Ayrton Senna, McLaren, Monaco, 1990

Senna and Schumacher faster than Hamilton, official F1 study claims

2020 F1 season

Posted on

| Written by

A study comparing Formula 1 drivers across the past 37 years has determined Ayrton Senna to be the fastest of all time, ahead of Michael Schumacher and Lewis Hamilton.

The research produced by Formula 1 sponsor Amazon Web Services used machine learning to compare the qualifying performances of drivers since 1983.

Senna, who scored 65 pole positions during the course of his 162-race career, was judged to be one tenth of a second per lap faster than Schumacher, who collected 68 poles over 308 appearances.

Hamilton, who took his 92nd pole position in his 256th race appearance last weekend, is ranked third, 0.275s behind Senna.

Three other drivers from the current field appeared in the top 10: Max Verstappen fourth – five-thousandths of a second slower than Hamilton – Charles Leclerc seventh and Sebastian Vettel 10th. Fernando Alonso, who will return to F1 next year with Renault, is fifth.

F1’s director of data systems Rob Smedley, a former Ferrari and Williams engineer, said the model used was similar that adopted by teams to compare driver performances when choosing who to hire.

“Within the team environment this type of modelling is used to make key decisions on driver choices,” said Smedley. “As drivers are more often than not the most expensive asset of the team it is important that the selection process is as robust as possible. A process such as this therefore would be deployed by the F1 team’s strategists in order to present the most objective and evidence-based selection possible.

“‘Fastest Driver’ enables us to build up a picture of how the drivers compare, by analysing the purest indication of raw speed, the qualifying lap – and it’s important to note this pure speed is the only element of the vast driver armoury we are analysing here, to showcase the quickest drivers ever, which is very exciting.”

10 fastest Formula 1 drivers since 1983, according to F1 and AWS

RankDriverGap to best (seconds)
1Ayrton Senna0.000
2Michael Schumacher0.114
3Lewis Hamilton0.275
4Max Verstappen0.280
5Fernando Alonso0.309
6Nico Rosberg0.374
7Charles Leclerc0.376
8Heikki Kovalainen0.378
9Jarno Trulli0.409
10Sebastian Vettel0.435

Advert | Become a RaceFans supporter and go ad-free

2020 F1 season

Browse all 2020 F1 season articles

Author information

Keith Collantine
Lifelong motor sport fan Keith set up RaceFans in 2005 - when it was originally called F1 Fanatic. Having previously worked as a motoring...

Got a potential story, tip or enquiry? Find out more about RaceFans and contact us here.

255 comments on “Senna and Schumacher faster than Hamilton, official F1 study claims”

  1. Goes without saying….

    You only need to look at the teammates they had.

    1. OK, Think you need to explain that one. How did teammates impact Heikki coming in at 8 and Jarno 9th for example?

      1. The only thing right about this list is that Senna is at the top.

    2. Also

      A study comparing Formula 1 drivers across the past 37 years has determined Ayrton Senna to be the fastest of all time

      >Looking at about half of the history of F1
      >finding the GOAT
      >goAt
      >ALL time

      1. If they’d have looked at all time Jim Clark would have been in front, looks to me more like GITT… Greastest in those times.

        1. The explanation is probably very simple in that this is the timespan where exact timing data is available vs Clark had his times stopped by his girlfriend with a stopwatch. Still wouldn’t hurt to communicate this properly.

          1. Biskit Boy (@sean-p-newmanlive-co-uk)
            18th August 2020, 15:46

            The girlfriends with stopwatches didn’t do the official timing! They did the timing for the team, probably timing many cars at the same time.

        2. Perhaps more significantly, after his first of 25 victories, Jim Clark was never beaten by his team mate when his car was healthy. That includes Graham Hill of course. He also only finished 2nd once, and his lower placings were dwarfed by his win tally.

    3. Fred, given this gives us no information about how they came up with any of the figures, right now I would be very sceptical of the worth of a study that is not providing the information to allow for peer review.

      Indeed, in general the worth of this seems to be fairly questionable at best – what even is the “gap to best” really meant to mean? How exactly are they taking into account variations in regulations, such as qualifying with race fuel onboard or being unable to change set up between qualifying and the race, or changes in technical regulations that altered the performance of the cars?

      I half wonder whether the performance algorithm that they optimised is going to be the one that gets their name promoted as much as possible by this study, rather than the study itself…

      1. Precisely.

      2. Peer review would be essential for that study to be taken seriously. You can’t compare drivers from different eras without also comparing their cars and the tracks they raced on. None of that has been mentioned.

      3. When you know you know. The Sennas and the Schumachers are head and shoulders above the rest!

    4. This is kind of silly if it doesn’t account for the cars and the regulations which limits what can and can’t now be used on the car. eg turbo vs hybred era cars.

      1. Totally agree – trying to compare a driver whose tyres and engine were completely totalled by one flying lap, with others that whose engines alone had to last multiple events is going to be nigh on impossible for any algorithm to be particularly accurate.

        1. Part of the reason for Hamilton success is that is a far more sensible driver than his rivals. So he wont take unnecessary chances. He’s very pragmatic on the risks he takes and doesn’t take. As we saw in spain, he will drive within himself, doing only what he needed to stay ahead and secure the win, whilst preserving the engine and managing his tires. Come the end of the season he’ll have changed fewer parts.

          I wonder if there’s a comparitive stat on engines blown, or gearboxes changed?

    5. Hamilton has been teammates with three of the people on the list & has beaten them all. Rosberg beat Schumacher comprehensively as well. Jarno & Heikki, top ten ahead of Vettel? This list is weird as hell……

      1. Did Button not outscore Hamilton over the 3 years they were in the same team?

        1. In terms of the championship (the reason they race) Button outscored Hamilton in 2011. Over the three years they were together, Hamilton finished ahead of Button 2 of those years. Adding up all the points of the seasons just because Lewis had a terrible season (he admits it… everyone knows it… the whole world was aware of his personal life falling apart at the time… it’s not exactly a secret…) has been the desperate attempt over the years to discredit Lewis. Nobody ever mentions any other metric: not qualifying, fastest laps, fuel usage… not even the fact that Button repeatedly had to copy Hamilton’s setup. Somehow, (& rather simplistically) it always boils down to “hang on… but if we add up all the points, though…”
          smh

          1. Still not impressive considering Button was not an all time great either.

          2. @fasterpod Who on this list has had stronger opponents than Hamilton? Reigning world champion teammate on debut… even Button was a world champion. I don’t really care whether you think it’s impressive or not: that says more about your lack of knowledge & appreciation for the sport & those that compete. Getting beaten only twice by a teammate in literally DECADES of racing (both times due more to personal problems, external influences & reliability issues than anything else) is an achievement that stands on it’s own whether you approve or not.

          3. Button was a world champion? He was not even supposed to be driving in 2009 had Brawn not bought the team. Even while winning the title he could barely beat old Barrichello and got outqualified that year and over their 4 years together. Considering how Schumacher dominated prime Barrichello, Button argument doesn’t work against Schumacher. Barrichello is on this list as well and ahead of Button. Rosberg is higher than Barrichello but he was much closer to Hamilton anyway.

      2. Tommytintop (@)
        18th August 2020, 20:37

        This means nothing…. as the drivers are all in different cars, the cars are in different eras on different tracks plus about thirty other things that don’t add up.
        You just can’t compare them, no matter how good the technology is.

        1. The authors of that study seem to be comparing them just fine (not really though). Maybe you should be directing this response at them? I don’t even know where to begin with that list… I’d honestly have a hard time keeping a straight face wile debating anyone seriously trying to convince me that Heikki Kovalainen is really the 8th fastest driver in F1 over the last 4 decades. I’d be really interested to find out more about their methodology (“machine learning” is way too vague a description for me to take seriously without even an idea of the points of data made available)

      3. F1 fans: argue over who is the best of all time on forums after every race

        Amazon: creates algorithm

        F1 fans: you can’t compare between eras

        Me: 🙄🔫

        1. Nailed it!

          I think they probably produced this as a bit of fun, that’s being lost a little here.

        2. The list is fine and all positions are more or less explainable even Kovalainen’s. It is just Hamilton fans that don’t like it. They think Hamilton should be above Schumacher or Senna or both.
          OK, while we are at it, let’s put Rosberg in top3 then.

    6. Maybe they included it as well? And it became even more clear that Hamilton was not that great compared to them?

  2. kovalainen and trulli ahead of vettel? come on…

    1. AWS said it, so it must be true!

    2. Didn’t trulli get a lot of his best qualifying done when at Toyota on low fuel? Hard to take that into account I guess.

      1. Trulli was quit good in qualifying but we know the Trulli train his racecraft was average.

    3. Just what I was thinking. They really wouldn’t figure in any list of the top two hundred drivers lol.

  3. Kovalainen: gets hammered by Lewis and still makes it onto the top 10.
    Trulli: gets hammered by Kovalainen and still makes it onto the top 10.
    Meanwhile, Mansell, Prost, Piquet, Hakkinen, Hill, Villeneuve, Raikkonen and Button wonder what good their championships did…
    Such an absolute joke.

    1. Jarno Trulli at his peak was one of the fastest qualifiers in F1. He was considerably faster than Ralf Schumacher, faster than Jenson Button and Heinz-Harald Frentzen and even marginally faster than Alonso, making his position in the top ten probably correct. In his final year, 2011, however, his skills had declined considerably, and was destroyed by Kovalainen. The study probably didn’t fully take into account how much Trulli had declined in his Lotus years, but I still think Kovalainen could legitimately be placed in the top 10 qualifiers since 1983, as he was closer to Hamilton in qualifying than many people remember.

      1. Some stats about how many points drivers would have scored if points were done on grid positions (current points system):
        2001
        Jarno Trulli 94 (at the point when Frentzen left Jordan)
        Heinz-Harald Frentzen 44

        2002
        Jarno Trulli 65
        Jenson Button 38

        2003
        Jarno Trulli 127
        Fernando Alonso 123

        2004
        Jarno Trulli 150
        Fernando Alonso 141

        2005
        Jarno Trulli 202 (3rd in points)
        Ralf Schumacher 92

        2006 (much closer)
        Jarno Trulli 81
        Ralf Schumacher 79

        2007
        Jarno Trulli 48
        Ralf Schumacher 33

        Heikki Kovalainen (rookie) 46
        Giancarlo Fisichella 46

        2008
        Lewis Hamilton 292
        Heikki Kovalainen 214 (4th in points)

        Jarno Trulli 109
        Timo Glock 29

        2009
        Lewis Hamilton 149
        Heikki Kovalainen 56 (admittedly a terrible year for him)

        Jarno Trulli 135
        Timo Glock 59

      2. @f1frog I’m not denying Trulli’s abilities, but we need to be real and actually see what his qualifying performances would be, fuel-corrected. Trulli’s qualifying performance relative to his teammate dropped starkly at the start of 2010, which could be indicated by a lot of factors, including that maybe refuelling was flattering his qualifying performances. Also, arguing for Trulli and Kovalainen’s merits still doesn’t justify them beating out eight world champions to this top-10 list.

      3. Absolute nonsense where was this study when they just simply said Micheal Schumacher the fastest F1 driver of all time with 68 poles greatest driver of all time with 91 1st place podiums it’s just another study with them trying to come up with statistics to deny Lewis his rightful spot, all this scrutiny of generations to present day which is impossible really to do with the constant evolving of
        F1 never went on when Schumacher was crowned king but now Hamilton is close to dismantling everything these people hold dear everything has to now come into play, I wonder why ?

    2. Yeah, if I got this as the result of my “machine learning” clever algorithm, I would probably go back and check my assumptions for flaws and then have a good look at more data to base the learning on for another good set of learning @wsrgo.

      Montoya clearly wasn’t fast either, nor J. Villeneuve, and in the 1980 it seems there weren’t any super quick drivers at all, or at least not as good as midfield drivers in the early 2000s.

      1. Completely agree @bascb. I really hoped they used an algorithm that is easier to revisit rather than a black-box one, in which case it’s going to be really hard to interpret the mechanism and what caused the bias and what tweaks need to be made.
        While these exercises would be interesting, but this is kinda like p-values making not much sense on their own, interpretations should be strengthened with actually looking at the error bar overlaps or other thing. You can’t (or shouldn’t) make informed decisions based on one thing alone, especially if it’s such a complex algorithm giving clearly biased results. I’m surprised with Smedley saying this is something teams have used to select drivers.

        1. @wsrgo you are being very generous here. let’s call it for it is – a laughable waste of time. Check out f1metrics for a serious attempt at objective cross-era comparisons. It makes for fun reading even if you dispute the results (and indeed the model itself – after all “every model is wrong, but some are useful”. I don’t know where that quote comes from but it’s endlessly useful in the field I work in, health economics).

      2. @bascb The average driver skill has gradually improved. Back in the 1980s Senna was considerably faster than Prost on average in qualifying, as shown by the model, even though Prost was one of the quickest drivers of his time. Nowadays there are many insanely fast drivers, so it’s not a surprise to see many of them in the top 10.

    3. Maybe Hamilton didn’t put that much of a gap on him?
      And maybe they included Trulli being off his prime and he was pretty good until then. His gap to Alonso was bigger than Hamilton’s.

  4. Kovalainen in 8th a Trulli in 9th? Defintely lost lots of credibility with those results. Another incorrect AWS insight

    1. Two drivers who should never have been in F1, never mind a fastest list.
      This venture its crap, I fear.

      1. Jarno Trulli

        should never have been in F1

        In case you have actually seen any race featuring Trulli (you haven’t): Have you been facing the back of the TV?

        1. Lol. I’ve seen plenty of races with JT, probably all. They all took twenty minutes longer than necessary.

  5. How can Trulli and Kovalainen be there, LOL? I don’t really get this measurement. What does Gap to ‘best’ mean in this context?

    1. Kovalainen is a bit of an odd pick (but it’s AWS “insights”, which are complete nonsense anyway) but Trulli had a reputation for being a superb qualifier, hence the Trulli trains.

      1. He was also a legendary train in every race that the others had to pass early or be held back and the race carry on without them.
        As for Kovaleinen…… words fail me how useless he was, and I don’t mean to be unkind!

        1. His race pace was poor, but in qualifying he was quite good.

  6. Don’t ever tell about anything reliable or real coming from such an atrocity which is AWS!

    Awful service, providing totally fake data based on some perverted algorithms.

  7. Kovalainen himself recently admitted he had no business racing against Lewis Hamilton. Yet AWS claim he is the 8th best driver since 1983? The calculation is probably correct, but the data input is clearly wrong. Much too much credit is given to the car if these stats add up.

    1. Well, one point is that Kovalainen was quite fast in one lap pace, at times @chrischrill. That is also where Trulli comes up. Since this has nothing to do with racing, just about qualifying speed.

      Still it seems to ignore different formats of qualifying, ignores just about all drivers from the 1980s quite a few top drivers from the 1990s and a decent amount from the 2000s too. To me it does show their dataset, their assumptions or their “machine learning” algorithm need a lot of work to find reasonable results that are less skewed.

    2. Not “best”, but “fastest”. There is a clear difference.

  8. Trulli, however, may deserve his spot. Remember Trulli trains? They came about because Trulli could qualify the car much higher than where it should be.

    1. I vaguely remember that was when teams could choose the starting fuel load, so Trulli would go light into qualifying to over achieve the cars potential then crawl through the race fuel saving.

    2. Trulli was lightning fast in qualifying. It had nothing to do with qualifying with too low fuel loads. He qualified the car where it shouldn’t be.

  9. Doesn’t this system mainly use gap to team mate to create a delta time, and then look at how far above, on average, a given driver is? It therefore is meaningless as it does not take into account drivers, such as Schumacher, who were clear team number 1s, given better equipment than their teammates, and paired with average drivers.

    It’s quite telling that Hamilton, in 3rd place, has been the team mate of 5th, 6th and 8th place on this list and has beaten them all.

    1. You think so because Schumacher destroyed his teammates all unlike Hamilton could.
      Now, we have to act like Button, Rosberg, Bottas are all time greats.
      And if there is limited equipment, of course the better drivers gets it. Did they give Nr2 drivers the bad equipment on purpose? It doesn’t make sense.
      Check 2002 brazilian GP. There was just one F2002 chassis available. What was Ferrari supposed to do? Wait until two are ready? And both Schumacher and Barrichello drove old F2001 for the first two races. So, did Ferrari give Schumacher the bad equipment, too?
      This entire “Ferrari hampered Schumacher’s teammates” narrative is a logical fallacy. The guy was getting paid 10 times more. What would Ferrari gain from hampering them to make Schumacher look good??
      And Schumacher would love his teammates to be faster so that he could steal points from his title rivals just like he did in in his return at Sepang99.

  10. Rule changes make this meaningless.

  11. Keep cashing the cheques Rob, I’m sure you know this is as much BS as we do.

    The desperation in humans to quantify and to attach numbers to absolutely everything depresses me greatly.

    “Computer says no Mr Prost, goodbye.”

    1. Absolutely right. Who cares? They don’t give points out for qualifying.

      AWS predictions in races are nonsense, even Pirelli couldn’t understand their “tyre condition”.

      The sad thing is that HMRC alone pay AWS 5x as much as Amazon pay in tax in the UK.

      I was going to write something more about this, but frankly this is such excrement that I’m losing the will to live.

    2. Actually this statistic seem to reflect quite much whether one had good teammates, or whether one had been a dedicated 1st driver, although of course it reflects whether someone is or was a good qualifier too. To some extent, because to create a really good model that quantifies and outputs really relevant information is not that easy.
      I’d say it’s not the worst thing ever seen, but as Amazon is a top tier company, they should have worked a bit more on it.
      As many says I don’t see what this gap to best means, but likely some quite complex combination of differently weighed functions (so it would be difficult to name it). And they emphasize that it only reflects raw pace. Therefore there is Trulli, whose showing up here surprised me, but thinking a bit I recalled that he was not that bad (but the fact he beat Alonso in 2 seasons considering quali performance shocked me even as a historical fact).

      So I think gap vs teammate is likely weighted quite heavily in this, and therefore it gets distorted if someone had equal or at least strong teammates throughout a carreer or when someone was a dedicated 1st or 2nd driver of a team for a long while. This obviously takes teammate comparison into account because Kovalainen had a longer stint in lesser teams too, and there he had teammates to beat more easily than many on this list. Or Hamilton had stronger teammates than Schumacher, and therefore this distorted the result for Hamilton slightly negatively and for Schumacher slightly positively.

      It’s very obvious one can’t put together a model like this without taking a composition of many kinds of input statistics. So I’d say they should have asked as many drivers from modern era of F1 and as many race engineers and similar highly competent trackside engineers to list what factors to take into account, and when that list is assembled, in an other round ask them to weigh them numerically.

      And even if they done so, teaching an AI, or a neural network to produce results is a lengthy iterative process, where they feed it with data, and check whether the output of the algorithm is close enough to reality or the desired result. And if not, they modify the weights of different kind of inputs and repeatedly feed it with the input data. They really try to simulate the learning process and interoperation of multiple (layers of) neurons.
      I think it’s not bad, as to have a nice simulator many try to model the car parts in as many aspects and as high detail as they can, so they literally build the car’s model of screws, springs, tyres and model their physical properties and so on. And if that’s done quite well and is backed by a nice physics engine then the outcome might be a nice and enjoyable simulator, or at least a simulator that works and can be fixed to some extent with modifying properties and weights of parameters.

      But what’s the purpose of AI here if many engineers with relevant knowledge of F1 provide the weights? Because taking the average of the weigths provided by them should be accurate. So they can make an AI and fiddle with parametrizing and feeding it with data until it provides the same results as the opinion of those engineers, instead of using their averaged opinion to simply create a statistics. Feels a bit l’art pour l’art, the snake have bitten itself as one characteristic of real tasks for AI that they are hard to solve. But algorithms for choosing the driver to hire are in existence for a while. Probably it’s just an experiment on the route of creating something bigger, for example a really capable and much more general AI in the field of human recruitment. Hopefully they will create very nice HR workforce, but there is a chance of creating a master spy :)

      1. Hehe, I not knew the word fiddling has such a negative tone (instead of something like experimenting).

    3. So, instead of facts, would you like to consider Button and Rosberg as all time greats just because they beat Hamilton?
      What fact would you like to see?

  12. I totally curious about the software for analysing driver performance to help with recruitment, it’s almost like a moneyball situation seeing value where others don’t.

    1. I can see how you can use machine learning to analyse a driver’s overall performance level when they’re competing in more or less the same era. Where I think this likely falls down is in comparing drivers who raced under vastly different technical conditions. All drivers will have good and bad performances – the variation between good an bad is likely a factor, but there needs to be some kind of equivalence factor applied; Senna never drove a car anywhere near as fast as the current F1 cars, and wouldn’t have a clue about modern hybrid systems. Nor would he know the first thing about how to extract the best out of modern tyres. But likewise, Hamilton would likely struggle with a manual shifter, unassisted steering, and comparatively terrible brakes.

      TL;DR – you can’t realy compare drivers from different eras because the skills were different.

      1. You are right, I suppose that means the same with all sport though as styles and techniques change over time.

  13. So the fastest Finn is Kovalainen? Not Mika, not Kimi, not Bottas.

    OK.

    1. Häkkinen wasn’t that fast in qualifying. He only looked good because he had Coulthard as a teammate. Räikkönen is also stronger on race day than in qualifying. Bottas is 13th, though. Bottas and Kovalainen are similar drivers. Good in qualifying, not so good in the race.

  14. Quite interesting, and then I read number 8.

    1. Absolutely waste of time different cars different set ups different tyres how can anyone just compare anything in different eras is just stupid it’s like comparing pele to Ronaldo complete stupid

  15. Replace Kovalainen with someone like Mansell and promote Hakkinen into the top 10 in place of Trulli (who should be top 15, but not quite top 10), and it looks considerably more realistic.

  16. As this is just based on qualifying lap I would have probably come up with the same top three; but overall it does look like a typical AWS model that has a 100% success rate of being wrong.

  17. As if it was that easy to condense to all the countless variables across 70 years of racing.
    If F1 was actually serious about this, it would be a spec series, which as all know will never happen.
    Total nonsense.

  18. “Data” – “Modelling” – “Robust” – Hmm, in what context have I heard these words before? Was An Algorithm involved?

    1. Paolo (@paulsteward40)
      18th August 2020, 21:50

      It was almost certainly the same algorithm used to calculate A level grades in UK. 😂

  19. Where is Daniel Ricciardo who beat vettel and verstappen in qualifying as teammates. Really bizarre ‘study’. Also where is George Russell, 26-0 against teammates in F1.

    1. Jose Lopes da Silva
      18th August 2020, 11:10

      People saying that Ricciardo beat Verstappen in qualifying does not make it true. You can’t count pole positions alone.

      1. Kpcart you seem to not understand this study, nor the reality of Max and DR at RBR.

        So firstly this is not about teammate comparisons.

        Secondly, yeah DR beat SV in 2014 quali 11-8. This was the season SV had had his 4-time WDC/WCC car ripped away from him and replaced by a car that was nothing like he had become accustomed, as the plan to stop the RBR juggernaut worked, and at the same time DR was promoted to the best car he had ever had, as terrible as it felt to SV. You had an extremely frustrated and dejected SV vs a stoked DR, all the pressure on SV.

        And thirdly, DR beat Max in quali 11-6 in their first season together when Max had joined the team after the first four races and didn’t have the luxury of pre-season testing, nor the same time the more experienced DR had to gel with the team. And still Max won his first race with RBR. And Max had outqualified Sainz 3-1 at Torro Rosso to start off that season.

        Then the quali head to head goes 13-7 for Max in 2017, and 15-5 for Max in 2018. But yeah, you cling to whatever facts make you happy.

  20. I always knew Kovalainen and Trulli better than that Prost guy. Nobody would believe me now I have official confirmation. How ’bout that.

  21. This is so wrong that it *even* put Kovalainen in front of Trulli. It’s like if the AWS was given a mission to misrank a driver in every possible way with Kovalainen.

    1. Trulli outqualified everyone except Kovalainen, who he faced much after his prime.

  22. I’m only mildly interested in the order of drivers here, but with all of the variables (and there are many …), they feel able to draw a distinction between drivers to the 3rd decimal place – really? After observing the veracity of the AWS produced graphics during the race, I have my doubts

  23. I’m not going to discuss the order, especially after the top 5!

    The thing I will add is that it seems ridiculous for F1 to be doing this “Fastest Driver”. I don’t think F1 fans want to be told the best (especially with machine learning – which is only as good as the data), I think they want to work it out for themselves. It is the same during the races, I don’t want to be told via an AWS graphic that Verstappen will be 1.5 second clear of Bottas after a pit stop, I want the unknown of looking at the gap thinking “this will be close” and watching the unpredictability of live sport – not being told what’s going to happen or what to think.

    The other factor they haven’t considered is what car they’re driving. The cars changed a lot over the years! Even getting the current 20 drivers to set lap times in all current 10 cars, you would get pretty different results depending on the car driven. This is why F1 should say out of the debate, and let the fans decide

  24. I wonder if Kovalainen got on this list only by beating a fast Trulli, who was unmotivated or otherwise underperformed by the end of his career :)

    Schumacher was probably noticeably faster than Senna once he hit his stride. Certainly from 1994 onwards.

    1. Zlate Sitnikoski
      18th August 2020, 13:13

      How??? Senna out qualified Schumi in all 3 races before he died

      1. Schumacher was in a much slower car than the Williams in 1994.

    2. Hard to say. Both Senna and Schumacher had amazing speed. I don’t think Schumacher would have been out-qualified by Berger, for example, which happened to Senna from time to time. Especially in the mid-1990s he was much faster than anyone else, but his rating probably suffers from his relatively poor performances in 2005 and 2006. However, Schumacher certainly was a better racer.

  25. Hmmm. Does this mean Ham beats Ver in both the RBR or Merc, if they raced for the same team?

    1. By a whisker.

  26. Qualifying is nice and all but race day pays the bills

  27. If someone is thinking on using AWS machine learning for something professional they would lost all confidence on it after this weird results.

    Hakkinen did 20 poles in two seasons when he had car for that, he was probably the best qualifier in the late 90s.
    Still, Kovalainen who drove a world champion car to nowhere is in front of him.

    1. Schumacher was by far the fastest driver in the late-1990s and therefore well ahead of Häkkinen.

    2. Hakkinen won all those poles with a clearly faster car.
      He outqualified Schumacher over a season only in 1998 and 1999, when even Coulthard could do so.
      Schumacher was undoubtedly the fastest driver over a lap during his first career.

  28. Yes (@come-on-kubica)
    18th August 2020, 10:47

    I mean the best driver of all time is Jim Clark there’s no disputing that.

    1. But this isn’t a list of “best drivers”. Best is subjective, whereas fastest is at least objective and measurable (if you ignore the hundreds of variables).

    2. Zlate Sitnikoski
      18th August 2020, 13:16

      I won’t dispute that because it’s true. Jimmy and Ayrton are unmatched even today.

    3. I agree that Jim Clark was the best of all time but it’s definitely a subject that’s up for debate and is disputable.

    4. I notice non-British never chose him. Over-inflation by Brits or devaluing by non-Brits?

      1. I’m not british and I consider him at the top of my list! The stats are phenomenal and his ability to drive around mechanical issues was just insane. I suspect he’d dominate in modern F1 with his ability to drive insanely fast but take nothing out of tyres.

    5. Sadly he isn’t in the data, so who knows.

  29. Like a good accountant a good statistician can make the numbers add up to anything the customer wants.
    I’ve always thought it futile to compare drivers from different eras. Different cars, tyres, track surfaces, track lay outs, driver fitness, the level of information current drivers have at their disposal the variables are huge. At best these figures are a curiosity at worst click bate.

    1. Exactly. People decry modern racers and say they wouldn’t be able to drive cars from ‘the old days’. Equally, drivers from those days would be nowhere near fit and strong enough to drive today’s cars on the limit.

      It’s fun to play ‘is Hamilton better than Fangio’ type games, but the reality is it’s impossible for it to be anything other than subjective opinion.

  30. Why 1983 onwards? If this is the first batch of analyses, then don’t bother doing any more. I feel embarrassed for Rob Smedley (& Heikki Kovalainen to an extent, but he’s blameless). No numerical analyst should publish such flawed “research” (it makes me a bit angry actually having worked in analysis myself).

    Why has F1 become so backward? I loved the (apocryphal?) anecdote that Enzo Ferrari wanted to crush last year’s cars because he was only interested in the future performance.

    1. Because there is very little actual news with the majority of F1 players locked in there bubble. I’d say about about 80% of the news is garnered from the live interviews that we all have access to and then edited and packaged up as news or an exclusive.

    2. Might be they lack some of the data from pre-1983, or couldn’t bother converting it into digital form it it exists only on paper.

    3. Timing was more accurate.

  31. We’ve seen all too recently just how bad all this machine learning and algorithm nonsense is, with the A-Level and GCSE debacle here in the UK… therefore, (and aside from this being generated by AWS), this has even less credibility than a Ferrari rules protest

    G

  32. Jose Lopes da Silva
    18th August 2020, 11:08

    Never a comment section has been so blindfully disappointing as this one. Even in this politically polarized times.
    This is about qualifying performances. Says nothing about overall performance as a racing driver. What’s people’s problem?
    Older guys are not aware that Prost lost 26-6 (or something) to Senna in their 32 intra-team qualifying battles? How does that make Prost less of the worthy 4-time champion he is, one of the all time greats?

    Let me guess. Emotional people who know zero about statistics feel offended and have to scream. The others say nothing.
    It’s a good example of what happens when the right people keep silent.
    Thanks to RocketTankski for not keeping silence.

    1. Even taking into account what you said the list is very dubious, especially. Also your opinion is not better than anyone else’s

      1. @kpcart

        Also your opinion is not better than anyone else’s

        I’m not sure that @jose‘s comments are “opinions” though. I would be the first to laugh at these AWS stats, however what he states is that the article contains a direct quote from F1/AWS saying that

        “‘Fastest Driver’ enables us to build up a picture of how the drivers compare, by analysing the purest indication of raw speed, the qualifying lap – and it’s important to note this pure speed is the only element of the vast driver armoury we are analysing here”

        The very fact you seem to have jumped on him, stating that his “opinion” is no better than anyone else’s, just goes to cement his other comment about current levels of discord and polarisation of views

        G

      2. @come-on-kubica Well plenty will dispute that. In my opinion any all time great list is entirely subjective because there is no way to quantitatively compare drivers across different eras. Everyone will have their own opinions, although as a Scotsman I tend to have Clark as my number 1 too.

        But for the record, this analysis only featured drivers from the past 37 years so several notable drivers including Clark are not included.

        1. Comment in wrong place obviously. Was a reply to Yes a few comments up.

    2. Seemed easy to me as well. They laid it out clear. Quali, one lap performance, data corrected to make them comparable.

  33. What a load of old tripe.
    How can Verstappen (2 poles) be on that list, when the closest driver to Hamilton in the last 3½ years, has been Bottas (13 poles). Who isn’t even mentioned.

    1. Its not about counting Poles, you need a way to exclude the cars potential.
      Bottas is in the fastest car and can occasionaly extract the maximum performance of the car (thats why he got 13 poles), but more often than not he is unable to do it.
      Verstappen is in a far inferior car, no driver in the world would be able to get 13 poles with the RB against the Mercedes.

      If all drivers had the same car Bottas wouldnt be near the 13 poles he currently has.

      1. That doesn’t make sense. Bottas regularly out qualifies Lewis and yet…

  34. Here is the rest of the list racefans didn’t include:
    11 Rubens Barrichello 0.445s
    12 Nico Hulkenberg 0.456s
    13 Valtteri Bottas 0.457s
    14 Carlos Sainz 0.457s
    15 Lando Norris 0.459s
    16 Daniel Ricciardo 0.461s
    17 Jenson Button 0.462s
    18 Robert Kubica 0.463s
    19 Giancarlo Fisichella 0.469s
    20 Alain Prost 0.514s

  35. GIGO

    An attempt to justify a data service which gives us as much amusement as the tyre wear graphics and frustrates us as much as race stats which can’t always keep up with track action despite the wads of cash put into it.

    If teams really rely on this stuff to choose a driver is it any surprise the grid is so pear shaped.

  36. Well this data is meaningless until AWS gives us the accurate calculation of tyre performance the drivers were using when they set their qualifying laps.

    In all seriousness though, while I can believe the teams use some sort of similar algorithm to compare current drivers, it seems to have fallen apart badly when trying to compare drivers across different eras, much like every other similar analysis I’ve seen. In my opinion it’s impossible to accurately quantify the performance of drivers from different eras, there are just too many different variables and no common datapoint to compare them to.

    Drivers’ performances also vary throughout their years due to age, confidence, compatibility with their car etc, and usually that throws up some strange results. Want to feature high on this list? Go up against a great driver who is either past their best, or not comfortable with the car/team and outperform them – that’ll gain you a bunch of performance ‘points’ that are based on nothing but your teammate’s reputation.

  37. According to my mum’s algorithms I’m the best looking guy in the world.

    1. Zlate Sitnikoski
      18th August 2020, 13:20

      That’s not what she told me lol

      1. @dot_com. According to your mom I am too….. Sorry, couldn’t help myself

        1. Haha. :-)

    2. Paolo (@paulsteward40)
      18th August 2020, 21:53

      They are the same algorithms that gave your sister a D grade in her A levels this week 😂

  38. Heikki Kovalainen? I think AWS must also have done the algorithm for the UK A level results.

  39. To think a few minutes ago I was just thinking about research fraud and check in and find this.
    I’m sure someone some where will keep a straight face and call it exact science.
    Next week’s new finding :- dinosaurs were able to watch Netflix

  40. Odd results to say the least, were they using the same algorithm used for the A-Levels/GCSE’s results for School kids here in the UK?

  41. Fair play Heikki!

    BTW if you don’t understand the data, that doesn’t automatically mean it’s wrong. That’s how religion persists.

    1. As a scientist myself, there’s a lot of algorithm revision we do when the results turn out to be completely unexpected and different from a hypothesis, especially if it’s a black-box algorithm that doesn’t allow informed, interpretative tweaks to models and variables used. This isn’t post-hoc modifications to prove a hypothesis, it’s a proper re-examination of every step followed.

      Interpretations are as, if not more important, than the algorithm operation themselves.

      I understand the temptation to use a rationalists’ dichotomy here, but it’s not exactly accurate to do so always.

    2. By the way, if you believe what they tell you without any proof it’s true that’s how religion persists

  42. It is hard to take such ‘studies’ seriously, isn’t it. As we celebrate 70 years of the World Championship, only 37 years are considered. Moreover, five of the top-ten are essentially current drivers. Nonsense…

  43. Think the AI needs to be checked. The gap between 1st and 3rd is .275 and .161 between 2nd and 3rd yet the gap between 3rd and 10th is .160. So the first two are just fast but too fast to be believed.

  44. I think Heikki Kovalainen is going to trend now more than he ever did during his racing career.

  45. There is this other rock solid study, which can be checked by the fastest quantum computer, listing these drivers as follows:
    Rank Driver
    1 Ayrton Senna
    2 Charles Leclerc
    3 Fernando Alonso
    4 Heikki Kovalainen
    5 Jarno Trulli
    6 Lewis Hamilton
    7 Max Verstappen
    8 Michael Schumacher
    9 Nico Rosberg
    10 Sebastian Vettel

    1. @coldfly Nah i cant believe that list either. Heikki still way too low. Unless your algorithm has Kovalainen, Trulli one and two, I’m not interested ;)

      1. @keithedin, maybe if we try a different alphabet.

        1. @coldfly I’m afraid your comment was too subtle for me and at a glance I didn’t even work out your ‘algorithm’. But your system might have as much merit as the AWS one, and probably a bit cheaper to implement.

  46. Don’t need an algorithm to know that over one lap, JP Montoya and Mika Hakkinen would both breeze into my top 10.

  47. Problem is these tools fall on their face if there is a #1 driver situation. Schumacher’s team mate was never allowed to challenge Schumacher. Like the example when Herbert was faster through practice in Monaco, they took away his access to the telemetry and then obviously Schumacher ended up being faster.

    Plus the car was always designed around his particular driving style. Strongly disadvantaging his team mates. We see the same with Verstappen since Horner stated they would build the team around Verstappen. When it was still a fair play team, Ricciardo wasn’t slower than Verstappen at all. In fact Ricciardo performed better than Verstappen (ie Ricciardo wasn’t slower, but he made much less mistakes). In 2014, Vettel and Ricciardo were shown to be evenly matched for qualifying.

    Besides, most of Hamilton’s team mates are in that list. So he’s apparently compared to team mates who were great qualifiers too. Yet he beat them. It would be easier for other drivers who have less fast team mates.

    If you start with bad data then it’s a case of garbage in, garbage out.

    I will believe that Senna was the fastest driver ever though. We have seen him actually stepping into his team mate’s car and going a second faster. Or beating Prost by 1.5 seconds in the “same” car.

    Qualifying was a totally different game back then though. They were using special qualifying tyres and even qualifying engines. If we had that today perhaps different drivers would be faster.

    1. Aah, good ol F1oclown clutching at straws:
      https://www.racefans.net/2017-f1-season/statistics/qualifying-data/
      https://www.racefans.net/2017/12/01/2017-f1-team-mate-battles-verstappen-vs-ricciardo-at-red-bull/

      “In the end Red Bull’s finishing rate in 2017 was so poor the cars only took the chequered flag together seven times in 20 races. “Had that been better, Verstappen’s margin of superiority might have been reflected in the points tally.”
      Perhaps the most telling detail was that Red Bull chose to extend his contract before Ricciardo’s.

      Please, keep making a fool of yourself! It’s the stuff of legends.

      1. Oconono

        So you give two links that barely show a difference between the two if you take out the issues/outliers. Which means you have simply proven my point!?!?!

        Besides, I was talking about 2018. You know … the season it actually happened?

        So yeah keep making a fool of yourself man.

    2. What would Ferrari gain from not allowing his teammates to challenge Schumacher? Even Schumacher would love his teammaets to be faster so that they could steal points from his title rivals just like he did in Sepang99 for Irvine.
      I thought that telemetry hiding happened in Argentina. The only source for that is a 25 year old interview by Herbert. And I don’t think Herbert became a more succesful driver after 1995. I wonder if his next teammates hid their telemetry, too. And the same logic applies here. Why would Benetton sabotage its own drivers?

      1. @fasterpod By contracting teammates like Barricello to not compete they then could build the car totally with MS in mind. So not only did MS have a designer car, he at no point had the psychological war that many drivers have with their teammates. Even if the odd time RB outqualified MS, he was not to finish the race ahead of MS if MS found himself behind the rare time and within a shout. See Austria 2002 for the most glaring example of the skewing towards MS and the post-race interview in which RB admits ‘obeying his contract.’

        Why would they do that? It was their business model to help MS end the Ferrari WDC drought.

        1. Austria2002 is the Most? and only. Tell me the other times it happened.
          That is a horrible business model. Hampering his teammates to make him look good while paying him 10 times more.

          I told you my reasoning already.
          Had Barrichello been faster, it would be great for Schumacher. Because, even if he were faster, he’d let Schumacher pass, right?

          They could build the car for Schumacher’s likening?
          If an oversteery car is faster and his teammates couldn’t adapt, then it is not Schumacher’s or Ferrari’s problem. They could make a stable car on rails, the gap between Schumacher and his teammates would be slower that way, but Schumacher would lose the title. Which F1 team would risk a WDC just to make their drivers close to each other.

      2. You do realise that MSC had his own bespoke tyres that none of the other drivers (even his team mate) had access to?
        He had seriously preferential treatment and his records have to be taken in context.
        Remember he also took Hill out in Australia when he knew he was retiring after he broke his suspension.
        Any other driver would have been stripped of that WDC, but not Ferrari and Msc.
        Watch it on YouTube. It is absurd that he was allowed to win that WDC.

        1. Sorry Benetton not Ferrari

        2. You think Schumacher didn’t deserve in 1994? Because some LCS code was found after the races he had a lot of wheelspin?
          Yeah I watched it a lot of times and couldn’t care less even if he did it 200% on purpose. Enough was enough that year.
          Hill, who I think definitely deserved 96 title, shouldn’t have won in 1994. That’d be the absurd thing.

          That preferential treatment is called Nr1 status and every team does it when needed. Alonso era Ferrari and 2017 2018 Mercedes are just another examples.
          I have already told you that hampering his teammates (Barrichello was not given access?) wouldn’t help Schumacher either. For example, if Irvine could steal 1 more point from Hakkinen in 1998, winning at the last race would be enough for Schumacher to win. With 4 points in 1997, a 2nd place at Jerez would be enough as well. These are very basic examples.

          1. So do explain why Ferrari and Bridgestone provided him with unique tyre designs that Irvine et Al didn’t get?
            Please explain how this isn’t an unfair advantage.

            There were all sorts of other objections to Schumacher. Senna was convinced he had drivers aids like traction control illegally on his Benneton.

            If a driver did what schumacher did to hill and Villeneuve today he would be excluded from the championship and receive a ban on his super licence.
            He certainly did not deserve that wdc in 94. He cheated.

            His legacy is tainted by proven disparity in tyres, dangerous and unsporting behaviours, team orders to his advantage, team mates who didn’t even know he was getting different tyres, but were contracted to let him part in raced- see Baricchellos quote after he stopped on track to give a win to him and persistent rumours of technical cheating by his team.
            Once he got equal equipment with Nico Rosberg his illusion was shattered, all notions of being the best ever were gone.

        3. First, bridgestone started working with Ferrari in 1999. That special tyre rumour came around 2002, when Irvine was not there. That shows you don’t know much about the subject. And, Irvine never mentioned anything you said, he always praised Schumacher’s talent. I will not get into the details of the tye thing, you already didn’t read what I previously said. but what do you mean? Why would Ferrari give its own driver especially while paying Schumacher 10 times more?
          Give me the source, it’ll be about 2003 season probably.

          He couldn’t win the title in 1997 anyway.
          Would you exclude him? Good luck with that. If you think 1998 season would be watchable without Schumacher, then you should watch some other sports.

          He deserved in 1994 after all those bans, which you don’t know. I couldn’t care less if he crashed into Hill on purpose or not. Hill, who I think definitely deserved in 1996, hadn’t deserved in 1994. It would be a shame if Schumacher lost that year with all those bans.

          Senna heard TCS sounds and he was convinced? So what? LCS code was found in the car but his starts had been very bad until Imola when the code was found. Then, how can he have TCS?
          And TCS sound is pretty understandable. If he were using it, thousands of people there can hear it. Or we could just basically on videos of those races.
          By the way, do you know what TCS or LCS are?

  48. What a load of nothing.
    Meaningless as it can be.
    Verstappen top 4 without achievieng nothing takes any credibility away.

    1. This is not about achievieng something (9 wins i would that call soemthing) but all about relative speeds compared to all drivers.

    2. What more could have he done since he joined in 2015??

  49. No Maldonado on this list, I’m disappointed!

  50. Whilst I regard Alonso as one of the best drivers of his generation, I don’t recall qualifying actually being that impressive even at his peak. But then I see kovalainen in the list and have now disregarded the entire “study”… I wonder how much it cost to conduct?

    1. Few cents for electro costs.
      Amazon just decided to showcase their useless AWS services… and of course miserably failed.

    2. Coz Kova beat Trulli, Trulli beat Alonso!

    3. Tommy-c. Your comment is THE correct answer.

  51. Looks like OFQUAL went to AWS for their algorithm…

  52. Wonderful to see such scepticism around here among the RaceFans, instead of kowtowing towards the sponsorship like F1 obviously has chosen to do.

  53. Michael (@freelittlebirds)
    18th August 2020, 13:59

    It’s impossible to compare drivers from other generations to new generations especially their qualifying in terms of speed. If anything, putting Lewis in 3rd and Alonso in 5th completely discredits this study. Also there’s race pace and Lewis and Alonso are pretty excellent in that regard.

  54. I really cannot wait for AWS and their ridiculous, incompetent, pointless statistics to get as far away from Formula One as possible.

  55. It’s a nice discussion-generator, but I can’t convince myself that the computer is able to account for enough variables to make this list anything more than that. The varying influence of driver ability on final lap time (has changed a lot since 1983), qualifying session format, qualifying rules, tyres, cars more suited to one lap, cars more suited to race distances, cars more suited to different tracks, team politics, team-mate quality and consistency, and so on.

    I’ll just park this one in my AWS drawer alongside the tyre graphic and its mates.

  56. Is this the same AWS that tells us Hamilton’s tyres are at 10% performance when he pulls out a fastest lap?

    1. Where have you been man? Lewis doesnt need tyres to win races!

  57. I think a big factor to modern day drivers scoring so high are the smaller qualifying margins. Back in the 80s and 90s, differences of well over a second per lap were quite common, even between competitive drivers, whereas in current days regular performance differences of sometimes less than half a second are reasons for driver seat rumours (Kvyat, Albon, Stroll, to name a few). The reduced margin has partly to do with the increased driver quality standard, but likely more so with the increased involvement of pitwall communication. Drivers who would have struggled in the past to get the tires and the brakes up to temperature, are in current days far more competitive. Or, in terms of this research, lose less time per lap compared to top level drivers.

    The result of this model is that high level drivers of the 80s and 90s will score lower than their modern day peers and possibly, that top level drivers of the 80s and 90s might in fact score better than their modern day peers, since inter-era links (Senna generation vs. Schumacher generation and Schumacher generation vs. Hamilton generation) will also be affected. The difference between Schumacher and the drivers from a previous era (Brundle, Patrese) are larger than between Schumacher and drivers from a later era (Massa).

  58. Just some random thoughts on this…

    The problems with machine learning seem to arise moreso if one is trying to use the findings for predicting the future, whereas this list is not about prediction.

    Nobody seems to have a problem with the list other than Kovalainen and Trulli being there, so they got those wrong but the others right?

    The reaction to this list or it’s accuracy would be different if it was LH on top.

    I have faith that Smedley is not just playing with crayons here and would be well aware of the different eras and rules between 1983 and now.

    I think it is so cool to see Max at 22 in the midst of a bunch of WDC’s and has never had a top car, unlike them.

    1. Yes, If Hamilton was one step above, people would love it.
      I don’t know what is surprising here. Verstappen beat all his teammates including Ricciardo who beat Vettel.
      Schumacher beat all his teammates. Haters love acting like his teammates were garbage but Ferrari management of the time and the numbers don’t agree with that.
      Hamilton may have the pole record but even Rosberg could pass Schumacher’s pole record had he stayed. He got 29 compared to Hamilton’ 35 during their time.
      Hamilton and his teammates had 76 front row lockouts compared to Schumacher’s 23 with his teammates.
      Mercedes advantage Hamilton enjoyed is clear as day and night.

    2. @robbie I would say that there are a lot of people who would question the validity of the whole premise and the whole list, as there seem to be some rather more noticeable trends within that data.

      If we look at what is apparently the top 20 list, it plays out as follows:

      1) Ayrton Senna: 0.000s
      2) Michael Schumacher: 0.114s
      3) Lewis Hamilton: 0.275s
      4) Max Verstappen: 0.280s
      5) Fernando Alonso: 0.309s
      6) Nico Rosberg: 0.374s
      7) Charles Leclerc: 0.376s
      8) Heikki Kovalainen: 0.378s
      9) Jarno Trulli: 0.409s
      10) Sebastian Vettel: 0.435s
      11) Rubens Barrichello: 0.445s
      12) Nico Hulkenberg: 0.456s
      13) Valtteri Bottas: 0.457s
      14) Carlos Sainz: 0.457s
      15) Lando Norris: 0.459s
      16) Daniel Ricciardo: 0.461s
      17) Jenson Button: 0.462s
      18) Robert Kubica: 0.463s
      19) Giancarlo Fisichella: 0.469s
      20) Alain Prost: 0.514s

      The first thing that strikes me about that list is that the data set appears to be very heavily biased in favour of drivers from more recent years – 40% of the drivers on the list are active drivers, and the data set as a whole seems to be skewed towards drivers who hit the peak of their careers in F1 within the last 10-15 years.

      Whilst it might be the case that increasing professionalism might have resulted in some bias towards more recent times, it comes across as a bit suspect that it is so heavily skewed towards current and very recent drivers – to the point where it suggests that whatever algorithm that was used to produce these results has been programmed in a way that potentially gives too high a weight to performances from modern drivers.

      1. anon, and it is not just the current drivers who show up disproportionately more, it is the teammates who are the overwhelming majority on this list.
        It seems the machine was leaning something like: Vettel as multi WDC was very good. Leclerc is beating him and should be higher. Ricciardo was his teammate and was within 3/100th of Vettel. Verstappen over the years beat Riccciardo and is 2/10th faster. Hulk was just a tad faster than Ricciardo. Sainz raced Verstappen during 2 years and was not that far behind. Norris is slightly faster than Sainz. etc. etc.
        Same for Schumacher, Barichello, Rosberg, Hamilton, Kovalainen, Bottas, Button, Alonso, Trulli, Fisichella.
        It doesn’t seem that the AWS machine didn’t take youngster development, or elderly deterioration, into account.
        They even had to build in a hard-break for any driver being more than 3 years out of the sport, otherwise Russell would have been at the top of this list by beating Kubica last year (and Latifi would be around #10).

        I think there is still an opportunity for a list like this, but it should say: Vettel (’12) was XYZ faster than Vettel (’19) or ABC faster than Button (’09), etc. etc.

        1. @coldfly it appears that the use of relative performance to their team mate was a major component of the analysis method used for this algorithm, so you are right that it does then also result in a disproportionately high number of paired drivers appearing in the list (e.g. Sainz and Norris).

          I have to agree that, the more you start looking at it, the more it looks like the problem is that the analysis is based on a comparatively limited set of metrics that then gives disproportionate weight to both modern drivers and result in too many drivers being paired up together.

    3. @robbie,

      I think it is so cool to see Max at 22 in the midst of a bunch of WDC’s and has never had a top car, unlike them.

      Why would the car matter?

      Just think how Kovalainen could be on that list. That’s by destroying people like Chandhok and Petrov in qualifying. Not by getting beaten by Hamilton.

      Likewise Verstappen is on that list because of Albon and Gasly.

      1. @f1osaurus I would think they are considering every qualifying rivalry from every race weekend the drivers have participated in, so in the case of Max much of his weighting is from his heading up DR on Saturdays as much as he did. DR is a high quality driver himself with more experience, so Max’s greater record against him, at his young age, elevated him in this ranking moreso than his handling of the lesser experienced Albon and Gasly. And Max has so far had more races with DR as his teammate than with Gasly and Albon combined, so no I highly doubt this study has taken Max’s beating of Gasly and Albon as enough to put him 4th overall.

        In the case of Kovalainen Vs Hamilton, while LH bested HK in their two seasons together it was particularly in 2008 that HK had pretty strong performances against LH and those would bode well for him wrt this study. Even if HK lagged behind it is about by how much that he lagged and not just by number of times he lagged, and it is about that HK was up against an eventual multiple WDC driver. The more success LH has had, the more that bodes well for those who have been his teammates, for the sake of this report.

      2. @robbie You stated:

        I think it is so cool to see Max at 22 in the midst of a bunch of WDC’s and has never had a top car, unlike them.

        So then I asked:

        Why would the car matter?

        Never mind I guess.

        1. @f1osaurus Why wouldn’t the car matter?

          1. @robbie Why would it?

            Just think how Kovalainen could be on that list. That’s by destroying people like Chandhok and Petrov in qualifying. Not by getting beaten by Hamilton.

            Kovalainen was destroyed by Hamilton, so he got his rating from driving at Lotus.

  59. Must be a slow news day………..

    1. Bunch of old….:)

  60. What do the times actually represent specifically? Do they represent the peak performance of a driver (ie their best ever single performance relative to each other), their expected performance (ie the expected gap in any given qualifying session) or something else?

    If it is the former looking at peak performance then I maybe able to reconcile these numbers (at least at the top of the list) but it would be a bit far-fetched to suggest that Senna would on average outqualify Hamilton by 0.275s and Schumacher outqualify Hamilton by 0.175s over a long period of time.

    Also in my experience “machine learning” is just a buzzword for we tried to do something fancy and probably reduces credibility of this study if anything.

    1. @chimaera2003 I think they are meant to represent average deltas to the best performing driver dictated by the calculation (in this case Senna). So they are saying that Hamilton would on average be 0.275 seconds behind him. So I agree, amongst other things that the deltas from Senna to Schumacher and Schumacher to Hamilton are much too large. In reality, if those three are the fastest qualifiers of the periods analysed (not an unreasonable top 3), I expect the gaps would be much closer, all three within a tenth of each other most likely.

      1. @keithedin I have tried to find a link to the complete study to confirm the delta meaning with no success. I feel that a clarification is important given what the study is trying to say.

        I heard that the difference between a good and great driver is 2 tenths (can’t find a source anywhere for this though), so on that basis the deltas are insane. If Lewis is an extraordinary driver (undoubted) then if my assertion holds then Senna must be an adjective that hasn’t even been invented!

  61. Jeffrey Powell
    18th August 2020, 15:52

    I didn’t know Jim had one girlfriend that did the official timing, this certainly would explain away those incredible performances. She must have had use of the chequerd flag and the television cameras as well, perhaps it was actually Sir Jackie he has always said Jimmy was the best, now we know why.

  62. I really can’t wait for AWS to stop spoiling the sport and coverage with their ridiculous stats. Always great to see a tire graphic showing 70% of tire left, while the driver is on the radio complaining his tyres are gone (eg Verstappen in Spain last weekend).

  63. To me its about dragging inferior cars to positions they shouldn’t be in that is more impressive than putting the best car on pole. That would be an interesting list if possible,d and i think Senna and Schumacher would still top the list but below them would look very different.

    1. F1oSaurus (@)
      19th August 2020, 9:00

      It’s about neither

  64. I have my doubts about the authenticity of these insights. Lets look at it from a layman’s perspective:
    Is it possible to differentiate mathematically the contribution of car and driver? Even if a number is achieved, is it constant as there are several external factors such as weather, mechanical gremlins, the equation between the driver and the teams and so on. Racing lines preference could also be another factor.
    If you see the speed traps, the fastest car is often not the best car. How can we therefore assume that one driver is faster than another unless its a pure spec series with all possible factors constant? The only way to properly assess is if each of the drivers took spend a minimum of 5 hotlaps in a car of the same make and tyres in one race track on one day under same conditions. Rest is all perception.

    1. I like that kind of comparisons between drivers, but its hard to say if they are completely right 😉
      There is one really interesting blog where the author claimed that he managed (with special system) to put drivers in different tiers, 100,00 meaning the fastest (tier 1), 100.1 means 0.1s slower (tier 2),…

      TIER 1
      Fangio 1950, Ascari 1951, Moss 1956, Surtees 1961, Clark 1963, Stewart 1968, Lauda 1975, Villeneuve G 1981, Piquet snr 1982, Prost 1983, Senna 1986, Schumacher M 1995, Raikkonen 2005, Alonso 2005, Hamilton 2008, Vettel 2011.

      TIER 2
      Rindt 1970, Andretti Mario 1970, Peterson 1973, Reutemann 1978, Berger 1989, Hakkinen 1998, Schumacher R 2004, Montoya 2004.

      What do you think?

      http://grandprixratings.blogspot.com/2014/06/formula-one-drivers-and-speed.html?m=1
      © Patrick O’Brien.

  65. Ambrogio Isgro
    18th August 2020, 16:46

    The most surprising thing is the absence of Hakkinen and Mansell from the list.

    1. Mansell was slightly better than Berger. Elio De Angelis smashed him when they were teammates. Mansell won only because of the car.

  66. Well it’s a talking point isn’t it, can’t agree with any top 5 that doesn’t feature Clarke and Fangio though.

    1. @f1bobby The data is from 1983 onwards ,if I saw it right (probably due to timing data limits until then?) so they are automatically not in the list of fastest from then on.

  67. this shows why smedley is working for nobody

  68. Amazon artificial consciousness trying to divide and conquer F1 community.

  69. This analysis looks really interesting. Too bad there is no link to the paper, so it remains unclear whether the effect of age and experience were taken into account, but overall, these results seem plausible.

    1. Ok, I found the link.

      Team mates had to have completed at least five qualifying sessions against one another before they were compared. Age was also factored in, while if a driver was coming back to the sport after three or more seasons out of F1, that was also considered.

      Good to see some F1 research, we need more of that. People are biased, models aren’t.

      1. @f1infigures Every algorithm result depends on weighed set of parameter set by human. It’s clear that certain age range result has been erased base on what Rob Smedley and Dean Locke believe as ‘prime age’. Because that the only explanation why Schumacher was so far ahead of Rosberg and why there’s no Mansell in the list.

        1. @ruliemaulana As far as I understand, the age effect is estimated as well by the model. Race-result analysis shows that drivers are best between roughly 30 and 35 years old, when they have the optimal combination of outright speed and experience. Their outright speed may peak in their 20s, after which it slowly declines. By the time a driver is in his 40s, his outright speed will have deteriorated considerably. However, there are not too many drivers older than 40 years in the dataset for comparison, which makes it harder to accurately assess Schumacher’s performance at the end of his career. This may have biased Rosberg’s ranking. Mansell’s ranking might be negatively affected as he partnered Piquet before the Brazilian’s horrible 1987 San Marino crash, after which he was never as fast as before. Something similar happened to Wendlinger and Panis, for example. In Panis’ case, this may have inflated Trulli’s ranking, and thereby indirectly Alonso’s and Hamilton’s as well.

          1. @f1infigures It’s scary if the parameter such as range of prime age is decided by the model. Skynet is near, man.

        2. @ruliemaulana I don’t think I understand this comment, but at the same time I don’t think you know anything about modeling. F1 metrics has a similar model for driver performance, you might want to check that out. It has a very detailed description of the data and the underlying assumptions.

        3. Of course Schumacher would be ahead. He was ancient.
          In Valencia2012, he became the oldest podium sitter since 1970. The same would be for his Monaco2012 pole. Even that tells you how age is important.

  70. Heikki faster than Vettel? Yeah very believable. Lol.
    Lewis and Max are the fastest of them all. Senna would be a tied Top 2.

    1. Is Rosberg also as fast as Schumacher and Senna?

  71. More meaningless statistics from AWS. Does this algorithm take into account car performance? Relative tyre performance (especially during tyre war eras)? Engine performance? Gaps to team mates? I could go on with a hundred different variables affecting a driver’s one lap pace. I see no way on earth that a computer algorithm could decisively prove these statistics. Nor would we want it to. To prove that Senna was fastest, ‘end of discussion’ means we should all go and do something more productive with our Sunday afternoons. Furthermore, endless statistics as provided by AWS takes all the uncertainty, human element and ultimately the fun out of watching a race.
    Amazon should stick to destroying the book industry and leave me to enjoy my F1!

  72. I’m genuinely confused why Kovalainen is on here, then again I didn’t really watch F1 between about 2000-2009 so missed his ‘best’ years I guess. On the random races I checked on wikipedia though, he seems to have been out qualified by LH most of the time? So yeah, still confused. Maybe someone can fill me in.

    1. It’s about qualifying speed only….Kov out qualified Trulli, Trulli out qualified Alonso

  73. This is where I would rank top 3 aswell. Hamilton is continuing to advance his career forward and should finish up as GOAT, but to me atleast he is not outright faster than Senna or more complete than Schumacher.

    Also Hamilton faced nothing like the competiton Senna faced in Prost.

    1. “Also Hamilton faced nothing like the competiton Senna faced in Prost.”

      I’d say 2007 Alonso v Ham is very similar

    2. It is pretty easy to advance with the most dominant car in the history after getting beaten by Button and Rosberg of all drivers.

  74. From my experience of machine learning/data mining, they will have basically taken a training data set of how drivers performed against all but there last one or two team mates, then tested it’s accuracy predicting known data they can compare the accuracy of for how they performed against their last team mate or two.

    So I’ve no doubt when given say Hamilton’s data against Alonso, Kovalainen, Button, and Rosberg, they’ve produced a predictive model that got very close to how he performs against Bottas. But I highly doubt that tasked with predicting a true unknown it’s all that accurate. It’s easy to tweak until you get something accurate from your training data set for a known test set where you can know if it’s accurate, but I doubt it holds up to scrutiny with real unknown data.

    Otherwise why don’t they tell us what the gaps will be for Alonso/Ocon, Leclerc/Sainz, and Ricciardo/Norris now and we’ll see how good their model is next year?

  75. We can’t know that for sure

    I mean he faced up against the only driver to consistently beat Schumacher and the only season Hamilton didn’t beat Rosberg he lost more points to events or reliability issues outside his control

    If Hamilton hadn’t moved to Mercedes we’d be debating if Rosbergs 4-6 titles put him up in the all time greats list

    It’s near impossible to conclude that Prost was a more worthy challenger than the 3 champions (Alonso, Button, and Rosberg) that Hamilton came out ahead more often than not against

    1. Meant as a reply to @jureo

    2. Schumacher was pretty even with Rosberg in 2012 despite being 42 years old (at least 10 years past his best).

      Quite an incredible achievement. I think it was 10-10 in qualifying in 2012. Rosberg beat Hamilton 11-8 in qualifying 2014.

      1. Pretty even??? Come on!!
        Rosberg is a damn fine driver and doesn’t get the recognition he deserves but in his battle with Schumacher he won comfortably in every metric. Schumacher was only gone for 3 years too. It wasn’t that long.

        1. Schumacher was in his 40’s after coming back from three years not racing.

          Someone like Alonso has continued to race at high levels during his break.

        2. Only for 3 years? He had to learn driving all over again with the new regulations.
          Can you say he was ONLY in his forties?
          If his age was not a handicap, how did he become the oldest to get to the podium since 1970?
          Even that proves that age was a huge disadvantage and he did well.
          Prime Schumacher would dominate Rosberg.

      2. F1oSaurus (@)
        19th August 2020, 9:06

        Rosberg qualified in front in about 3 out of 4 times. That’s not “even”.

        We’ll see how well Alonso does at a similar age I guess.

  76. I’d agree with the top 3…but in this order

    Senna fastest ever over 1 lap

    2nd Hamilton

    3rd Schumacher…..Schumacher had weaker teammates than Hamilton, and Schumacher wouldn’t share his data with his teammates. Rosberg, Bottas etc have all said they improved their speed by studying Hamilton’s data. If Hamilton had did what Schumacher did & refuse to share his data, he’d have beaten his teammates in qualy by larger margins

    1. Who said that he hid his data?
      What makes you think Irvine and Barrichello are weaker than Button, Bottas and Kovalainen?

      1. Well apparently Kovalainin is the 8th fastest driver ever to live😂, so Hamilton must be amazing to smash him as he did. 🤣
        Also Msc did have all kinds of advantages over team mates–Google his name and ” special tyres”
        We saw at 70th Ann GP just how important those are!

  77. I think that F1 metrics has the best model & thus ranking, with the caveat that no model is perfect:

    https://f1metrics.wordpress.com/2019/11/22/the-f1metrics-top-100/

  78. Another effort to “McDonaldise” F1 through oversimplified insights brought to you by Amazon (which you all know and trust).
    Results not worth commenting, for so many reasons…

  79. “Utilizing machine learning”? Done by Amazon??? This smells of “Garbage In, Garbage Out”.

  80. If Schumacher is only a tenth behind Senna in qualifying then he’s easily the best driver of the last 37 years because Schumacher was even stronger in the race than qualifying.

    Hamilton was outqualified 11-8 by Rosberg in 2014.

    1. Lewis is also the holder of the all time pole position record. So Yeh. ..he’s terrible/slow/clumsy etc etc

      1. He’s had the most dominant cars of all time for 7 straight seasons. No-one has ever had the advantages Hamilton has had.

        1. F1oSaurus (@)
          19th August 2020, 9:09

          Schumacher had a lot more advantages than Hamilton ever had. A bespoke car to his liking alone, contractual #2 drivers, FIArari helping out wherever they could, a huge budget advantage (aided by a $100 million annual bonus) which allowed them to almost constantly test. etc etc etc

          1. Car to his liking? If oversteery car is the way to go and his teammates cannot drive it, it is not Schumacher’s problem. They don’t have to design the car to make the gap smaller between drivers.
            Contractual #2 drivers like Bottas? Nr1 status is applied whenever needed.
            Schumacher was already was faster than them. That’d be great if they could be faster and steal points from his title rivals, just like he did in Sepang99.
            Or, do you think Ferrari were hampering them to make Schumacher look good despite paying him 10 times more?
            In fact, he gifted his teammates more wins than they did.
            I hope you don’t think
            FIArrari helping? They banned front wing attachment in 2006, xWings in 1998, you know 2003 and 2005.
            FIA made him lose almost 40 points in 1994 with the silliest penalties ever.
            What exactly do you mean by help?

      2. Pole record doesn’t mean anything.

        Rosberg won 29 poles against Hamilton’s 35 during their time. had Rosberg stayed, even he could pass Schumacher’s pole count.
        Vettel also has much more poles than Verstappen and Ricciardo combined but I don’t think it would still be that way if they team up next year.

        Hamilton and his teammates have had 76 front row lockouts so far compared to 23 of Schumacher and his.
        So, Hamilton’s pole record only means that Mercedes was super dominant.

  81. Also keep in mind that Verstappen is still 22 and will only improve so will pass Hamilton on that list quite easily.

    That 0.05 gap to Hamilton includes a lot of races as a very raw 17, 18 year old trying to learn the ropes of motor racing let alone F1.

    Right now Verstappen would blow away Bottas in qualifying 6-0 not 4-2.

    1. Ifs and buts. Yet again VER comes up second to Lewis (even in its laughable list)

      1. Despite being 22.

        Hamilton at his peak was outqualified by Rosberg 11-8 in 2014 he’s not beating Verstappen.

        1. Even in the ifs and buts study that you promote constantly he still comes second behind the ifs and buts champion; Alonso.

          1. Hamilton was the ifs and buts champion at the end of 2013.

            Got the clear best car on the grid for 7 consecutive seasons yet only just now approaching Schumacher’s records.

          2. What 7 consecutive seasons?

            Hamilton had the best car only in 2014 and 2015.

            2007 and 2008 Ferrari had the better car
            2009 Red Bull had the best car but Vettel threw it all away
            2010-2013 Red Bull had the better car
            In 2016 Hamilton’s car broke down a lot more than Rosberg’s so Rosberg picked up that title.
            2017 Ferrari car was better, but Vettel threw it all away
            2018 Ferrari car was a lot better, but Vettel threw it all away
            2019 Mercedes and Ferrari were equal but Vettel and Leclerc threw it all away. Plus Verstappen threw races like Hungary, Monaco and Mexico away.

    2. F1oSaurus (@)
      19th August 2020, 9:12

      Verstappen was blown away by Ricciardo in 2016 and they were equal in 2017. He’s only on that list because Red Bull create a car only for Verstappen’s driving stule and Gasly and Albon cannot drive with it. Which creates a huge gap making Verstappen look good.

      1. Verstappen was an 18 year old in 2016. 2017 Verstappen established his superiority just had more bad luck and made some silly mistakes, but the raw speed advantage was there.

        Ricciardo ran from a fight in the end.

        1. Geez, so you claim he was better, but then whe he wasn’t you come up with excuses.

          Ricciardo ran from unfair treatment yes. With Verstappen clearly unable to lead the car development in the right direction that cost Red Bull a lot …

      2. @f1osaurus Wrong. DR beat Max 11-6 in quali in 2016, but Max didn’t have the luxury of starting off the season with RBR nor testing with them. Max started with them in race 5 in Spain and won that race. But anyway this is about qualifying performances and I fail to see how Max beating DR 13-7 in 2017 makes them equal other than in your world. Then in 2018 it became 15-5 in favour of Max in quali. No, Max is on the list far more because of his domination of DR in quali over the nearly three full seasons together. The study would not be putting much weight into Max beating two teammates that he should have been beating anyway. If more experienced DR had stood up better against Max then perhaps Max would be lower on the list, and DR would be higher.

        1. @robbie So you mean Right? Indeed Ricciardo destroyed Verstappen in 2016. For 2017 there is no point in taking all races since there were issues on both sides with drivers not finishing qualifying. Their lap times would be very close (if both managed to get a lap in Q3)

          But yeah, I’m sure you make the same mistake as the algorithm does. Just look at the scoreboard and ignore external factors influencing results.

          1. @f1osaurus Rich of your to accuse me of ignoring things when you don’t seem to want to acknowledge that Max wasn’t at RBR in 2016 for pre-season learning and only started at the fifth race, which he won. Then you’ve now decided conveniently to ignore 2017 because of them ‘not finishing qualifying?’ Huh? Nice try at making excuses for DR but you’re fooling nobody.

            I’m sure you don’t even know what all went into the algorithm, but you will conveniently claim they didn’t consider this or that. And you yourself are the master at ignoring external factors at your convenience and when it best suits your argument. DR ‘destroyed’ Max, end of. DR ‘destroyed’ SV end of. But you never touch on how Kvyat destroyed DR in 2015.

          2. @robbie So what? He was destroyed by Ricciardo.

            Alonso did not destroy Hamilton in 2007.

            Sigh, Ricciardo destroyed Kvyat in 2015. Are you seriously kidding?

            Are you just being dumb for a reason? Is this that your joke? You think being that stupid is funny? It is not. It’s just incredibly tiresome that you always want to claim the opposite of what the truth

            But never mind, I will not react to your dumb bs anymore. There is just no point. you will never learn.

          3. @f1osaurus Right back at ya! When you’ve got nothing you resort to the insults as usual.

      3. Are you a Hamilton fan? Your entire argument revolves around the complete hoaxes like “Verstappen can beat his teammates just because he likes the car, Schumacher sabotaged his teammates, they let him pass all time, Button is an all time great but Barrichello sucks, Ferrari paid Schumacher 10 times more money for nothing, He hid his telemetry, etc”

  82. This list made sense till I got to Heikki Kovalainen.

  83. Anything Rob Smedley promotes has to be questionable!

    1. Just because Schumacher is ahead of Hamilton?
      Maybe not doing so great (42-36 quali, 35-29 pole, which also proves that his pole record is down to the car) against Rosberg of all drivers just didn’t help Hamilton’s stats.

  84. Totally flawed, Kovalainan 8th, is a joke. All the listed drivers are from the 2000 apart from Senna but there was other great drivers around at that time. It’s too early to place Max in fourth place.

  85. I think that F1 Metrics has the best model & thus ranking, with the caveat that no model is perfect.

    1. Yeah, f1metrics is getting better and better. In the previous versions, Rosberg was 6th or something. Now, it is getting more realistic.

  86. and after some seriously good candidates, there you have it: the most pointless, useless, stupid and unnecessary thing of 2020.

  87. Next, apples and oranges brought to you by AWS

Comments are closed.