Jump to content


Photo

Some hack data analysis of Practice Two


  • Please log in to reply
22 replies to this topic

#1 maccaFTW

maccaFTW
  • Member

  • 704 posts
  • Joined: October 09

Posted 17 October 2009 - 11:56

I'm playing around with some empirical modelling of lap times to predict which teams are going to be fast. I scratched together a system last night, and I've been collecting data and getting the descriptive statistics together today. There is a lot of refinement that needs to occur, so this is just a rough go.

Basically, I'm taking the top ten times from every driver from practice runs of five laps or more, the hypothesis being that if a driver does a practice run for five laps or more it is intended to gauge race pace. (To get a sense of how the car performs with laps on the tires and brakes, etc.) Usually, qualifying runs are no more than 2-3 laps, especially if they're run on the option tire. So if drivers are doing mock up qualifying runs, it's a good bet they won't be going more than 5 laps.

I'm going to analyze several factors, including:

* Mean time over the set of ten laps
* Standard deviation
* Differences from the drivers with the top car by ordinal lap (further description of my method when i get to it)
* Means and st. devs thereof

As I said, this is a very elementary empircal analysis. I'm using this as a foundation, and I intend to build on it.

I haven't yet computed differences in lap times; I'll do that later. Here's the raw data on the mean and standard deviations of each driver's set of times, by constructors standings and car number:

Brawn

Button: Mean = 72.922 seconds; St. Dev. = .188 seconds
Barrichello: 73.860; .825

Red Bull

Webber: 72.683; .102
Vettel: 72.935; .111

Ferrari

Fisichella: 73.528; .132
Raikkonen: 73.189; .100

McLaren

Hamilton: 72.997; .133
Kovalainen: 73.158; .135

Toyota

Trulli: 73.474; .324
Kobyashi: 73.598; .270

Williams

Rosberg: 73.815; .319
Nakajima: 73.462; .294

Renault

Alonso: N/A (Didn't do a run longer than 5 laps in practice; I'm trying to work my way around this)
Grosjean: 74.444; .561

BMW

Kubica: 73.358; .149
Heidfeld: 73.334; .240

Force India

Sutil: 73.582; .639
Liuzzi: 73.369; .376

Toro Rosso

Alguersuari: 73.253; .121
Buemi: 72.780; .25

Analysis

* Unless Brawn and McLaren come up with some very good setup changes from Practice Two or get strategy and qualifying right, they may not see the Red Bull drivers again after the first lap.

* The Red Bull Technology package as a whole seems to be very quick this weekend. Not only does Red Bull look to be by far the best car in race pace, but Toro Rosso has great race pace, as well. Buemi set some times on race runs that were actually quicker than Webber's top times, but there was a larger spread among his times, indicating less consistency. Alguersuari looks slower than his teammate, but his consistency is very good.

STR could well be in for a surprisingly good weekend. Will they pull a "Force India at Spa?" Probably not, but at least one of them should score some really good points ***PROVIDED*** the car stays on the track.

* Webber looks like he is on it this weekend. He has the best mean time and the second best standard deviation, indicating that he was both quick and consistent. He gave the impression that he's going for a win and doesn't intend to help Vettel, and he's certainly shown he means it with his second practice performance.

* Barrichello looks to have had an off-kilter session. He set some quick times, but none of them were on extended runs, and his times were all over the place.

* Button said he's happy with his race pace, and my data backs him up. He looks to have good enough pace to challenge for the podium. The bad news, though, is his laps were less consistent than laps from the Red Bull and McLaren drivers.

* Raikkonen once again looks like he could pull some decent points out of a car that isn't worthy of it. Not only is he right on the doorstep of Kovalainen in an inferior car, his standard deviation statistic is by far the lowest, indicating he was the most consistent driver in practice.

* Kobyashi is really doing a very good job. This is is his first full grand prix weekend, and not only is he just a little over a tenth per lap off of Trulli's average time (a side-by-side comparison of all their lap times mirrors this number), but as seen by his lower standard deviation statistic, he was also more consistent in practice. If he keeps it under him, he might have a nice weekend.

* Alonso didn't do any real race runs, but even looking at runs that look like they could have been race runs (say, runs that were aborted because he wanted things changed on the car?), his laps weren't much better than Grosjean's race fuel laps. In any event, Renault really look horrible this weekend so far.

My assessment of the current competitive order:

1. Red Bull- The quickest and most consistent
2. McLaren, Brawn- The STR might actually set some quicker race laps, but these two have better consistency. Along these lines, the BGP001 looks like it might set some quicker laps in the race, but the McLaren looks like the more consistent car over a race run.
4. STR- Really impressive pace so far. As I said, they could shock some people this weekend.
5. Ferrari, BMW- The Beemer might be a little quicker as a car. But Raikkonen is very quick and consistent and probably gets the car to figure on equal ground with the Beemer.
7. Toyota, Force India, Williams- They seem to constitute the second part of the midfield. Force India may have the most pace, but they're really all over the place.
10. Renault- They're really far off right now. They're going to need a significant setup change to get them back in business.

