Overtaking in 2012: statistics and analysis

[LionKing]

It is funny that Vettel leads all categories of overtaking this season and still some people try to tell us that he is useless because he cant pass and only wins with the fastest car.

He does only win in the fastest car

It still beats having the best car and coming 4th on WDC ...

[Nando]

on paper at least.

[JimClarkFan]

Never likeD the overtaking statistic, it would mean something if the cars were equal.

As it stands it seems to be a reflection on some kind of bad qualifying or race incident rather than anything meaningful.

Your dislike towards anything RB/Vettel related is so obvious in this one. If alonso/ham would lead the numbers, you'd be full of prasie. Vettel drove past his cars lack of top speed in spa briliantly, whereas webber was stuck behind cars lap after lap.

You know nothing about me, because I wouldn't go wax lyrical about stats like this..

The numbers for overtakes by and large are meaningless because it tells us very little. This should be obvious because we know that Vettel made 20 odd passes in Abu Dhabi, and a similar amount in Brazil and I don't see that there is much skill in just sailing past an opponent who is in a much slower car.

The question is what have we actually learned from these stats? Nothing really is the answer. Vettel made more overtakes than anybody, but it doesn't take into account that he was demoted to the back of field on two separate occasions. It doesn't take into account his poor qualifying at the start of the season when he couldn't turn on the tyres. Lewis made more overtakes than Button, but it doesn't take into account him being demoted to the back of grid in Spain or his several lengthy pit stops, or the fact that he was on pole 7 (well 8,) of the 20 races.

Without being able to factor in the value of each overtake we can know nothing.

The only way I can think of doing this is to divide the points totalled by a driver vs their total overtakes.


And for the most part this is still useless however it is more useful than gross overtakes alone. What we can get from the above numbers is the relative number of points averaged per overtake. The higher the ratio of points won per overtake, the higher up the field the overtake took place, and thus the faster the cars you were racing. What this tells me is that on average Alonso's overtake were much more valuable than Massa's or just about anyone else's.

Yet again this does not take into account all the other circumstances, such as car failures, DNF's, being taken out of the race or who had the fastest car.

It still beats having the best car and coming 4th on WDC ...

So the best car has several DNFs this year, and other problems beside times in a single year? Ham also got taken out in 2 races... some people have very short memories.

[mbvinnie]

And for the most part this is still useless however it is more useful than gross overtakes alone. What we can get from the above numbers is the relative number of points averaged per overtake. The higher the ratio of points won per overtake, the higher up the field the overtake took place, and thus the faster the cars you were racing. What this tells me is that on average Alonso's overtake were much more valuable than Massa's or just about anyone else's.

In terms of analysis I can't think of many more useless ways of representing this in terms of overtaking ability. This is definitely worse than just looking at the number of overtakes, it actually penalizes the best overtakers. And it rewards reliability of car in terms of 'driver overtaking ability' which are absolutely unrelated (i.e. ALO made no overtakes in his 2 DNF's, HAM/VET made many when they got no points, thus lowering their overtaking ability???).
All else being equal with your score, someone who makes zero/low overtakes in your score ranks highly as an overtaker, and someone who makes a lot of overtakes for whatever reason scores lowly as an overtaker. WTF?

You are right that you have to consider other things (being out of position, race pace vs qualy pace, top speed of car etc), but this is absolutely NOT the way to do so.

The real trick is not to put so much weight on any one analysis. The original stats show Vettel in a good light, but it doesn't mean he is the best overtaker. It just one piece of evidence that he seems to do alright, contrary to what some people believe. There are many other things to look at also.

[mbvinnie]

In China, some dude named "Kimi Raikkonen" was overtaken at least nine times on one lap. He's clearly an awful driver, and I don't know why he's allowed to race in "the pinnacle of motorsport."

You are absolutely right that the number of overtakes does not take into account all the circumstances around each. It is quite raw data. But that just means it has LESS validity (not NO validity). Feel free to refine it and analyze it further, but don't just dismiss it.
And it does try and address some issues, like excluding backmarkers, and only counting top 5 teams, and excluding passes where tyres are off cliff (i.e the passes on Raikkonen you mention).
The original post puts too much weight on these statistics, but it feels like you put too little. Bias goes both ways.

I, for one, interpret this information to mean that Red Bull foolishly adopted strategies that required a lot of overtaking. The best and safest strategies are those that put a driver in clear air to make the most of his car's performance. Maybe such poor tactical vision is why Red Bull only narrowly escaped with the Driver's Championship despite clearly having the best car, as evidenced by their easy win in the Constructor's Championship.

??? Is this serious or sarcastic?
Red Bull's strategy is to the extreme of NOT making overtakes. They are all about qualifying high on the grid, ultimate lap time at expense of top speed, and creating and managing a gap.

