23 replies to this topic

[tkulla]

Since it's that time of year to reflect back on the season that was, I got to thinking about all the talk about how the McLaren was the fastest car for much of the season until the Red Bull took over for a while and then McLaren fought back. But with Fernando finishing just a few points behind Seb and with Ferrari beating McLaren to second in the WCC I wondered if there was more to it than just Woking's operational errors and mechanical failures and Alonso's strong season.

So I decided to take a look at the fastest laps for the four top teams (Red Bull, McLaren, Ferrari, Lotus) to see who was fastest in the races. Now of course this tells only a sliver of the total picture and there are a lot of variables to take into account (race strategy, motivation to go faster near the end of the race, traffic, etc.) but I still think it's useful information that compares the cars over a 20-race season with both drivers from each team contributing fastest laps. Each team had races where they were way off the fastest lap (amongst the four teams - other teams are ignored for our purposes) and the results are quite interesting.

To start, here is the race-by-race graph, which shows the difference from the fastest of the group at each race (so if Red Bull was quickest their difference is 0).



As you can see, Red Bull was the quickest of the four team in fastest laps for exactly half the races. Here's the full breakdown

Fastest of the Four
10 Red Bull
5 McLaren
3 Lotus
2 Ferrari

No surprises there. With Seb's propensity to go for fastest laps I expected Red Bull to edge McLaren.

Comparing the average fastest lap times for the four teams was a bit more surprising though:

Average Fastest Lap Time
90.265 Red Bull
90.635 Ferrari (+.370)
90.770 Lotus (+.505)
90.797 McLaren (+.532)

Shockingly, McLaren is the slowest of the four! Looking at the data for bad races and we see that Mclaren had a few stinkers where there were more than one second off the fastest of this group. Checking all teams for such bad races gives us:

Number of Races > 1 sec off pace
7 Lotus
6 McLaren
5 Ferrari
1 Red Bull

So except for Red Bull they all had roughly the same amount of "bad" races when it comes to fastest lap. Any way you look at it Red Bull clearly had an advantage over the rest. But was Ferrari really quicker than McLaren on Sundays? Looking at which of those two teams did the faster race lap in each race yields:

Ferrari vs. McLaren
McLaren 11, Ferrari 9

I must say I find this very surprising. I expected a clear margin between McLaren and Ferrari and Lotus but it's not really there. McLaren certainly had the better car on Saturdays (though this was mitigated somewhat by Ferrari's blinding starts) and when I put the numbers together Massa contributed far more than I expected him to (fastest lap for Ferrari) which may have been due to driving with more midfield-style strategies (a lot of teams had the absolute fastest laps this year too).

Thoughts?

[ViMaMo]

Appreciate the effort but its so skewed. Id rather not touch THE fastest lap for any stats sake. Example, X has 10 odd laps to go, tyres falling off, he can finish race, but he pits for fresh rubber and pulls in fastest after fastest lap. Then you have someone building up a cushion and controlling the pace.

I think a better analysis would be to go for the qualifying laps where everyone (in top 10) is on a more level playing field.

Edited by ViMaMo, 14 December 2012 - 02:52.

[tkulla]

Appreciate the effort but its so skewed. Id rather not touch THE fastest lap for any stats sake. Example, X has 10 odd laps to go, tyres falling off, he can finish race, but he pits for fresh rubber and pulls in fastest after fastest lap. Then you have someone building up a cushion and controlling the pace.

I think a better analysis would be to go for the qualifying laps where everyone (in top 10) is on a more level playing field.

The biggest problem I have with qualifying as a measure of the cars is the free DRS use. Since they can't do this during the race qualifying times aren't representative of real race pace at all. Fortunately they've changed that rule for next year so qualifying will be more relevant to race speed.

I agree that there are some flaws with fast lap as a barometer, but the scenario you paint (late pit stop followed by fast laps on fresh tyres) was more likely to be seen by lesser teams than these four. I'm sure it happened a couple of times, but not enough to greatly modify the results here.

[BenettonB192]

How can the McLaren set a fastest lap when it falls apart in the middle of the race?

[Wander]

There is some truth in these statistics, but I think it makes Mclaren's car look far worse than it was due to a few things:

1. Spanish Grand Prix stands out like a sore thumb and distorts the statistics a fair amount. Hamilton had to start last, because of the fuel problem despite being driving the fastest Q3 time (by a large margin) and this forced him to drive a 2-stop race while almost everyone else did three stops. Button was dreadfully slow in both qualifying and the race as this GP was a part of his mid-season slump in form. He did a three-stopper and his fastest lap-time was still less than 0.3s faster than Hamilton's best.

To add to facts about this Spanish GP, it is hardly surprising that Grosjean's time was overwhelmingly the fastest lap time of the race when his last stint was only 15 laps long whereas Massa did 20, Alonso 22, Vettel 24, Webber 25, Button 28 and Hamilton 31. All these drivers except Vettel did their fastest time quickly after starting their last stint so fuel corrected they probably wouldn't be that far off each other. This was of course a result of the Lotus being the best car in terms of tyre degradation in that race, but seeing how well Hamilton actually did with two stints, he might have been close to doing equally well had he started on pole (doing three stints).

Clearly Button was not any better a benchmark for Mclaren than Massa was for Ferrari in that race. Massa's fastest lap was over a second slower than Alonso's fastest (despite a shorter last stint).


2. Canadian GP is again misleading. Hamilton won this race, because of a superior strategy, but this strategy involved him driving 20 laps on soft tyres for the last stint while Vettel did 7 on supersofts and Massa 12 also on super softs. The most impressive car in the race here was actually again Lotus who had amazingly low tyre degradation and posted very competitive fastest laps despite doing them on very long last stints. Too bad they were average in qualifying (like they were all season).

3. Belgium is obviously not representative of Mclaren at all. Button was so superior in the race that it's not even funny. He did one stop whereas all those others did two stops. I believe you might also have a mistake there as the source I am checking right now ( http://en.mclarenf-1...sa#.UMqgWm9mtuI ) says that Button's fastest lap was under a second slower than Massa's fastest lap, not over a second like your graphics appear to claim. Please tell where you are getting your data from so we can double check, cause if there's a mistake there, there might be many others as well.

I don't have time to raise other similar examples, but I'm sure that you knew that there would be distorting statistics like this already when you compiled this graphic. I think I would still agree with you that

1. Mclaren was the best qualifying car over the season, but not quite as good in the race

2. Red Bull was overall the best car to have in the race

3. Lotus and Ferrari were both stronger cars on the race days than on Saturdays, often matching and sometimes bettering the RBRs and Mclarens starting ahead of them.

I would like to see how much the statistic would change if you did something like ignoring the worst weekend of each car (4 races). I think they might end up closer to each other.

[abc]

1. Mclaren was the best qualifying car over the season, but not quite as good in the race

2. Red Bull was overall the best car to have in the race

I dont know about point 2. McLaren had definitely better pace till Spain. Between Monaco and Silverstone (only 4 races) they curiously fell even behind Ferrari. Since Germany they had best speed on Sunday and could win all except 3 races (Japan - India). In Hungary there were probably quicker cars on Sunday but they won it anyway.
In races from Singspore onwards I feel RBR were relatively stronger on Saturday not only in comparison to Ferrari but to lesser degree to McLaren, too. They strategy was often build around to be quickest in Q.

[Wander]

I'm not completely sure about number two myself either. Mclaren and Red Bull were very close over the whole season.

Edited by Wander, 14 December 2012 - 08:13.

[tkulla]

I would like to see how much the statistic would change if you did something like ignoring the worst weekend of each car (4 races). I think they might end up closer to each other.

Fastest Lap Averages with worst race removed
89.729 Red Bull
90.018 Ferrari (+.289)
90.176 McLaren (+.447)
90.255 Lotus (+.527)


So this is over 17 races (Bahrain, Spain & Japan were removed since they gave the biggest outliers for each team). Not much has changed. Ferrari and Mclaren closed up a bit on Red Bull, McLaren gave a small amount on Ferrari and Lotus fell to the back (Spain was their best and a bad one for the others).

It still looks like Ferrari had an awfully good car on Sundays. Perhaps our perception is skewed by the first/last effect somewhat. McLaren came out of the gates fastest, but Australia is often not a representative circuit. Add that to their strong finish and it's easy to see how we make the jump to "Mclaren fastest" or "equal fastest" but these statistics don't bear that out.

[LiJu914]

Add that to their strong finish and it's easy to see how we make the jump to "Mclaren fastest" or "equal fastest" but these statistics don't bear that out.

Because the statistic is useless.

Just one example:
Fastest Lap in India: Jenson Button. McLaren was nowhere near Red Bull in that race...

Edited by LiJu914, 14 December 2012 - 13:52.

[tkulla]

Because the statistic is useless.

Just one example:
Fastest Lap in India: Jenson Button. McLaren was nowhere near Red Bull in that race...

Okay, that's annoying. The statistic of the fastest lap by either member of team in actual race conditions is useless? I'm surprised that a poster who is normally more thoughtful would say such a thing.

Exactly how is it less relevant than a low-fuel lap on fresh tyres (which happens exactly never in race conditions) with full use of DRS over the course of a lap on Saturday when it comes to ultimate race pace?

Your one-race example is not terribly compelling either, considering it's cancelled out by similar races in the other direction, including the very next race (Abu Dhabi).

[LiJu914]

Okay, that's annoying. The statistic of the fastest lap by either member of team in actual race conditions is useless? I'm surprised that a poster who is normally more thoughtful would say such a thing.

Exactly how is it less relevant than a low-fuel lap on fresh tyres (which happens exactly never in race conditions) with full use of DRS over the course of a lap on Saturday when it comes to ultimate race pace?

Your one-race example is not terribly compelling either, considering it's cancelled out by similar races in the other direction, including the very next race (Abu Dhabi).

I´m sorry if my choice of words may have come across too harsh.

I didn´t claim that Qualifying would be a better indicator for the respective race-performances, did i?

Anyway...That this is all cancelled out over a whole season is just your assumption and i see not much evidence for that.

I used the example for a reason as Button had an even worse strategy to achieve that fastest lap (his tyres were 8 laps older than Vettel´s) but his FL is still 0.5sec faster than the race leader´s, who finished half a minute in front of him...

In Spa - a race JB completely dominated - he drove only the 10th fastest lap.

I could go on and on, but i think it´s easier to suggest that you just look up the statistics again and think about how they match with the actual performances, we could see on race-day... One example is, that the actual race winner almost never set the FL of the Grand Prix.

Edited by LiJu914, 14 December 2012 - 16:19.

[Sakae]

I am also doubtful that top speed clocked is really a reliable indication of overal race related potential performance. A faster car could have simply a different setup which is good for race, but not so much for quali (Ferrari comes to mind). It cannot be concluded however that Ferrari was a slow car. Then another factor, driver's performance; his consistency how he prints repeatedly lap times, his skills in navigating track might create false impression that car is faster, yet some reasons for growing gap might be attributted to lesser effectivness of a driver that follows which start dropping off due to personal errors, degradation of tires, etc. A pertient question to ask is a car not only fast, but is it also drivable? Only drivers knows, thus where it leaves us? Qualification times are also limiting, because of traffic, and several other factors that are often blurring the picture, and skewing results.

Analysis of this kind I think is actually quite complex, and requires credible data, sector by sector, of the whole sample under study, and over whole season, just as understanding of noises on the patern. I am not sure if it is possible for one person to do all of this in Excel spreadsheet, but I tried anyway. Long time ago I thought that creating efficiency pattern would help me to understand where cars were standing in relationaship to each other considering combined effect of best of both cars and their drivers. I shall explain my approach:

We know purple times in each race for each sector, even if generated by different vehicles, thus theoretical best lap time for a race on that day can be set as benchmark (a team's car = 100% efficiency), to which I compared best sector time for each vehicle in the sample (regardless whether it was set by one driver or another, since they have relatively very similar vehicles). Delta in times (best theorethical - best sector by the team) will provide sector efficiency for the team, then recalculated for a race efficiency, and ultimately a season efficiency. One number for a team per season, as effort by either of drivers. Result - I was hoping - should have been indicative how team's vehicle compared to rest of them. (Unfortunately I got sidetracked by my paying job, then later data were hard to find, and analysis was never completed)...

I might still do that next year. Numbers will hopefully tell me car potential in each race in comparison to its actual race placing. It's a lot of numbers to evaluate, but for top five it could be reduced to about 300 results in all for final analysis.

Cheers (winter is loooong)

_______

For statistical purists, of course I may increase my sample size for each deviation (say average of best 10 sector times, etc.) to get more accurate profile. We shall see.

Edited by Sakae, 14 December 2012 - 16:05.

[tkulla]

I didn´t claim that Qualifying would be a better indicator for the respective race-performances, did i?

No, you didn't. But it is generally accepted as the way to evaluate the pace of the cars. What it actually is measuring the pace of the cars in a state in which they don't actually race. My main point is that this method is at least as valid as qualifying averages at measuring race pace. That said, neither is perfect - in fact both have major flaws. But I still see value in it, especially as it debunks the "Ferrari is a bad car" myth a bit.

The analysis that Sakae mentions would be a better indicator than either for sure. That's an awful lot of work so I can understand not having time for such a thing.

I've though of other possible (and simpler) indicators as well. Perhaps the fastest lap for each team in the middle quintile of each race to try to capture the pace of the cars on medium fuel levels. But strategy and traffic would likely have an even greater effect on this number than the overall fastest lap.

[tkulla]

This week's Autosport did something similar in a sidebar about how good the Ferrari was this year (page 38).

In "Raw Pace" (which I assume means qualifying speed) the Ferrari was the 4th best car after the McLaren, Red Bull and Lotus.

In "Race Pace" (which appears to be based on Fastest Laps) it was the third best car after the Red Bull and Sauber (!) and ahead of McLaren and Lotus.

In "Reliability" it was first with zero mechanical race failures, followed by Lotus (1), Red Bull (3) and McLaren (4).

[V8 Fireworks]

In "Reliability" it was first with zero mechanical race failures, followed by Lotus (1), Red Bull (3) and McLaren (4).

Amazing from the mighty Scuderia

For McLaren team, I am not sure if problems with air guns or trolley jacks in the pits should also count as mechnical ailures - as that also cost them many points.

[PassWind]

The data means something, how you interpret the data is another thing entirely. As always people point out singular instances in a attempt to debunk hang on debunk nothing as the OP doesn't make any real assumptions. It is just data after all however I think you can look at some broad brush observations other than simply oh Redbull had the best car so end of chat. I find the best car claims quite annoying in particular when denigrating a driver which only reflects a posters bias rather than a well considered reason for a teams sucess. There in lies in my opinion the main error in these types of arguments, the omission of team and the data presented can in some way represent the efforts of a team.

A designer can make a car that is in theory, that which perfectly represents his design intent, now whack in the constant variables, tires, weather, track surface, grid position and racing position. Those couple of hundred bodies employed to deal with those come into play, even though they are rarely mentioned or given much in recognition in most posters arguments for and against a team or driver.

So to the data which in my opinion broadly illustrates a teams overall ability to get the car closest to its performance delta, lap after lap with two main goals to achieve, one lap pace and race pace. One lap pace can drastically reduce the complexity of working through the race variables of course coming to fruition with adequate control of race pace by being consistently closer to the cars performance delta than other teams. That juggle goes on between drivers, engineers, designers well gee the whole team.

The data is relevant and one take away I get is the strength of a team for the whole season and why they were strong or weak.

[Petroltorque]

Fastest Lap Averages with worst race removed
89.729 Red Bull
90.018 Ferrari (+.289)
90.176 McLaren (+.447)
90.255 Lotus (+.527)


So this is over 17 races (Bahrain, Spain & Japan were removed since they gave the biggest outliers for each team). Not much has changed. Ferrari and Mclaren closed up a bit on Red Bull, McLaren gave a small amount on Ferrari and Lotus fell to the back (Spain was their best and a bad one for the others).

It still looks like Ferrari had an awfully good car on Sundays. Perhaps our perception is skewed by the first/last effect somewhat. McLaren came out of the gates fastest, but Australia is often not a representative circuit. Add that to their strong finish and it's easy to see how we make the jump to "Mclaren fastest" or "equal fastest" but these statistics don't bear that out.

From a statistical point removing outliers is frowned upon. It's cooking the books and in academic circles it would discredit all your research. The more data points, the more valid the final average statistic.

Edited by Petroltorque, 23 December 2012 - 06:53.

[tkulla]

From a statistical point removing outliers is frowned upon. It's cooking the books and in academic circles it would discredit all your research. The more data points, the more valid the final average statistic.

Thanks for the peer review. I happen to agree with you, which is why my original post was just the pure data with no interpretations or modifications of any kind. The outliers removed data was by request, and as it turns out made little difference.

[Petroltorque]

Thanks for the peer review. I happen to agree with you, which is why my original post was just the pure data with no interpretations or modifications of any kind. The outliers removed data was by request, and as it turns out made little difference.

My apologies. I hadn't seen the original post.

[boldhakka]

We know purple times in each race for each sector, even if generated by different vehicles, thus theoretical best lap time for a race on that day can be set as benchmark (a team's car = 100% efficiency)...

You would be underestimating the theoretical best lap time by quite a margin with this approach. But it would give you an excellent lower bound on that measure (not the same thing as theoretical best lap time).

Edited by boldhakka, 23 December 2012 - 15:00.

[Wander]

The problem is really just that the data is thoroughly misleading. Sauber did not have the 2nd fastest race pace overall. Maybe they did lots of short stints near the end or something. I can recall a few such occurences.

Edited by Wander, 23 December 2012 - 15:02.

[ViMaMo]

Take the difference between lap times for every lap between driver A and B. Then sort the difference in ascending or descending order. The skew gives a much better idea as to how fast there were relative to each other. Leave out the pit stops and the larger differences, it should be a smooth curve.

[tkulla]

The problem is really just that the data is thoroughly misleading. Sauber did not have the 2nd fastest race pace overall. Maybe they did lots of short stints near the end or something. I can recall a few such occurences.

I wonder how many of these fastest laps were set on the harder of the two tyres on each weekend. Some cars may have had trouble extracting speed from the options (especially in qualifying) but maybe have been well suited to the primes and set their fastest race lap on those.

Sauber was a team that would often go for one-stop strategies. They often started on the harder tyres and ran long, and then used the softs for the final stint. This could explain their pace on the Autosport table.

Edited by tkulla, 25 December 2012 - 15:20.

[Szoelloe]

I wonder how many of these fastest laps were set on the harder of the two tyres on each weekend. Some cars may have had trouble extracting speed from the options (especially in qualifying) but maybe have been well suited to the primes and set their fastest race lap on those.

Sauber was a team that would often go for one-stop strategies. They often started on the harder tyres and ran long, and then used the softs for the final stint. This could explain their pace on the Autosport table.

The problem with the fastest lap thingy is that it is meaningless, and it is near-impossible to draw any conclusions from, because of the delta time racing enforced upon the teams by this type of construction and rubber. Fastest race laps were usually achieved when there was no need anymore for whatever reason to nurse the tyres. That situation was never simultaneous between teams, and the difference between outright pace of the cars in 2012 and the delta times on a stint, was HUGE IMHO. It's basically my only wish for next season: big difference between the two compounds, and the ability to push them for at least 70-80% of their lifespan. And then the fastest race-laps should bore some meaning by the end of the season.