuilding it in a different style using the same wood again and declaring "look, we fixed it."
Dr Z said:
Dr. Phillips said:
Subjective ratings of the best drivers in the history of Formula One are common, but objective analyses are hampered by the difficulties involved in comparing drivers who raced for different teams and in different eras. Here, we present a new method for comparing performances within and between eras. Using a statistical model, we estimate driver and team contributions to performance, as well as the effects of competition with other drivers. By adjusting for team and competition effects, underlying driver performances are revealed. Using this method, we compute adjusted scoring rates for 1950–2013. Driver performances are then compared using: (i) peak performances for 1-year, 3-year, and 5-year intervals; and (ii) number of championships. Overall, these comparisons rank Clark, Stewart, Fangio, Alonso, and Schumacher as the five greatest drivers. We confirm the model’s accuracy by comparing its performance predictions to 2010–2013 lap-time data. The results of the analysis are generally in good agreement with expert opinions regarding driver performances. However, the model also identifies several undervalued and overvalued driver performances, which are discussed. This is the first objective method for comparing Formula One drivers that has yielded sensible results. The model adds a valuable perspective to previous subjective analyses.
(my emphasis)Paper: Uncovering Formula One driver performances from 1950 to 2013 by adjusting for team and competition effects
So, it has also been validated, in which case I believe the GIGO charge is a bit unjustified. Let's just agree to disagree on this.
Re: Hamilton. In the only season he competed against Alonso as team mate, he only equalled him on points and wins in one of the most messy F1 seasons in history in terms of team harmony. To then conclude that he dominated Alonso in that season...it is much shakier ground than that list of the greatest drivers. IMHO. Before you say, he had more poles. Yes, he's fast but is that enough to make one great? I think not.
hairyben said:
Dr Z said:
I really don't think you understand what the model does, to be honest.
I appreciate what the model is aiming to do, I just place far more emphasis on how flawed the input data is. And if it wasn't flawed enough already, he then "flawifys" it some more by converting modern scoring to the old 10-6-4 system; any tr00 fan will have no trouble explaining the problem here. wink smiley included to indicate my use of tr00 was a light-hearted way of making a serious point and shouldn't detract from it.Re interpreting F1 results using the same data is like deciding to knock a house down and rebuild it on the basis the wood it's made from is all rotten, then rebuilding it in a different style using the same wood again and declaring "look, we fixed it."
hairyben said:
Dr Z said:
Re: Hamilton. In the only season he competed against Alonso as team mate, he only equalled him on points and wins in one of the most messy F1 seasons in history in terms of team harmony. To then conclude that he dominated Alonso in that season...it is much shakier ground than that list of the greatest drivers. IMHO. Before you say, he had more poles. Yes, he's fast but is that enough to make one great? I think not.
Okay the hamilton things taken a life of it's own and you've missed my point by quite a margin. I never use the word dominated - the slang-ness of spanked was deliberate but has made you think I'm just being a fanboy - Yes I'm an unashamed fan but also objective and there was a very valid challenge to the concept within there (and if adding hamiltons latest years alters his status, that just proves the models flaws, given the model should already be "accurately ranking him" no?)
It basically takes the whole career of each driver and finds his peak form (1, 3 and 5 year intervals). It then compares the peak form between drivers to come up with the ranking. The model was published half way through 2014, so the ranking was based on Hamilton's performance from 2007-2009 which the model considered his best years at the time of publication.
The arguments here demonstrate the futility of even attempting to apply statistical analysis to such a feat.
Those of a certain age will remember a song by Joe Dolce called Shudupa your Face, a song that kept Ultravox's Vienna off the number one slot. Statistics suggest that the latter was therefore more popular than the former - which it wasn't as Vienna went on to sell significantly more, just not concentrated enough to hit number one, and onlookers without knowledge of either song might therefore deduce that Shudupa Your Face is a better song than Vienna.
If you are unfamiliar with either, go have a listen and let us know whether you consider this to be right.
Those of a certain age will remember a song by Joe Dolce called Shudupa your Face, a song that kept Ultravox's Vienna off the number one slot. Statistics suggest that the latter was therefore more popular than the former - which it wasn't as Vienna went on to sell significantly more, just not concentrated enough to hit number one, and onlookers without knowledge of either song might therefore deduce that Shudupa Your Face is a better song than Vienna.
If you are unfamiliar with either, go have a listen and let us know whether you consider this to be right.
Dr Z said:
Alright, this proves you don't understand the model!
It basically takes the whole career of each driver and finds his peak form (1, 3 and 5 year intervals). It then compares the peak form between drivers to come up with the ranking. The model was published half way through 2014, so the ranking was based on Hamilton's performance from 2007-2009 which the model considered his best years at the time of publication.
Sorry I haven't the time or TBH the interest to read and digest the whole rambling piece; I did scan for a concise outline of how he makes his calculations but he doesn't try to make it easy or convenient, all I found were reference to using the points themselves, thus you can't get away from the fact that the entry data is still just as flawed. It basically takes the whole career of each driver and finds his peak form (1, 3 and 5 year intervals). It then compares the peak form between drivers to come up with the ranking. The model was published half way through 2014, so the ranking was based on Hamilton's performance from 2007-2009 which the model considered his best years at the time of publication.
I still don't see how you moved past the flaw that a "drivers best years, measured by points" are still car and teammate dependant?
taking our best friend hamilton,he's nearly always had world class team-mates, except arguably kovalinen, who quite pertinently to the discussion (but O/T) was only actually there as flabio felt he would'nt make a suitably flattering no2 at renault for alonso. It's well known Schumacher had powers of veto over potential team mates at fez and we actually saw the safer team mates he agreed to physically relinquish race wins to him.
Team mates you/the team select to be no2's who let you by/crash on purpose to help your race will flatter your apparent "peak form" more than team-mates selected by the team to win WDC's so I fail to see whats been corrected. you've just re-arranged the same flawed data.
hairyben said:
It's well known Schumacher had powers of veto over potential team mates at fez and we actually saw the safer team mates he agreed to physically relinquish race wins to him.
It's well known is it? Please give me a source for this well known information because I've never seen it.And unless I'm very much mistaken, Schumacher relinquished more wins than he was given during his time at Ferrari at 2 - 1.
The survey has been misnamed as it wasn't intended to arrive at a "Greatest driver of All Time" finding. Apparently it was a demonstration of statistical techniques used by the university to promote its capabilities for compiling statistics. The person who organised the survey had no intention of arriving at a "greatest" result.
As ever, it's all about spin and publicity and getting the masses all worked up.
As ever, it's all about spin and publicity and getting the masses all worked up.
Gassing Station | Formula 1 | Top of Page | What's New | My Stuff