Advertisement

#2 Madras

Madras
  • Member

  • 3,911 posts
  • Joined: November 07

Posted 17 October 2009 - 12:01

Analysing free practice sessions....when will people learn?

#3 stevvy1986

stevvy1986
  • Member

  • 3,168 posts
  • Joined: October 07

Posted 17 October 2009 - 12:07

Given the ever changing conditions in both sessions it's impossible to really predict anything, plus you've got to bear in mind the weather for the next 2 days, which again makes it impossible to be certain of anything.

#4 bond

bond
  • Member

  • 2,554 posts
  • Joined: October 07

Posted 17 October 2009 - 12:09

Analysing free practice sessions....when will people learn?


Tell it to the teams, they do analysis on friday and saturday morning...
:cool:

#5 Madras

Madras
  • Member

  • 3,911 posts
  • Joined: November 07

Posted 17 October 2009 - 12:11

Think about it - some teams will run heavy in practice, some will run light, some in between, some will take fuel out and some will fuel up at some point - times will end up not very consistant and you wont know why.

#6 Madras

Madras
  • Member

  • 3,911 posts
  • Joined: November 07

Posted 17 October 2009 - 12:12

Tell it to the teams, they do analysis on friday and saturday morning...
:cool:


They do, and they have access to far more data, and even then they arent always right.

#7 maccaFTW

maccaFTW
  • Member

  • 704 posts
  • Joined: October 09

Posted 17 October 2009 - 12:14

Given the ever changing conditions in both sessions it's impossible to really predict anything, plus you've got to bear in mind the weather for the next 2 days, which again makes it impossible to be certain of anything.


Sure.

I hope to control for changing track conditions by taking top lap times from all of the long runs combined, though. Usually, they'll be setting their best lap times around the same period of the session.

Wet weather will surely change things.

One of the things that strikes me, though, is that my analysis seems to flesh out what Hamilton and Button are saying as far as expectations go. And I think I give yet more evidence to what seems to be the consensus that Red Bull could well run and hide this weekend.

#8 maccaFTW

maccaFTW
  • Member

  • 704 posts
  • Joined: October 09

Posted 17 October 2009 - 12:15

Think about it - some teams will run heavy in practice, some will run light, some in between, some will take fuel out and some will fuel up at some point - times will end up not very consistant and you wont know why.


I tried to account for all of this.

Like I said; I've got a really rudimentary approach. If anything, I hoped it could spark some interesting discussion.

#9 Madras

Madras
  • Member

  • 3,911 posts
  • Joined: November 07

Posted 17 October 2009 - 12:16

One of the things that strikes me, though, is that my analysis seems to flesh out what Hamilton and Button are saying as far as expectations go. And I think I give yet more evidence to what seems to be the consensus that Red Bull could well run and hide this weekend.


Well, Red Bull have the best car - we've seen in the last few races, so that could be predicted by most people. I'm expecting it to be quite similar to the last race tbh.

Edited by Madras, 17 October 2009 - 12:17.


#10 grunge

grunge
  • Member

  • 4,162 posts
  • Joined: May 09

Posted 17 October 2009 - 12:19

They do, and they have access to far more data, and even then they arent always right.

cut him some slack wise guy :down: .the analysis in an interesting read in any event..u dont necessarily have to draw conclusions from it.

#11 Sunflower

Sunflower
  • Member

  • 171 posts
  • Joined: July 05

Posted 17 October 2009 - 12:35

cut him some slack wise guy :down: .the analysis in an interesting read in any event..u dont necessarily have to draw conclusions from it.


True it is a better discussion then the Fan Fights...

Thanks MaccaFTW

Why dont we just chip in and see whoever has a better idea or could help to add to the mans appraoch he has already done and make even better comparisons.


#12 Blue

Blue
  • Member

  • 1,219 posts
  • Joined: January 00

Posted 17 October 2009 - 12:36

maccaFTW, Do you think St. Dev. can be used directly to estimate the ‘goodness’ of the setups?

#13 KWSN - DSM

KWSN - DSM
  • Member

  • 7,822 posts
  • Joined: January 03

Posted 17 October 2009 - 12:41

cut him some slack wise guy :down: .the analysis in an interesting read in any event..u dont necessarily have to draw conclusions from it.


I second that.

This is a F1 BB, why the need to degrade an earnest effort in reading insight from the scenario which is exactly what the teams have as well.

Obviously the teams know their program, however I think that any driver will be trying to find the limits of the car, performance of the car regardless of weight and tires, that is what practice is for.

