[stat]Team mates comparison 2008[/stat]

[deeks6]

Fastest race lap is the most meaningless stat in F1 and has no place in any comparison IMO.

There have been many occaisions where a teammate has been on a different strategy (i.e 3 stops to 2) or has had an incident and done an extra pitstop so it can hardly count. There are other occaisions where a teammate is so far ahead of the other and bringing the car home safely for points whereas the other has nothing to lose (e.g. KR last year when Massa won the race). It is completely meaningless.

Qualifying is where both are driving to their maximum capacity and is the true test of speed over a lap.

Finishing ahead of your teammate only counts if both have finished - or if you have a stat that says X was ahead of Y when he retired. And even that stat is flawed if one driver constantly retires, indicating that he may be hard on cars or has incidents forcing him out.

F1 is about winning - there should be some weighting on the stats so that there is a good indication of which teammate is fastest on raw speed, which one has better racecraft (i.e. positions made relative to qualifying with consideration to retirements), which one is more reliable (i.e. number of driver errors versus things beyond their control).

These are interesting and probably indicate a trend however.

[Notso Swift]

How can anything be proven until the time when the team mate can no longer recover to be (At the minimum) equal?
In other words, until after the mid point of the season.

[vinegar]

Originally Posted by Schummy

I don't think so. Looking into Q2 data, Kubica beats Heidfeld 4-1, so Robert is being really faster than NH in Saturdays.

But the strategy bias in qualifying positions is a valid point. In fact I'm doing a stat with Q2 along the season, and I'm thinking about using it instead grid position. But, of course it is more work... , so I don't know when (or if) I'm going to change it.

Problem is, an intelligent driver in a BMW, McLaren or Ferrari knows he doesn't need to shave off the last couple of tenths to make the top ten, so is quite unlikely to go flat out in Q2.

I don't think there is any reliable and measurable indicator of comparative pace since 'proper' qualifying was killed. It's all subjective now. Unless you have the kind of data the teams have, and even then it's a bit of a stab in the dark because they can't read the driver's intention at any given time.

[Schummy]

Originally Posted by deeks6

Fastest race lap is the most meaningless stat in F1 and has no place in any comparison IMO.

There have been many occaisions where a teammate has been on a different strategy (i.e 3 stops to 2) or has had an incident and done an extra pitstop so it can hardly count. There are other occaisions where a teammate is so far ahead of the other and bringing the car home safely for points whereas the other has nothing to lose (e.g. KR last year when Massa won the race). It is completely meaningless.

(I have "bolded" some words)

"Most meaningless" or "completely meaningless" is an exaggeration, I'm afraid.

You point several scenarios where FLs are not realistic:
- Different strategy, extra pitstop: it doesn't influence a lot their FL. Fastest lap is done in an optimal window of laps in a stint. Usually, but not always, in the end, when drivers go with few fuel. The difference between 2 or 3 pitstops is the driver has 3 or 4 "opportunities" to set his fastest lap (and the optimal phase of his stints). Every stint is different, in every track, in every GP. Sometimes the first stint is the overall optimal, sometimes it's the second, etc. Most of time, this variation introduces "noise" in the data, i.e. makes harder to decide who is the best.

-Race incident: if the incident happens early, it obviously affects FL. But if it happens in the second part of the race, the driver at least has had an opportunity to set a FL in the earlier laps. Anyway, race's final position is more dramatically influenced! Almost every stat is influenced by a early incident.

-Driver "cruising" along: it is going to happen mostly in the last stint, and no in every race or in every team. So in the rest of the race (stints) drivers has to go reasonably fast.

Respecting other issues:

-Qualifying: Q3 is, as mentioned by others, not totally reliable because of the variation in fuel loads. Q2 (or Q1 in lesser teams) is more reliable (but not absolutely). But the main point is qualifying is different that FLs, we have seen it many times (Senna-Prost, Trulli, Webber,...). So an excellent qualifier is not always a great race driver. Qualifying is just an important factor.

-Comparisons and retirements: Most of retirements are probably "random", but some drivers are more prone to errors or are harder with the mechanical elements than others. It is part of the racecraft and have to be considered. Random luck with retirements are factored out by the statistical analysis (significance).

[Schummy]

Originally Posted by Notso Swift

How can anything be proven until the time when the team mate can no longer recover to be (At the minimum) equal?
In other words, until after the mid point of the season.

"Proven" doesn't mean in this context "to win in the season" but to statistically "prove" than one is better (in those stats) than the other (with a reasonably high probability).

So if A beats B 19-17 along the whole season, it doesn't prove A is better than B, because if A and were exactly equal, 19-17 could happen by chance (as heads and tails in coin tossing). But, say, 8-0 is very unlikely to be caused by chace, so while is *possible* that B recovers in the rest of the season, it is highly probable he doesn't make it.

[Schummy]

Originally Posted by vinegar

Problem is, an intelligent driver in a BMW, McLaren or Ferrari knows he doesn't need to shave off the last couple of tenths to make the top ten, so is quite unlikely to go flat out in Q2.

I don't think there is any reliable and measurable indicator of comparative pace since 'proper' qualifying was killed. It's all subjective now. Unless you have the kind of data the teams have, and even then it's a bit of a stab in the dark because they can't read the driver's intention at any given time.

Maybe Ferrari drivers can do it at some extent. I doubt Heidfeld is regulating his pace in Q2 taking in account the narrow gap he has had in some occasions respect the 11th driver.

Anyway it is a matter of opinion.

[deeks6]

Originally Posted by Schummy

You point several scenarios where FLs are not realistic:
- Different strategy, extra pitstop: it doesn't influence a lot their FL. Fastest lap is done in an optimal window of laps in a stint. Usually, but not always, in the end, when drivers go with few fuel. The difference between 2 or 3 pitstops is the driver has 3 or 4 "opportunities" to set his fastest lap (and the optimal phase of his stints). Every stint is different, in every track, in every GP. Sometimes the first stint is the overall optimal, sometimes it's the second, etc. Most of time, this variation introduces "noise" in the data, i.e. makes harder to decide who is the best.

-Race incident: if the incident happens early, it obviously affects FL. But if it happens in the second part of the race, the driver at least has had an opportunity to set a FL in the earlier laps. Anyway, race's final position is more dramatically influenced! Almost every stat is influenced by a early incident.

-Driver "cruising" along: it is going to happen mostly in the last stint, and no in every race or in every team. So in the rest of the race (stints) drivers has to go reasonably fast.

Respecting other issues:

-Qualifying: Q3 is, as mentioned by others, not totally reliable because of the variation in fuel loads. Q2 (or Q1 in lesser teams) is more reliable (but not absolutely). But the main point is qualifying is different that FLs, we have seen it many times (Senna-Prost, Trulli, Webber,...). So an excellent qualifier is not always a great race driver. Qualifying is just an important factor.

-Comparisons and retirements: Most of retirements are probably "random", but some drivers are more prone to errors or are harder with the mechanical elements than others. It is part of the racecraft and have to be considered. Random luck with retirements are factored out by the statistical analysis (significance).

I have bolded some words too ... a lot of "if" s and "sometimes" and "usually" and "not always" ... just backs up that the fastest lap is a useless statistic.

For example, you are saying that Webber is "not proven" to be better than Coulthard yet has outqualified him and finished ahead in every race ... what more do you need?

Your stats are also flawed with finishing position if a driver retires (through no fault of his own).

A better way to do it is by weighted indexes.

1. Qualifying - Time and position .. . can then be averaged over a year .. gives indication of raw speed
2. Finishing position compared to teammate and qualifying ... also averaged over a year ... gives indication of racecraft and consistency (i.e average position compared to qualifying)
3. Finishing record ... weighted to driver error or mechanical failure ... gives indication of judgement and car care.

Here's an example:

2001 - Montoya versus Ralf Schumacher

Average Qualifying Position - JPM 5.4 Ralf 3.4
Qualify In Front - JPM 6 Ralf 11
Average Qualifying Speed - Ralf faster by 0.2899
Average Finishing Position (if finished) - JPM 3.17 Ralf 3.4
Finished in front (if both finished) - JPM 3 Ralf 1
Races Finished - JPM 6 Ralf 10
Retirement Cause - JPM 4 Driver Error 7 Mechanical
Retirement cause - Ralf 3 Driver Error 4 Mechanical
WDC Points - JPM 31 Ralf 49

Now, those stats tell me that Ralf was faster, more reliable and had better racecraft but there are others on this forum that would argue until they are blue in the face that JPM was unlucky or something. He set a faster lap in more races than Ralf and ... so what?

A simple addition of the "good" stats and minus the bad stat (driver error) shows a total of:

Ralf 106.923
JPM 45.91

Fast forward to 2002:

Average Qualifying Position - JPM 2.94 Ralf 3.47
Qualify In Front - JPM 8 Ralf 9
Average Qualifying Speed - Ralf faster by 0.0418
Average Finishing Position (if finished) - JPM 3.85 Ralf 4.0
Finished in front (if both finished) - JPM 5 Ralf 4
Races Finished - JPM 13 Ralf 12
Retirement Cause - JPM 1 Driver Error 3 Mechanical
Retirement cause - Ralf 1 Driver Error 4 Mechanical
WDC Points - JPM 50 Ralf 42

Those stats tell us how much JPM improved. Despite 7 poles, Ralf was still quicker than him more of the time (more consistent) but JPM had better race finishes (even though it was close), had less mistakes and was kind to the car so the indexes now result in:

Ralf 68.7106
JPM 88.56

Now they are far more indicative of the true teammate comparison than the simpler stats IMO.

[Sev]

Originally Posted by deeks6

A simple addition of the "good" stats and minus the bad stat (driver error) shows a total of:

Ralf 106.923
JPM 45.91

Fast forward to 2002:

Ralf 68.7106
JPM 88.56

Out of interest, what you do mean by simple addition? Ie, your normalize all the stats and take their average?

Anyway, I think that comparison is pretty flawed as well however. I think it very nicely shows how well each driver performed in a given year. But pointing out the inconsistency as a concept that such measures are flawed is hardly fair.

2001 was Montoya's first year, Ralf was already well established. Now, I'm not going to say anything about who I think was better etc, but it is well recognized that there is a learning curve in F1.

I also think if you use your more advanced stats and compare them to Schummy's over a large dataset, I think you'll find they say more or less the same thing.

My favourite F1 team-mate comparison web site uses a system similar to Schummy's (although final classification is worth double), and says this about the Montoya/Ralf situation:

season	Highest	Points	wins	poles	class	quali	fl	'score'	Normalized
2001	1vs1	31vs49	1vs3	3vs1	5vs7	6vs11	7vs10	 -12	-0.706
2002	2vs1	50vs42	0vs1	7vs0	9vs7	9vs8	7vs8	 +4	+0.235

One thing that stands out here to me is the fact that fl actually seem to fairly well represent the actual out-come. In 2001 Ralf undeniably beat Montoya. In 2002, it was too close to be sure.

[deeks6]

Originally Posted by Sev

Out of interest, what you do mean by simple addition? Ie, your normalize all the stats and take their average?

Anyway, I think that comparison is pretty flawed as well however. I think it very nicely shows how well each driver performed in a given year. But pointing out the inconsistency as a concept that such measures are flawed is hardly fair.

2001 was Montoya's first year, Ralf was already well established. Now, I'm not going to say anything about who I think was better etc, but it is well recognized that there is a learning curve in F1.

I also think if you use your more advanced stats and compare them to Schummy's over a large dataset, I think you'll find they say more or less the same thing.

My favourite F1 team-mate comparison web site uses a system similar to Schummy's (although final classification is worth double), and says this about the Montoya/Ralf situation:

season	Highest	Points	wins	poles	class	quali	fl	'score'	Normalized
2001	1vs1	31vs49	1vs3	3vs1	5vs7	6vs11	7vs10	 -12	-0.706
2002	2vs1	50vs42	0vs1	7vs0	9vs7	9vs8	7vs8	 +4	+0.235

One thing that stands out here to me is the fact that fl actually seem to fairly well represent the actual out-come. In 2001 Ralf undeniably beat Montoya. In 2001, it was too close to be sure.

I had'nt thought about it enough to put any weighting or apply any other formulas but I think even the simple method amply demonstrates where the teammates sat at the time. In 2001, Ralf was miles ahead - yes, he was established and more experienced but he drove better and more consistently too. And there is an indication that he was kinder to the car.

However, in 02, Montoya improved every part of his game - less errors, more finishes, more consistency et5c etc. The "learning curve" as you put it is well demonstrated in these stats.

A similar picture would be shown with Raikkonnen and Coulthard, I think

[Greenback]

Schummy, pitstop strategy does have an impact on the driver's ability to achieve fastest lap. As you have said, FL is usually acheive towards the end of the stint when the car is light on fuel, hence when a driver is running longer and heavier, his tyres will be very worn out at the end of the stint.

[deeks6]

OK, how about this?

Multiply the average qual. position x WDC points x no of GP
Add bonus point for each time beatinf teammate
Multiply the avg speed difference x no of GP
Multiply the avg finishing x no of GP
Bonus point for each finish in front of teammate
Bonus point for race finish
Subtract avg. finishing position points for driver error = .5 point for mechanical
Double the WDC points

Using the Ralf/JPM model:

2001

Qual Position - JPM 76.5 Ralf 93.5
Bonus Points - JPM 6 Ralf 11
Qualifying Speed - Ralf 4.93
Avg Finishing - JPM 3.9
Finish in Front - JPM 3 Ralf 1
Races Finished - JPM 6 Ralf 10
Retirement Penalty - JPM (27.5) Ralf (18.5)
WDC Points x 2 - JPM 62 Ralf 49

Total "Index" - JPM 129.9 Ralf 204.93

2002:

Qual - JPM 102 Ralf 93.5
Bonus - JPM 8 Ralf 9
Qual Speed - Ralf 0.71
Avg Finish - JPM 2.55
Finish Front - JPM 5 Ralf 4
Races Finished - JPM 13 Ralf 12
Retirement Penalty - JPM (6.5) Ralf (7)
Points - JPM 100 Ralf 84

Total "Index" - JPM 223.95 Ralf 197.21

I think that reflects the true situation and allows for comparison of "indexes" from year to year. This clearly shows that Ralf had a slight fall in overall performance but JPM improved out of sight in the second year to surpass even Ralf's 2001. Probably helps to compare teams indexes too.

[deeks6]

Heres the 2008 to date using that method:

Raikkonnen 128.0
Massa 154.47
Kubica 128.14
Heidfeld 76.0
Kovalainnen 65.5
Hamilton 109.42
Trulli 65.96
Glock 1.5
Rosberg 39.2
Nakajima 15.0
Coulthard 5.0
Webber 53.47
Vettel 7.11
Bourdais 5.5
Sato 7.0
Davison 7.57
Alonso 61.84
Piquet -0.5
Sutil 0.0
Fisichella 27.75
Barrichello 6.0
Button 14.19


The Team Indexes are:

Ferrari 141.24
McLaren 87.46
BMW 102.07
Toyota 33.73
Williams 27.1
Red Bull 29.24
STR 6.31
Super Aguri 7.29
Renault 30.67
Force India 13.87
Honda 10.1

Those indicate to me that there is little difference between RB, Williams, Renault and Toyota (duhh) however Alonso, Webber and Trulli are far more superior to their teammates than Rosberg. Lower down the order Fisichella dominates Sutil.

Significant are Hamilton over HK, Kubica over Nick and Button over Rubins. The others are all in dispute. I consider this looking at the driver versus his team index %.