planetf1.com

It is currently Fri Jul 19, 2019 8:09 am

All times are UTC


Forum rules


Please read the forum rules



Post new topic Reply to topic
Author Message
PostPosted: Tue Jun 25, 2019 6:34 am 
Offline
User avatar

Joined: Thu Jan 08, 2015 7:54 am
Posts: 2510
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 6:43 am 
Offline
User avatar

Joined: Thu Feb 13, 2003 9:39 pm
Posts: 3440
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

And also Hill is faster than Villeneuve, Villeneuve is faster than Frentzen, Frentzen is faster than Hill.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 6:58 am 
Offline

Joined: Tue Dec 23, 2014 1:04 pm
Posts: 786
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.


Yes it is absurd.
If it was the way you say it is, VER would have had at least 2 poles by now instead of 0.

If you reversed the role, so HAM in RB I am sure Lewis would have socred at least 2 pole positions.

And cross comparison doesn't work because the vs performances in the same team depend on too many variables that don't stay constant for example:
- psychology
- team dynamics
- tyres
- brakes/suspension geometry(an area where HAM has lost massively when they introduced the Rear Brake by Wire and the FRIC) ).

About Bottas,
Valterri is the first teammate who can match Lewis where he has always made most of his gaps in Q3 compared to his previous teammates(ALO, KOV, BUT, ROS): The first heavy braking zone after the main straight. Strange eh? altough to be fair, Mercedes is certainly not the car to have under braking(RB and Ferrari are the best here).


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 11:46 am 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:00 pm 
Online

Joined: Wed May 27, 2015 7:11 pm
Posts: 760
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:23 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:34 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

I'm wondering how any of that computes. I would hope you're not using something as unreliable as average gap in qualifying; a metric which will always be completely thrown off by outlier performances.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:39 pm 
Offline

Joined: Mon Nov 26, 2018 12:53 am
Posts: 257
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s



I remember Max coming to my attention as a stand-out performer pretty much immediately from his debut. Right off the bat he was doing Alonso/Minardi type things albeit with better relative equipment than the 2001 Minardi so he was scoring points and racing high up in the midfield.

It wouldn't surprise me if Max was a little bit quicker than Hamilton, he is that good.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:40 pm 
Online

Joined: Wed May 27, 2015 7:11 pm
Posts: 760
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:
If it's based on qualifying gaps then the Button Perez data makes no sense.
Perez out qualified Button 10-9 but his gap on average was only 0.19 sec. Button on his 9 out qualifying laps was on average 0.56 up.
Therefore Button is 0.37 up on Perez not equal to. (rough figures)


Last edited by Laz_T800 on Tue Jun 25, 2019 1:02 pm, edited 1 time in total.

Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:49 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

I'm wondering how any of that computes. I would hope you're not using something as unreliable as average gap in qualifying; a metric which will always be completely thrown off by outlier performances.

That's what I use but I ensure that drivers have fair chance to set a time and don't include sessions were 1 driver didn't compete in or wet sessions.

Like you I'm a Hamilton fan but primarily I'm a numbers guy, for the record it shows Hamilton to be quicker than Alonso and Vettel so it's not all bad, as for the comparison shown it's very much incomplete so not entirely accurate for example if we just took the first 8 races of each season then Hamilton's gap to Bottas would be 0.06s, however the overall 2017/2018 gap is 0.18s.

This is purely a qualifying comparison to try and ascertain the out and out fastest driver but not necessarily the best driver, it doesn't include race pace, race craft or wet weather driving.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:52 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
Laz_T800 wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:

That's your opinion but I find there's strong correlation to who the fastest drivers actually are.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 12:59 pm 
Offline

Joined: Sat Jan 28, 2012 4:13 pm
Posts: 15265
Laz_T800 wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:


It's pretty relevant TBH. If you add a little nuance and take out first and last season the cross team mate comparison is generally a good predictor. Obviously not fool proof though.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 1:03 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

I'm wondering how any of that computes. I would hope you're not using something as unreliable as average gap in qualifying; a metric which will always be completely thrown off by outlier performances.

That's what I use but I ensure that drivers have fair chance to set a time and don't include sessions were 1 driver didn't compete in or wet sessions.

Like you I'm a Hamilton fan but primarily I'm a numbers guy, for the record it shows Hamilton to be quicker than Alonso and Vettel so it's not all bad, as for the comparison shown it's very much incomplete so not entirely accurate for example if we just took the first 8 races of each season then Hamilton's gap to Bottas would be 0.06s, however the overall 2017/2018 gap is 0.18s.

This is purely a qualifying comparison to try and ascertain the out and out fastest driver but not necessarily the best driver, it doesn't include race pace, race craft or wet weather driving.

Sigh, not everyone thinks like you poker. My problem is not about what it shows; it's about the lack of reliability. That metric probably shows Vettel to be much slower than Ricciardo for example and if you were to add that to your model it would probably suggest that Max is half a second faster than anyone. I'm assuming that's why you chose not to include that matchup?

The transitive property is somewhat effective but not when you are measuring the actual size of the margin. Even between teammates, that can fluctuate quite a bit. We saw significant changes to the average margin between Raikkonen and Vettel from year to year and we have seen that with other drivers as well. When you jump around too much between match-ups and years; it breaks down completely.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 1:04 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
mikeyg123 wrote:
Laz_T800 wrote:
pokerman wrote:
Laz_T800 wrote:
I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:


It's pretty relevant TBH. If you add a little nuance and take out first and last season the cross team mate comparison is generally a good predictor. Obviously not fool proof though.