I find it interesting to see that Kimi appear the most consistent, which I to a point will argue is part of what makes a driver good, being consistent will outweigh the drivers who is less, and will enable a talented driver to outperform a slightly better car, simply through being raced 'better'.

It is also noteworthy that Kobayashi is as close to Trulli as he is, since the 'experiments' of Renault and Ferrari have show, that it takes a lot to step into a race seat midseason.

This weekend may not be the best for this exercise to be quite as correct as it could be, with the weather changing, however I think that this is a very worthwhile effort and hope that you keep it up, and share with us in Abu Dhabi, and again next season.

Cheers for the effort.

:cool:


#14 Atreiu

Atreiu
  • Member

  • 9,473 posts
  • Joined: May 07

Posted 17 October 2009 - 13:10

Nice thread, thanks for the numbers.
Has there ever been a race in which all 4 Red Bull cars scored points?

#15 maccaFTW

maccaFTW
  • Member

  • 704 posts
  • Joined: October 09

Posted 17 October 2009 - 13:10

maccaFTW, Do you think St. Dev. can be used directly to estimate the ‘goodness’ of the setups?


Maybe. Certainly if the driver has found a setup that clicks, then his times are likely going to be more consistent as the car suits his style and can hit a rhythm lap after lap, whereas a driver who hasn't found the right setup will be fighting the car for time; sometimes winning, sometimes losing.

To me, it looks like Red Bull and McLaren have really consistent cars at race pace, so their drivers are probably all happy with their race setups (although Vettel is probably looking for some pace at the moment to get on terms with Webber, who looks quick). Button had good consistency in his times, so he would seem to have found a good setup for him; he's said that he's happy with the car on primes for the race. Raikkonen was pretty consistent, but generally he sets pretty consistent times in practice (according to engineer accounts I've read).

#16 plastik2k9

plastik2k9
  • Member

  • 507 posts
  • Joined: June 09

Posted 17 October 2009 - 13:13

cut him some slack wise guy :down: .the analysis in an interesting read in any event..u dont necessarily have to draw conclusions from it.

Totally. I for one am happy that somebody has tried to make sense of the practice sessions for us all. It's not an exact science, far from it in fact, but hey, what else can we do with the times? Look at them and go "HA, meaningless!". No point watching practice if that's the case.

#17 Mat

Mat
  • Member

  • 7,674 posts
  • Joined: January 99

Posted 17 October 2009 - 13:35

great analysis. it would be great to have this kind of stuff done by the journos after all the sessions.

Unforunately i think the analysis will come to nought this weekend as it looks like it will be a wet one. And in that, all bets are off. Although, if wet, the money would still be on RB.



#18 Guizotia

Guizotia
  • Member

  • 1,274 posts
  • Joined: March 09

Posted 17 October 2009 - 13:41

cut him some slack wise guy :down: .the analysis in an interesting read in any event..u dont necessarily have to draw conclusions from it.


Yep very interesting. :up: Would like to see this every session!

Edited by Guizotia, 17 October 2009 - 13:42.


#19 Modern Lover

Modern Lover
  • Member

  • 727 posts
  • Joined: May 06

Posted 17 October 2009 - 14:22

Sure.

I hope to control for changing track conditions by taking top lap times from all of the long runs combined, though. Usually, they'll be setting their best lap times around the same period of the session.

Wet weather will surely change things.

One of the things that strikes me, though, is that my analysis seems to flesh out what Hamilton and Button are saying as far as expectations go. And I think I give yet more evidence to what seems to be the consensus that Red Bull could well run and hide this weekend.


Great analysis.

To see if your model holds water you could test it retrospectively, for example do it for the dry weekends of 2009.

Keep up the good work :up:

Advertisement

#20 mstar

mstar
  • Member

  • 1,105 posts
  • Joined: December 06

Posted 17 October 2009 - 14:32

great analysis its what the teams use they try to analyse the laptimes and predict who is going to be fast. I think Macca you done a great job and its a shame we cnt feed in Fp3 as well (if we can)

#21 agni

agni
  • Member

  • 224 posts
  • Joined: February 01

Posted 17 October 2009 - 14:36

Great work :up:

Looking at teammates the BMW drivers are very close to each other. Their StdDev is also not far off.

#22 KWSN - DSM

KWSN - DSM
  • Member

  • 7,822 posts
  • Joined: January 03

Posted 17 October 2009 - 14:44

I really think that it is a pretty cool project that you have there, and as one suggested if trying retroactively, then you could get a better feel for how valid your premise is, and I would suggest also enable that you create a bigger datapool to gain knowledge from.

Where are you getting all the lap times from?

:cool:

#23 Ray Singh

Ray Singh
  • Member

  • 35 posts
  • Joined: June 08

Posted 17 October 2009 - 15:04

Good effort I reckon, keep it coming :up: