A great F1 passing study (must read!)

[Richard]

As others have said above...

Would be interesting to see how much reliability has changed. I recall a lot more DNFs when I first started watching F1. If cars are struggling to maintain reliability, then I suspect they will be more likely to vary in pace during a race, hence more overtaking. Today's cars seem to be able to nail perfect lap times on every lap.

The cars seem to be much closer in pace. That hinders overtaking. You need some sort of performance differential to enable a fast car to overtake a slow car. Today's cars on similar pace rely more on driver error to allow overtaking (or banzai out-braking moves).

Remember when cars failed the 107% rule for qualifying? With that sort of diversity in performance you are bound to get more overtaking.

Having said that... If there are big performance differences then quali is going to sort the fast from the slow more effectively than today's closely paced cars?

Would be nice to see some stats on DNFs, and variation in pace from front to back of the grid.

Of course correlation is not causation


edit - typo corrected to 107%

[timbo]

Remember when cars failed the 120% rule for qualifying? With that sort of diversity in performance you are bound to get more overtaking.

Well, it was not THAT bad - it was 107%)))

Actually, I think there is one another reason - speed control at the pit-lane that greatly restricted strategy options (although being a right decision safety-wise).
At early 90's (pre 1994) I remember often seeing leaders taking a fresh tyres, loosing 10 secs and then quickly moving up on the grid, today with even fastest stops costing 20+ seconds teams are not as flexible with a strategy.

[xpensive]

Why not simply ban the front wing altogether? Beholding those endplates, it would probably save a fortune in the process.

[Ciro Pabón]

I'd love to see Ciro post here with his overtaking histogram.

Well, with my usual modesty, I can attest they started the discussion about overtaking in that forum when I posted those graphs you mention, the ones we posted here two or three years ago. I posted there like five months ago or so (in May).

http://www.cliptheapex.com/forum/viewto ... 5367#p5367

I don't agree with Brogan conclusions (altough I think he's a great guy, go Brogan! ).

To summarize my point of view, F1 is like piloting a helicopter in a livingroom, in the immortal words of David Coulthard. If you're in the mood for artificial overtakings, watch NASCAR (go, Juan-derful! ). Me, I teach kids to drive karts and there is plenty of true overtakings (for me) in that kind of racing. I adore F1 not for the overtakings but for the precision. Not everybody likes that, but, hey, suit yourselves (you're going to do it anyway).

[Richard]

Here's the "Retired" data for the key years previously noted.

Analysis only counts cars that made it to the grid, and how many of those made it to the finish line.

Expressed as %DNF per starting driver over whole season. For example, 2009 has seen 298 drivers start on the grid. In total 50 retired, 10 classified but did not finish, so total of 60 DNF. Hence 20% (ie 60/298).

1984: 54% (Fuel tank capacity reduced)
1989: 51% (Engine capacity changed)
1999: 45% (Flexible wings banned)
2003: 32% (Single lap qualification)
2006: 30% (Engine capacity changed)
2008: 24% (pre OWG)
2009: 20% (OWG New rules)

Notes:
Data sourced from wiki summary of each season.
2006, 08, 09 Cars completing 90% were classified, but deemed to have retired for this analysis

The above only considers cars that made it to the grid. It is worth noting:
1984 DNQ was in place - 4% of entrants.
1989 DNQ was in place - 23% of entrants.
1989 DNPQ was in place - 10% of entrants

So in 1989, a total of 67% of entrants didn't make it the finish. Compared to 20% this year.

As before, correlation of this data with overtaking trends does not prove causation, although I suspect there is a reasonable hypothesis to link them.

edit - corrected example in third paragraph.

[Richard]

Remember when cars failed the 120% rule for qualifying? With that sort of diversity in performance you are bound to get more overtaking.

Well, it was not THAT bad - it was 107%)))

See the DNQ and DNPQ data for 1989 in the post above. It was a big factor!

[Pup]

How refueling affects passing? AFAIK refueling was banned at 1984, so I'd like to see pre-1984 figures.

Well, we can certainly see the effect when it was brought in in 1994.

What you got was teams and drivers very reluctant to pass, angling their strategy to pass while in the pits and you merely got the top teams locking out the front of the grid only for them to run lower fuel to maintain track position and stay where they were. Carrying fuel from qualifying into the race was a way of getting around that but it merely reduced qualifying to a qustion mark about fuel, and I hate hearing about those boring published fuel weights.

No refuelling wasn't perfect, but refuelling was an ill-conceived idea brought in in 1994 as a response to William's dominaton.

It's not quite as clear cut as it seems. 1994 only saw a significant drop in passing at four races: Monaco, Hockenheim, Hungaroring, and Magny Cours. Take those away, and passing actually increased that year. So the most you could say is that introducing refueling had a dramatic affect on passing on the tightest, street-like circuits. Of course, the ban on traction control would also affect those races the most, so things still aren't completely clear. It's hard though to credit or blame traction control for anything since we never knew who was running it. FWIW, I don't see a clear drop or rise in passing at these circuits in either '02 or '07, when traction control was reintroduced then rebanned.

On the other hand, I don't see that the '83 season - with refueling - lacked overtaking.

[Pup]

Here's the "Retired" data for the key years previously noted.

Does that include accidents, or just mechanical failures?

[Mystery Steve]

Looking at the mean number of passes per race in each given year is OK, but it doesn't necessarily paint the whole picture. You may be able to add a little depth by at least including the standard deviation with the mean values, but even that leaves some variability. It would also be beneficial to normalize the overtaking data since having 50 passes in a race with 10 cars is different than the same number of passes among 20 cars.

For the data to truly be statistically relevant you need to isolate any extra variables (i.e. you want a single independent variable, the year, and one dependent variable, the number of passes). The best way I can think to do this is to look at the data on a track-by-track basis. As a starting point, pick a track that has been used consistently for the last 25-30 years, or has at least been used for the majority of those years and look at the number of passes for each race at that track normalized against the number of cars that started the race in each respective year. Looking at the number of DNFs (normalized again) on a track-by-track basis would also be interesting and could give at least a little insight into general reliability. The problem with just using DNFs is you have to assume that the number of cars retired by accident stays generally proportional to the number retired by mechanical failure.

Does anyone know where to find reputable data for the number of passes and the number of DNFs for each individual grand prix for the last 25-30 years? Doing it this way would at least isolate track variability and the size of the field, but a big variable that can not be eliminated is the range of driver ability.

Side Note: I'm assuming a "pass" in this context is only measured at the start-finish line? In other words, if a driver gets passed but repasses in the next turn before reaching the start-finish line again, then it would not count.

[timbo]

Well, it was not THAT bad - it was 107%)))

See the DNQ and DNPQ data for 1989 in the post above. It was a big factor![/quote]

I don't arguing that - I just noted that 120% is a bit too high - think about it, with 20% pace deficit you would be lapped within 5 laps from the start of the race!

[marcello]

And Here is my analysis:

There's a reason why they say statistics are some of the biggest lies in the world [...].


I mean, read this:

"During a period of pretty much unchecked technical development from 1989 to 1993 there was a decrease of 9 in the average number of overtakes over the whole period.
During a period of constant FIA rule changes from 2004 to 2008 there was an overall reduction of 1 overtake per race."

[...] By using two different units of measurement for the two different periods, it's clear that this study is trying to get a point accross, instead of reporting unbiased information.

I will give u the reason (or at least the biggest factor) why the amount of passing fluctuates in F1, and has taken an overall plunge: EQUALIZATION of the cars AND drivers!!!

I am sick of ppl turning the FIA into a big enemy, whoever wrote this study needs to go... study... a statistics book.

Long live Max Mosley.

[Richard]

I don't arguing that - I just noted that 120% is a bit too high - think about it, with 20% pace deficit you would be lapped within 5 laps from the start of the race!

I see your point now, it was the 107% v 120% typo


I'll correct it.

[Richard]

My analysis only looks at the number of "Ret" noted on wiki. So that includes crashes and car failure. It does not include disqualified.

I did the assessment to follow up my earlier idea that F1 is less spectacular than it used to be. Admittedly off topic compared to the OP

I can't help thinking that lack of variability in the cars is the root cause of less over taking and fewer retirements.

Put it simply, the mechanics are too good!

[gibells]

For me the most interesting racing this year was at the beginning, when every driver was moaning like hell about how little downforce they had. That was when cars were able to drive right up to the back of the leading car without apparently losing any speed or downforce.

Just an observation.

Could it have something to do with the DDDs?

[Brogan]

Hi guys,

I agree the data is a little simplistic at the moment but it is a work in progress

We will be feeding all the data into the database which will then allow us to manipulate it based on number of entries, RET, finishers, by circuit, etc.

That's going to take some time though.

Regards.