The first season being the rookie seasons in F1?

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 1:12 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
Laz_T800 wrote:
I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

I'm wondering how any of that computes. I would hope you're not using something as unreliable as average gap in qualifying; a metric which will always be completely thrown off by outlier performances.

That's what I use but I ensure that drivers have fair chance to set a time and don't include sessions were 1 driver didn't compete in or wet sessions.

Like you I'm a Hamilton fan but primarily I'm a numbers guy, for the record it shows Hamilton to be quicker than Alonso and Vettel so it's not all bad, as for the comparison shown it's very much incomplete so not entirely accurate for example if we just took the first 8 races of each season then Hamilton's gap to Bottas would be 0.06s, however the overall 2017/2018 gap is 0.18s.

This is purely a qualifying comparison to try and ascertain the out and out fastest driver but not necessarily the best driver, it doesn't include race pace, race craft or wet weather driving.

Sigh, not everyone thinks like you poker. My problem is not about what it shows; it's about the lack of reliability. That metric probably shows Vettel to be much slower than Ricciardo for example and if you were to add that to your model it would probably suggest that Max is half a second faster than anyone. I'm assuming that's why you chose not to include that matchup?

The transitive property is somewhat effective but not when you are measuring the actual size of the margin. Even between teammates, that can fluctuate quite a bit. We saw significant changes to the average margin between Raikkonen and Vettel from year to year and we have seen that with other drivers as well. When you jump around too much between match-ups and years; it breaks down completely.

How is Verstappen half a second faster than everyone when he was only 0.17s quicker than Ricciardo, I also said the model was incomplete, I was just showing an example to a poster what was behind KingVoid's thinking, also it's life time performance not merely short term performance.

Edit: With respect to Verstappen how can you be sure to pigeon hole just how fast he might be?

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Last edited by pokerman on Tue Jun 25, 2019 1:17 pm, edited 1 time in total.

Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 1:15 pm 
Offline

Joined: Sat Jan 28, 2012 4:13 pm
Posts: 15265
pokerman wrote:
mikeyg123 wrote:
Laz_T800 wrote:
pokerman wrote:
Laz_T800 wrote:
I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:


It's pretty relevant TBH. If you add a little nuance and take out first and last season the cross team mate comparison is generally a good predictor. Obviously not fool proof though.

The first season being the rookie seasons in F1?


Yes sorry,

I mean rookie season and final season of a drivers career.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 1:40 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
Laz_T800 wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:
If it's based on qualifying gaps then the Button Perez data makes no sense.
Perez out qualified Button 10-9 but his gap on average was only 0.19 sec. Button on his 9 out qualifying laps was on average 0.56 up.
Therefore Button is 0.37 up on Perez not equal to. (rough figures)

7-6 to Perez but with Button having a better average of 0.06s, I tip you just took 5 minutes to get all that off a website that just throws everything into the mix whereas I discount wet sessions and were drivers had to be in the same sessions when they set their times, so basically I throw out all the junk, in Australia because of the wet qualifying Button managed to post a time over 9 seconds quicker than Perez, this was included in the website you visited seeing as there was 19 races that year.

I take hours compiling these stats, you spent 5 minutes on a website to make fun of my results.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 1:42 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
mikeyg123 wrote:
pokerman wrote:
mikeyg123 wrote:
Laz_T800 wrote:
pokerman wrote:
It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s


Ok, so no relevant data then. :lol:


It's pretty relevant TBH. If you add a little nuance and take out first and last season the cross team mate comparison is generally a good predictor. Obviously not fool proof though.

The first season being the rookie seasons in F1?


Yes sorry,

I mean rookie season and final season of a drivers career.

Yeah I take that into account as well, not so much final seasons but definitely rookie seasons.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 2:33 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
That's what I use but I ensure that drivers have fair chance to set a time and don't include sessions were 1 driver didn't compete in or wet sessions.

Like you I'm a Hamilton fan but primarily I'm a numbers guy, for the record it shows Hamilton to be quicker than Alonso and Vettel so it's not all bad, as for the comparison shown it's very much incomplete so not entirely accurate for example if we just took the first 8 races of each season then Hamilton's gap to Bottas would be 0.06s, however the overall 2017/2018 gap is 0.18s.

This is purely a qualifying comparison to try and ascertain the out and out fastest driver but not necessarily the best driver, it doesn't include race pace, race craft or wet weather driving.

Sigh, not everyone thinks like you poker. My problem is not about what it shows; it's about the lack of reliability. That metric probably shows Vettel to be much slower than Ricciardo for example and if you were to add that to your model it would probably suggest that Max is half a second faster than anyone. I'm assuming that's why you chose not to include that matchup?

The transitive property is somewhat effective but not when you are measuring the actual size of the margin. Even between teammates, that can fluctuate quite a bit. We saw significant changes to the average margin between Raikkonen and Vettel from year to year and we have seen that with other drivers as well. When you jump around too much between match-ups and years; it breaks down completely.

How is Verstappen half a second faster than everyone when he was only 0.17s quicker than Ricciardo, I also said the model was incomplete, I was just showing an example to a poster what was behind KingVoid's thinking, also it's life time performance not merely short term performance.

Edit: With respect to Verstappen how can you be sure to pigeon hole just how fast he might be?

You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 4:07 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
KingVoid wrote:
I think Verstappen is faster and a bigger talent than Alonso/Hamilton. This is partially gut feeling and partially based on cross comparison data.

To give you an idea of just how talented Verstappen is, he was already one of the best drivers on the grid in 2017 when he was still only a teenager. Has there ever been any other 19 year old driver who could make any case for being the best/fastest driver on the grid?

You're welcome to your own opinion but I think the first part is important. You may think that he's faster or a bigger talent but he still has to prove that. He has to establish that for himself. People's thoughts (whether supporting or refuting your view) are irrelevant.

What's clear is that he is the fastest driver to come along since Hamilton and is easily among the best in F1 already. Most polls rank him second best driver in the sport already at just 21 years of age. So there can be no denying that he has come farther at a younger age than anyone before him. The important stuff is yet to even begin though. He hasn't fought for a championship yet and he hasn't ever been in a situation where he's had much to lose. he's able to attack and race carefree most of the time and he hasn't had to balance risk/reward in order to maximize a WDC challenge yet. He hasn't faced that kind of pressure yet of needing to deliver week in and week out with the title resting in the balance.

Next year he will reach the age that Hamilton and Vettel were when they first began fighting for championships. The question is; will he have the car to do that himself? If not, will he leave RBR? If he does have the car; will he handle the situation well? I think it's always easy to talk about someone's potential or what you think they could or would do in a given situation. The hard part is actually doing it.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 6:53 pm 
Offline

Joined: Tue Dec 23, 2014 1:04 pm
Posts: 786
KingVoid wrote:
I think Verstappen is faster and a bigger talent than Alonso/Hamilton. This is partially gut feeling and partially based on cross comparison data.

To give you an idea of just how talented Verstappen is, he was already one of the best drivers on the grid in 2017 when he was still only a teenager. Has there ever been any other 19 year old driver who could make any case for being the best/fastest driver on the grid?


Just like with Alonso, with Verstappen it is all talk and he needs to do facts very soon.

The biggest thing going for him is his age. Stop for a moment and think about Hamilton first season. A season where he scored 6(SIX) pole positions against the defending 2 times WDC at the time.

Yet we have a driver who is still to score his FIRST pole position called the best driver.

At no point in time was HAM called the best, not at the beginning of his career nor now that he has achieved everything despite doing stuff VER is yet to ever think about.

F1 is all about hype, self marketing, potential and having the right looks.


Last edited by Pullrod on Tue Jun 25, 2019 7:06 pm, edited 2 times in total.

Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 7:05 pm 
Offline
User avatar

Joined: Sun Feb 23, 2014 3:53 am
Posts: 7012
Location: Michigan, USA
sandman1347 wrote:
You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

:thumbup:

It's a good enough way to get an idea of the comparison between teammates, but you lose me at the point you claim you can use a season or two of data and produce a two-decimal result for drivers who weren't directly involved. It's not that accurate.

_________________
PICK 10 COMPETITION (4 wins, 15 podiums): 3rd in 2016
TOP THREE CHAMPIONSHIP (No Limit Excedrin Racing): Champions in 2015 & 2018 | 2nd in 2017
AUTOSPORT GP PREDICTOR: 2017 USA & P-F1 Champion | #2 in the world in 2017


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 8:59 pm 
Offline
User avatar

Joined: Thu Jan 08, 2015 7:54 am
Posts: 2510
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

One quick sidenote: Hulkenberg outqualified Perez 35-24 and was actually 0.09s quicker on average, which makes Verstappen look even quicker. But that’s only a minor issue.

There’s also this one:

Hamilton > Alonso 0.07s
Alonso > Raikkonen 0.29s
Vettel > Raikkonen 0.22s
Ricciardo > Vettel 0.21s
Verstappen > Ricciardo 0.17s

Which again adds up to Verstappen having 2 tenths over Hamilton, and unlike the Hulkenberg/Ricciardo cross comparison, we have at least one full season of data for each of the above.

Again, I’m not saying that cross comparisons are a perfect measure, but they do give some idea. The limited data on Verstappen suggests that he’s an absolute monster.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 9:21 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
KingVoid wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

One quick sidenote: Hulkenberg outqualified Perez 35-24 and was actually 0.09s quicker on average, which makes Verstappen look even quicker. But that’s only a minor issue.

There’s also this one:

Hamilton > Alonso 0.07s
Alonso > Raikkonen 0.29s
Vettel > Raikkonen 0.22s
Ricciardo > Vettel 0.21s
Verstappen > Ricciardo 0.17s

Which again adds up to Verstappen having 2 tenths over Hamilton, and unlike the Hulkenberg/Ricciardo cross comparison, we have at least one full season of data for each of the above.

Again, I’m not saying that cross comparisons are a perfect measure, but they do give some idea. The limited data on Verstappen suggests that he’s an absolute monster.

The comparisons as far as "<" or ">" are useful. The comparisons as far as the actual numbers are not for the reasons explained in my above post. Even if we look at your post; you are using data from Hamilton's rookie year as your baseline. All drivers improve speed-wise from their rookie years.

The bottom line is that you cannot know how Verstappen would match up with Hamilton or how he would match up with Alonso. If you feel that 2014 was representative, you can be fairly confident that he would have the measure of Vettel but you yourself have often argued that, for whatever reason, 2014 was not representative between Vettel and Ricciardo. The figures you are using are not reliable in the way that you're using them.


Top
 Profile  
 
PostPosted: Tue Jun 25, 2019 11:47 pm 
Offline
User avatar

Joined: Thu Feb 13, 2003 9:39 pm
Posts: 3440
KingVoid wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

One quick sidenote: Hulkenberg outqualified Perez 35-24 and was actually 0.09s quicker on average, which makes Verstappen look even quicker. But that’s only a minor issue.

There’s also this one:

Hamilton > Alonso 0.07s
Alonso > Raikkonen 0.29s
Vettel > Raikkonen 0.22s
Ricciardo > Vettel 0.21s
Verstappen > Ricciardo 0.17s

Which again adds up to Verstappen having 2 tenths over Hamilton, and unlike the Hulkenberg/Ricciardo cross comparison, we have at least one full season of data for each of the above.

Again, I’m not saying that cross comparisons are a perfect measure, but they do give some idea. The limited data on Verstappen suggests that he’s an absolute monster.

This is all very nice, but it's ultimately got very little meaning in reality. For one thing, even if the fundamental idea behind the system had merit - the tracks are all different lengths, with different error margins so you can't just average up the time deltas and use them with any meaning. A slightly better approach would be to go and use percentage time, but that only marginally improves on an already very flawed system.

Different cars suit different drivers, different tyres suit different drivers, different qualifying sessions have different circumstances, the data is all very very very woolly. The best you can do is look at the overall scoreline between drivers and if one has consistently out qualified the other with equal machinery then you can say with a high degree of confidence that that driver is a faster driver than the other one.

What you can't go and do is start adding together times from different circuits in different seasons and say it represents the actual speed differences.

Yes, Max is fast. But until we see him go up against Hamilton in the same car over the course of a season it's impossible to say who is faster.


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 1:24 am 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
Laz_T800 wrote:
pokerman wrote:
Laz_T800 wrote:
mikeyg123 wrote:
It's pretty relevant TBH. If you add a little nuance and take out first and last season the cross team mate comparison is generally a good predictor. Obviously not fool proof though.


But it's all based on the Button/Perez season being seen as equal.
1/ they only had one season together
2/ They weren't equal!

With #1 I take your point, with #2 I basically just rubbished your post.


Nice guy. :lol:

Well it did lack a bit of effort on your part. :)

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Last edited by pokerman on Wed Jun 26, 2019 1:58 am, edited 1 time in total.

Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 1:34 am 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
That's what I use but I ensure that drivers have fair chance to set a time and don't include sessions were 1 driver didn't compete in or wet sessions.

Like you I'm a Hamilton fan but primarily I'm a numbers guy, for the record it shows Hamilton to be quicker than Alonso and Vettel so it's not all bad, as for the comparison shown it's very much incomplete so not entirely accurate for example if we just took the first 8 races of each season then Hamilton's gap to Bottas would be 0.06s, however the overall 2017/2018 gap is 0.18s.

This is purely a qualifying comparison to try and ascertain the out and out fastest driver but not necessarily the best driver, it doesn't include race pace, race craft or wet weather driving.

Sigh, not everyone thinks like you poker. My problem is not about what it shows; it's about the lack of reliability. That metric probably shows Vettel to be much slower than Ricciardo for example and if you were to add that to your model it would probably suggest that Max is half a second faster than anyone. I'm assuming that's why you chose not to include that matchup?

The transitive property is somewhat effective but not when you are measuring the actual size of the margin. Even between teammates, that can fluctuate quite a bit. We saw significant changes to the average margin between Raikkonen and Vettel from year to year and we have seen that with other drivers as well. When you jump around too much between match-ups and years; it breaks down completely.

How is Verstappen half a second faster than everyone when he was only 0.17s quicker than Ricciardo, I also said the model was incomplete, I was just showing an example to a poster what was behind KingVoid's thinking, also it's life time performance not merely short term performance.

Edit: With respect to Verstappen how can you be sure to pigeon hole just how fast he might be?

You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 1:51 am 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
KingVoid wrote:
pokerman wrote:
Laz_T800 wrote:
pokerman wrote:
KingVoid wrote:
How fast is Verstappen relative to Hamilton?

The Ricciardo/Hulkenberg and Ricciardo/Vettel cross comparisons basically suggest that Verstappen is around 2 tenths quicker than even Hamilton/Alonso tier

I know sounds absurd, but that is what the numbers say.

Yeah I'm aware of that, it's still too early to make those defining conclusion though, Ricciardo out qualifying the Hulk by 9 tenths in Baku is maybe a bit of an outlier that overly weights things towards Ricciardo at this moment in time considering we only have 7 sample cases?

One thing I would say is that it would be hard to make a case for Hamilton being quicker than Verstappen but then again I believe Verstappen will end up as one of the greats of the sport anyway, which we kind of see with him being 5 tenths quicker than Gasly.

It has to be said though that with things that I've read a decision was taken not to go up against Hamilton at Mercedes.


I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

One quick sidenote: Hulkenberg outqualified Perez 35-24 and was actually 0.09s quicker on average, which makes Verstappen look even quicker. But that’s only a minor issue.

There’s also this one:

Hamilton > Alonso 0.07s
Alonso > Raikkonen 0.29s
Vettel > Raikkonen 0.22s
Ricciardo > Vettel 0.21s
Verstappen > Ricciardo 0.17s

Which again adds up to Verstappen having 2 tenths over Hamilton, and unlike the Hulkenberg/Ricciardo cross comparison, we have at least one full season of data for each of the above.

Again, I’m not saying that cross comparisons are a perfect measure, but they do give some idea. The limited data on Verstappen suggests that he’s an absolute monster.

Your figures are similar to mine but with the Hulk I have him consistently 0.04s quicker than Perez season on season which puts him close to equal to Button.

I have to reiterate again that with the Hulk and Ricciardo we have only a sample of 7 qualifying sessions, that's too small a sample size to use to be jumping on the gun about just how quick Verstappen is, also Ricciardo beat the Hulk by 0.9s in Baku so the actual gap presently might be slightly inflated?

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 1:57 am 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
KingVoid wrote:
pokerman wrote:
Laz_T800 wrote:
I don't usually see the posts of the user you quoted.
Where does the logic come from that suggests Verstappen is 2 tenths faster than Hamilton?
Genuinely interested as I can't quite fathom how that can be computed by any kind of relevant data that exists?

It's multiple cross references between drivers but at this point far from conclusive with Ricciardo having done only 7 qualifying sessions with the Hulk one of which Ricciardo was 0.9s quicker which he may never repeat again, however in simplistic form:-

Hamilton > Button 0.2s
Button = Perez = Hulkenberg
Ricciardo > Hulkenberg 0.21s
Verstappen > Ricciardo 0.17s

One quick sidenote: Hulkenberg outqualified Perez 35-24 and was actually 0.09s quicker on average, which makes Verstappen look even quicker. But that’s only a minor issue.

There’s also this one:

Hamilton > Alonso 0.07s
Alonso > Raikkonen 0.29s
Vettel > Raikkonen 0.22s
Ricciardo > Vettel 0.21s
Verstappen > Ricciardo 0.17s

Which again adds up to Verstappen having 2 tenths over Hamilton, and unlike the Hulkenberg/Ricciardo cross comparison, we have at least one full season of data for each of the above.

Again, I’m not saying that cross comparisons are a perfect measure, but they do give some idea. The limited data on Verstappen suggests that he’s an absolute monster.

The comparisons as far as "<" or ">" are useful. The comparisons as far as the actual numbers are not for the reasons explained in my above post. Even if we look at your post; you are using data from Hamilton's rookie year as your baseline. All drivers improve speed-wise from their rookie years.

The bottom line is that you cannot know how Verstappen would match up with Hamilton or how he would match up with Alonso. If you feel that 2014 was representative, you can be fairly confident that he would have the measure of Vettel but you yourself have often argued that, for whatever reason, 2014 was not representative between Vettel and Ricciardo. The figures you are using are not reliable in the way that you're using them.

I made a post before the season started that for Ricciardo to fully validate the 2014 season he would have to beat the Hulk by 0.23s, he presently is beating him by 0.21s.

With the 0.07s figure between Hamilton and Alonso you can actually get there by cross comparison with Button.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Last edited by pokerman on Wed Jun 26, 2019 2:01 am, edited 1 time in total.

Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 1:57 am 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
That's what I use but I ensure that drivers have fair chance to set a time and don't include sessions were 1 driver didn't compete in or wet sessions.

Like you I'm a Hamilton fan but primarily I'm a numbers guy, for the record it shows Hamilton to be quicker than Alonso and Vettel so it's not all bad, as for the comparison shown it's very much incomplete so not entirely accurate for example if we just took the first 8 races of each season then Hamilton's gap to Bottas would be 0.06s, however the overall 2017/2018 gap is 0.18s.

This is purely a qualifying comparison to try and ascertain the out and out fastest driver but not necessarily the best driver, it doesn't include race pace, race craft or wet weather driving.

Sigh, not everyone thinks like you poker. My problem is not about what it shows; it's about the lack of reliability. That metric probably shows Vettel to be much slower than Ricciardo for example and if you were to add that to your model it would probably suggest that Max is half a second faster than anyone. I'm assuming that's why you chose not to include that matchup?

The transitive property is somewhat effective but not when you are measuring the actual size of the margin. Even between teammates, that can fluctuate quite a bit. We saw significant changes to the average margin between Raikkonen and Vettel from year to year and we have seen that with other drivers as well. When you jump around too much between match-ups and years; it breaks down completely.

How is Verstappen half a second faster than everyone when he was only 0.17s quicker than Ricciardo, I also said the model was incomplete, I was just showing an example to a poster what was behind KingVoid's thinking, also it's life time performance not merely short term performance.

Edit: With respect to Verstappen how can you be sure to pigeon hole just how fast he might be?

You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 2:14 am 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
Sigh, not everyone thinks like you poker. My problem is not about what it shows; it's about the lack of reliability. That metric probably shows Vettel to be much slower than Ricciardo for example and if you were to add that to your model it would probably suggest that Max is half a second faster than anyone. I'm assuming that's why you chose not to include that matchup?

The transitive property is somewhat effective but not when you are measuring the actual size of the margin. Even between teammates, that can fluctuate quite a bit. We saw significant changes to the average margin between Raikkonen and Vettel from year to year and we have seen that with other drivers as well. When you jump around too much between match-ups and years; it breaks down completely.

How is Verstappen half a second faster than everyone when he was only 0.17s quicker than Ricciardo, I also said the model was incomplete, I was just showing an example to a poster what was behind KingVoid's thinking, also it's life time performance not merely short term performance.

Edit: With respect to Verstappen how can you be sure to pigeon hole just how fast he might be?

You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.

That's all well and good if I wasn't finding that I could actually start to see some correlation like for instance:-

Alonso > Massa 0.27s
Bottas > Massa 0.17s
Therefore
Alonso > Bottas 0.1s

Hamilton > Alonso 0.07s
Hamilton > Bottas 0.17s

That's off the top of my head, I will come back and check tomorrow but I know I'm at max only 0.01s out.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Last edited by pokerman on Wed Jun 26, 2019 11:53 am, edited 1 time in total.

Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 2:32 am 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
How is Verstappen half a second faster than everyone when he was only 0.17s quicker than Ricciardo, I also said the model was incomplete, I was just showing an example to a poster what was behind KingVoid's thinking, also it's life time performance not merely short term performance.

Edit: With respect to Verstappen how can you be sure to pigeon hole just how fast he might be?

You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.

That's all well and good if I wasn't finding that I could actually start to see some correlation like for instance:-

Alonso > Massa 0.27s
Bottas > Massa 0.17s
Therefore
Alonso > Bottas 0.1s

Hamilton > Alonso 0.07s
Hamilton > Bottas 0.17s

That's off the top of my hand, I will come back and check tomorrow but I know I'm at max only 0.01s out.

The qualifying system when Hamilton and Alonso were teammates was one that had drivers using different fuel loads in Q3 and Alonso had the lighter load 11 times to Hamilton's 6. That was also Hamilton's rookie season (not a reliable year to use to judge his pace in his prime). The Hamilton Bottas gap will likely go up as you have just gotten through the part of the season where Bottas is closest to Hamilton and have yet to get through the part where he is furthest away.

Regardless, through sheer probability it's certainly likely that your method will work sometimes but there will be other times when it will not work. Use your method to explain why Hulkeberg is slower than Ricciardo but faster than Sainz. Use your method to explain why Rubens was slower than Button but faster than Hulkenberg. The bottom line poker is that it is not a reliable method. If you can point out even one exception, a method becomes unreliable but your method has many many exceptions.


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 5:43 am 
Offline

Joined: Sat Jan 28, 2012 4:13 pm
Posts: 15265
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.

That's all well and good if I wasn't finding that I could actually start to see some correlation like for instance:-

Alonso > Massa 0.27s
Bottas > Massa 0.17s
Therefore
Alonso > Bottas 0.1s

Hamilton > Alonso 0.07s
Hamilton > Bottas 0.17s

That's off the top of my hand, I will come back and check tomorrow but I know I'm at max only 0.01s out.

The qualifying system when Hamilton and Alonso were teammates was one that had drivers using different fuel loads in Q3 and Alonso had the lighter load 11 times to Hamilton's 6. That was also Hamilton's rookie season (not a reliable year to use to judge his pace in his prime). The Hamilton Bottas gap will likely go up as you have just gotten through the part of the season where Bottas is closest to Hamilton and have yet to get through the part where he is furthest away.

Regardless, through sheer probability it's certainly likely that your method will work sometimes but there will be other times when it will not work. Use your method to explain why Hulkeberg is slower than Ricciardo but faster than Sainz. Use your method to explain why Rubens was slower than Button but faster than Hulkenberg. The bottom line poker is that it is not a reliable method. If you can point out even one exception, a method becomes unreliable but your method has many many exceptions.


I don't think Poker is arguing it is in inflatable of offers difinative proof. More of an indicator. With that in mind it doesn't need to have 100% accuracy to offer value.

Like he said earlier he takes rookie seasons with a pinch of salt which would include the examples you give - Hulkenberg against Barrichello or Sainz against Verstappen.


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 12:05 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
You can't but that's not what I'm actually talking about. I'm trying to explain that your method is flawed because the size of the average gaps in qualifying times are not a reliable metric. Do you normalize the spread; throwing out outliers? Do you ensure that there is a representative and statistically significant sample size? Do you adjust for variations in approach? For example, Rosberg is on the record explaining that in 2014, he focused his setups on qualifying; believing that focusing on track position was his best strategy for winning the championship. It's the one season where he out-qualified Hamilton and it completely skews their head to head qualifying matchup and any data that can be taken from it. If you don't make a substantial effort to actually ensure the viability of the data; you can't put too much faith in it.

You also have to account for fluctuations from season to season between teammates. I believe Raikkonen was pretty much on par with Vettel in qualifying in 2016 after having been demolished in 2015 for example. There can also be one or two sessions that massively alter the averages; especially if a driver loses time with an error on their hotlap. Now making an error doesn't mean that you're slow; it means that you made a mistake.

I think the transitive property can only be applied in general terms. By that I mean that if driver A is faster than driver B and driver B is faster than driver C; driver A is also faster than driver C. Trying to use the exact margins will not work.

There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.

That's all well and good if I wasn't finding that I could actually start to see some correlation like for instance:-

Alonso > Massa 0.27s
Bottas > Massa 0.17s
Therefore
Alonso > Bottas 0.1s

Hamilton > Alonso 0.07s
Hamilton > Bottas 0.17s

That's off the top of my hand, I will come back and check tomorrow but I know I'm at max only 0.01s out.

The qualifying system when Hamilton and Alonso were teammates was one that had drivers using different fuel loads in Q3 and Alonso had the lighter load 11 times to Hamilton's 6. That was also Hamilton's rookie season (not a reliable year to use to judge his pace in his prime). The Hamilton Bottas gap will likely go up as you have just gotten through the part of the season where Bottas is closest to Hamilton and have yet to get through the part where he is furthest away.

Regardless, through sheer probability it's certainly likely that your method will work sometimes but there will be other times when it will not work. Use your method to explain why Hulkeberg is slower than Ricciardo but faster than Sainz. Use your method to explain why Rubens was slower than Button but faster than Hulkenberg. The bottom line poker is that it is not a reliable method. If you can point out even one exception, a method becomes unreliable but your method has many many exceptions.

Like I said previously the Hamilton and Alonso comparison can be done from 5 years of Button data, the Hamilton/Bottas comparison doesn't include this season as like you I'm aware how the seasons tend to play out.

Someone else also pointed out the need to avoid rookie seasons which the Hulk was when he was teamed up with Rubens, another example of this is Verstappen who clearly wasn't close to full speed until his third season in F1, not surprising perhaps for a 17 year old who only had one full season of car racing under his belt before he joined F1.

For me personally I definitely see some correlation while it's for others to criticise as they wish, if nothing else maybe people find it a bit interesting? :)

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 12:54 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
There are countless qualifying sessions that have gone into it, I don't believe that drivers mess up in qualifying as often as you might think and it's not one lap qualifying, I think it gives me a good idea of the pecking order but of course it's not a statement of fact especially with people who prefer to be subjective.

I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.

That's all well and good if I wasn't finding that I could actually start to see some correlation like for instance:-

Alonso > Massa 0.27s
Bottas > Massa 0.17s
Therefore
Alonso > Bottas 0.1s

Hamilton > Alonso 0.07s
Hamilton > Bottas 0.17s

That's off the top of my hand, I will come back and check tomorrow but I know I'm at max only 0.01s out.

The qualifying system when Hamilton and Alonso were teammates was one that had drivers using different fuel loads in Q3 and Alonso had the lighter load 11 times to Hamilton's 6. That was also Hamilton's rookie season (not a reliable year to use to judge his pace in his prime). The Hamilton Bottas gap will likely go up as you have just gotten through the part of the season where Bottas is closest to Hamilton and have yet to get through the part where he is furthest away.

Regardless, through sheer probability it's certainly likely that your method will work sometimes but there will be other times when it will not work. Use your method to explain why Hulkeberg is slower than Ricciardo but faster than Sainz. Use your method to explain why Rubens was slower than Button but faster than Hulkenberg. The bottom line poker is that it is not a reliable method. If you can point out even one exception, a method becomes unreliable but your method has many many exceptions.

Like I said previously the Hamilton and Alonso comparison can be done from 5 years of Button data, the Hamilton/Bottas comparison doesn't include this season as like you I'm aware how the seasons tend to play out.

Someone else also pointed out the need to avoid rookie seasons which the Hulk was when he was teamed up with Rubens, another example of this is Verstappen who clearly wasn't close to full speed until his third season in F1, not surprising perhaps for a 17 year old who only had one full season of car racing under his belt before he joined F1.

For me personally I definitely see some correlation while it's for others to criticise as they wish, if nothing else maybe people find it a bit interesting? :)

I was not trying to criticize you and it is an interesting thing to look at. My only point was that you're overreaching with your conclusions.


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 7:23 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
I think you're actually still missing the point poker with all due respect. Ask yourself this question; what are these numbers actually showing me? The red herring is that there is such a thing as being 2 tenths faster or 3 tenths slower than another driver. Bottas might be 2 tenths off of Hamilton in a place like Canada but he might be 7 tenths off in Singapore and 5 tenths off at Spa. At Russia, he might be a tenth ahead. Gaps will vary track to track for various reasons. Gaps may also vary from season to season for various reasons. A track like Singapore has so many corners and is so technical that there can be a lot of time in it. A track like Austria doesn't really offer that many opportunities for drivers to really separate themselves from each other.

You claim to be placing a high value on objectivity but, as with all statistics, you must first know whether the numbers you are collecting actually justify the conclusions you want to draw with them. For your method to generate accurate results; you would need qualifying to have rules where drivers can take as many laps and tires as they need to get their best time in so that you don't have outlier results based on errors. You would also need to ensure that the same tracks were used every season so that the time gaps were actually meaningful and fair. You would also need weather to be consistent for the same reason. You would need both drivers to have the same approach to their races with regards to setup and prioritization of qualifying or the race. You would need to normalize for variations in a driver's performance. You would need drivers to spend at least 100 races with each of their teammates to generate an even remotely statistically significant sample size. Essentially, you would need a lot of things to be the case that are very clearly not the case. Your method will allow you to make detailed observations about how a driver compares to his teammate in any given span of time. What it won't do is give you precise figures that can be useful in cross-comparisons (which is what you seem to want to use it for).

All that said, as I mentioned before, there is value in using the transitive property when looking at racing drivers. The value is NOT in finding precise time deltas between drivers who have never been teammates. We can assume, for example, that Verstappen is probably faster than Hulkenberg as Max was faster than Daniel and Daniel is faster than Hulkenberg. We can't know how much faster Max is though unless he and Nico actually team up.

That's all well and good if I wasn't finding that I could actually start to see some correlation like for instance:-

Alonso > Massa 0.27s
Bottas > Massa 0.17s
Therefore
Alonso > Bottas 0.1s

Hamilton > Alonso 0.07s
Hamilton > Bottas 0.17s

That's off the top of my hand, I will come back and check tomorrow but I know I'm at max only 0.01s out.

The qualifying system when Hamilton and Alonso were teammates was one that had drivers using different fuel loads in Q3 and Alonso had the lighter load 11 times to Hamilton's 6. That was also Hamilton's rookie season (not a reliable year to use to judge his pace in his prime). The Hamilton Bottas gap will likely go up as you have just gotten through the part of the season where Bottas is closest to Hamilton and have yet to get through the part where he is furthest away.

Regardless, through sheer probability it's certainly likely that your method will work sometimes but there will be other times when it will not work. Use your method to explain why Hulkeberg is slower than Ricciardo but faster than Sainz. Use your method to explain why Rubens was slower than Button but faster than Hulkenberg. The bottom line poker is that it is not a reliable method. If you can point out even one exception, a method becomes unreliable but your method has many many exceptions.

Like I said previously the Hamilton and Alonso comparison can be done from 5 years of Button data, the Hamilton/Bottas comparison doesn't include this season as like you I'm aware how the seasons tend to play out.

Someone else also pointed out the need to avoid rookie seasons which the Hulk was when he was teamed up with Rubens, another example of this is Verstappen who clearly wasn't close to full speed until his third season in F1, not surprising perhaps for a 17 year old who only had one full season of car racing under his belt before he joined F1.

For me personally I definitely see some correlation while it's for others to criticise as they wish, if nothing else maybe people find it a bit interesting? :)

I was not trying to criticize you and it is an interesting thing to look at. My only point was that you're overreaching with your conclusions.

Well for sure it's not the absolute truth, it's a prediction as much as anything, also I did say any kind of prediction regarding Verstappen and Hamilton is also a bit premature at this time.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Wed Jun 26, 2019 7:48 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
Well for sure it's not the absolute truth, it's a prediction as much as anything, also I did say any kind of prediction regarding Verstappen and Hamilton is also a bit premature at this time.

Out of curiosity; how did you accumulate your numbers for Max and Dan? Did you use their entire time as teammates?


Top
 Profile  
 
PostPosted: Thu Jun 27, 2019 1:29 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
Well for sure it's not the absolute truth, it's a prediction as much as anything, also I did say any kind of prediction regarding Verstappen and Hamilton is also a bit premature at this time.

Out of curiosity; how did you accumulate your numbers for Max and Dan? Did you use their entire time as teammates?

No just the last 2 seasons, in 2016 Verstappen joined the team mid season and was initially getting beat by Ricciardo, that year I just treat as a rookie year given the circumstances.

For the record I have Verstappen beating Ricciardo by 0.16s, KingVoids reasoning that Verstappen could well be 2 tenths clear of the field is what you might call the result of massively rounding up. :)

Also I need to repeat again that anything that relates to Hamilton is greatly immature at this present time.

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
PostPosted: Thu Jun 27, 2019 1:47 pm 
Offline

Joined: Sun May 25, 2008 7:55 pm
Posts: 6559
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
Well for sure it's not the absolute truth, it's a prediction as much as anything, also I did say any kind of prediction regarding Verstappen and Hamilton is also a bit premature at this time.

Out of curiosity; how did you accumulate your numbers for Max and Dan? Did you use their entire time as teammates?

No just the last 2 seasons, in 2016 Verstappen joined the team mid season and was initially getting beat by Ricciardo, that year I just treat as a rookie year given the circumstances.

For the record I have Verstappen beating Ricciardo by 0.16s, KingVoids reasoning that Verstappen could well be 2 tenths clear of the field is what you might call the result of massively rounding up. :)

Also I need to repeat again that anything that relates to Hamilton is greatly immature at this present time.

It's beyond immature. If you pay close attention to 2018; Daniel frequently faced reliability issues both on Saturdays and Sundays. In 2017, it was kind of the other way around. Anyway, I was curious about that.


Top
 Profile  
 
PostPosted: Thu Jun 27, 2019 3:45 pm 
Offline

Joined: Wed Dec 19, 2012 1:30 pm
Posts: 31524
sandman1347 wrote:
pokerman wrote:
sandman1347 wrote:
pokerman wrote:
Well for sure it's not the absolute truth, it's a prediction as much as anything, also I did say any kind of prediction regarding Verstappen and Hamilton is also a bit premature at this time.

Out of curiosity; how did you accumulate your numbers for Max and Dan? Did you use their entire time as teammates?

No just the last 2 seasons, in 2016 Verstappen joined the team mid season and was initially getting beat by Ricciardo, that year I just treat as a rookie year given the circumstances.

For the record I have Verstappen beating Ricciardo by 0.16s, KingVoids reasoning that Verstappen could well be 2 tenths clear of the field is what you might call the result of massively rounding up. :)

Also I need to repeat again that anything that relates to Hamilton is greatly immature at this present time.

It's beyond immature. If you pay close attention to 2018; Daniel frequently faced reliability issues both on Saturdays and Sundays. In 2017, it was kind of the other way around. Anyway, I was curious about that.

The issues were mainly on the Sundays I wouldn't include anything that was affecting performance on the Saturday.

I'm a bit puzzled why you would be wanting to refute the notion of Verstappen being quicker than Ricciardo?

_________________
PF1 Pick 10 Competition

2013: 5th Place
2014: Champion
2015: 3rd Place
2016: 4th Place

2017: 9th Place
2018: 7th place
2019: Currently 21st

Wins: Canada 2018, Abu Dhabi 2017
Podiums: (8)


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic

All times are UTC


Who is online

Users browsing this forum: Bing [Bot], Laz_T800 and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group