[JimClarkFan]

In terms of analysis I can't think of many more useless ways of representing this in terms of overtaking ability. This is definitely worse than just looking at the number of overtakes, it actually penalizes the best overtakers. And it rewards reliability of car in terms of 'driver overtaking ability' which are absolutely unrelated (i.e. ALO made no overtakes in his 2 DNF's, HAM/VET made many when they got no points, thus lowering their overtaking ability???).
All else being equal with your score, someone who makes zero/low overtakes in your score ranks highly as an overtaker, and someone who makes a lot of overtakes for whatever reason scores lowly as an overtaker. WTF?

That's not what I am taking from that, and not what I claimed.

I said it perhaps represents better the value of each individual overtake because there is no doubt it is harder to overtake a Ferrari than an HRT. And whilst the ratio is baised it at least takes into account where in the field the overtake was performed. Those with a higher ratio did most of their overtaking higher up the field, that's the point.

Besides, all this makes the point I originally made valid. Overtakes is a stupid method of trying to analyse anything other than who had the most overtakes, which means nothing in any event without context which is not factored into the statistics in any way.

[mbvinnie]

I said it perhaps represents better the value of each individual overtake because there is no doubt it is harder to overtake a Ferrari than an HRT. And whilst the ratio is baised it at least takes into account where in the field the overtake was performed.

It doesn't do that, It rewards people who score a lot of points per overtake. It rewards people who are consistent. It rewards people who make few overtakes. And it rewards people who are in the fastest car as they get the most points.
i.e. Why does A Toro Rosso overtaking a Sauber for 9th not count as much as a Ferrari overtaking a McLaren for 3rd?
The best overtakes are between cars of similar performance, right? That doesn't just happen at the front of the grid.

Your score essentially measures consistency of position (which is the opposite of overtaking), with a bias towards the fastest cars.

An simplified example to show the flaws in your score:
Two drivers in the same car average 5th on the grid. They are both perfectly reliable and have no incidents.
Driver A overtakes 1 car every races, finishing an average of 4th.
Driver B overtakes 2 cars every race, finishing an average of 3rd.
Driver A scores higher in your system (points per overtake)!?! Whereas in the simple 'number of overtakes' system Driver B scores higher.

And if you add in Driver C who starts 12th every race in a poor car, overtakes 6 cars per race and finishes 6th, he would be worse than both in your score.

You have to look at the real world application, your model is not modeling what you think it is modeling.
In the 'perfect' situation described above, the results from your score are inverted to what they should be, so with imperfect data who knows what it is really doing (on top of the criticisms outlined above).

Besides, the original analysis DOES take into account passes only on top cars (which especially counts for the top guys), it excludes backmarkers, it factors in old tyres etc. It is not perfect, but having the 5 categories is better than just the total number of overtakes, even though it can still be improved further with more analysis.

Besides the point, all this makes the point I originally made valid. Overtakes is a stupid method of trying to analyse anything other than who had the most overtakes, which means nothing in any event without context which is not factored into the statistics in any way.

Just because the analysis is not perfect does not make it worthless. The world is not black and white, this is not a binary situation. Number of overtakes is a valid input into evaluating overtaking ability.
There is SOME merit in the numbers, though further refining would be useful (i.e. we can't just exclude the races where VET makes a lot of overtakes, as we would have to exclude other races that were wet/crazy or people were out of position)

My takeaways:
It seems that VET is not a bad passer (not saying that he is a good passer).
Alonso seems to have been very consistent and reliable in 2012 (never racing through the field from the back).
There probably isn't that much difference between the top guys.

My opinion (not backed by anything in particular, and not certain if I can defend it):
-ALO, HAM and VET are all good overtakers, I don't know if any are 'better' than the others.
-HAM is blazingly fast and goes for the overtake ASAP (think Senna style) and uses good strategy to do so (KERS usage etc). Sometimes this hurts him through tyre drop-off etc if it doesn't work immediately and he has had incidents in the past (esp 2011).
-ALO bides his time more (think Prost), often waiting for the time when the overtaking opportunity is in his advantage. Few incidents
-VET is somewhere in the middle (more incidents than HAM, less than ALO). He has some great recovery drives, but sometimes his ability is not as apparent because he is either out front or he is down on top speed (see last 3 years).
(again, just my opinion)

[bhall]

[...]

I, for one, interpret this information to mean that Red Bull foolishly adopted strategies that required a lot of overtaking. The best and safest strategies are those that put a driver in clear air to make the most of his car's performance. Maybe such poor tactical vision is why Red Bull only narrowly escaped with the Driver's Championship despite clearly having the best car, as evidenced by their easy win in the Constructor's Championship.

??? Is this serious or sarcastic?
Red Bull's strategy is to the extreme of NOT making overtakes. They are all about qualifying high on the grid, ultimate lap time at expense of top speed, and creating and managing a gap.

I think that's nonsense, and the statistics agree with me.

"...lies, damned lies, and statistics."

I think our intrepid analyst just decided to flatter his favorite driver. How else could one decide to paint a "true" picture with subjectively collated and incomplete data (unmentioned here but readily acknowledged by Clip the Apex, and I'm not talking about the omission of first-lap passes, technical issues, etc.), an "equal tire" designation that completely ignores strategy (think one-stopper vs two-stopper), and arbitrary sample sizes that ignore circuit to circuit variations in the definition "top five"?

This "analysis" had an agenda from the outset, and it's not the first time we've seen that agenda put on display.

[CHT]

good stuffs...does this stats include overtaking on the starting grid..

[Nando]

Position changes on lap 1

+1,7 - Alonso (+30 in total)
+1,0 - Massa (+20 in total)
+0,7 - Raikkonen (+14 in total)
+0,2 - Button (+4 in total)
+0,1 - Hamilton (+2 in total)


-0,7 - Vettel (-14 in total)
-1,1 - Webber (-23 in total)
-2.7 - Grosjean (-47 in total) (understandable lol)

[mbvinnie]

??? Is this serious or sarcastic?
Red Bull's strategy is to the extreme of NOT making overtakes. They are all about qualifying high on the grid, ultimate lap time at expense of top speed, and creating and managing a gap.

I think that's nonsense, and the statistics agree with me.

"...lies, damned lies, and statistics."

I thought it was widely acknowledged that Red Bull (Seb in particular) used the strategy as I described.
Can you clarify what you think their strategy is? How do the statistics agree with that?

I agree the original stats probably had bias in how they were presented, but they were still partially useful.
And it's still the most useful analysis so far on this thread in terms of a starting point.

[mbvinnie]

Position changes on lap 1

+1,7 - Alonso (+30 in total)
+1,0 - Massa (+20 in total)
+0,7 - Raikkonen (+14 in total)
+0,2 - Button (+4 in total)
+0,1 - Hamilton (+2 in total)


-0,7 - Vettel (-14 in total)
-1,1 - Webber (-23 in total)
-2.7 - Grosjean (-47 in total) (understandable lol)

Can we agree that it looks like Ferrari has the best starts/first-lap performance, McLaren are neutral, and Red Bull the worst starters amongst those 3 teams?
Or is it coincidence that they pair up like that and it is the drivers who are responsible?

[mnmracer]

I think that's nonsense, and the statistics agree with me.

"...lies, damned lies, and statistics."

I think our intrepid analyst just decided to flatter his favorite driver. How else could one decide to paint a "true" picture with subjectively collated and incomplete data (unmentioned here but readily acknowledged by Clip the Apex, and I'm not talking about the omission of first-lap passes, technical issues, etc.), an "equal tire" designation that completely ignores strategy (think one-stopper vs two-stopper), and arbitrary sample sizes that ignore circuit to circuit variations in the definition "top five"?

This "analysis" had an agenda from the outset, and it's not the first time we've seen that agenda put on display.

The funny thing with you is that you are always ready to dismiss anything that doesn't fit your world view, yet you are never ready to substantiate your claims.

"I can't say why it's wrong, or how it could be better, I just know it's wrong."

It reminds me of how little children discuss: na-uh! why not? because!

[Nando]

Position changes on lap 1

+1,7 - Alonso (+30 in total)
+1,0 - Massa (+20 in total)
+0,7 - Raikkonen (+14 in total)
+0,2 - Button (+4 in total)
+0,1 - Hamilton (+2 in total)


-0,7 - Vettel (-14 in total)
-1,1 - Webber (-23 in total)
-2.7 - Grosjean (-47 in total) (understandable lol)

Can we agree that it looks like Ferrari has the best starts/first-lap performance, McLaren are neutral, and Red Bull the worst starters amongst those 3 teams?
Or is it coincidence that they pair up like that and it is the drivers who are responsible?

Yea looks to me Red Bull have trouble keeping their positions at the start despite having a faster car.

Also if they would have been included in page 1 post the results would be quite different with Alonso probably taking the top step.
And rightfully so as we have seen his magic on lap 1.

[mnmracer]

Position changes on lap 1

+1,7 - Alonso (+30 in total)
+1,0 - Massa (+20 in total)
+0,7 - Raikkonen (+14 in total)
+0,2 - Button (+4 in total)
+0,1 - Hamilton (+2 in total)


-0,7 - Vettel (-14 in total)
-1,1 - Webber (-23 in total)
-2.7 - Grosjean (-47 in total) (understandable lol)

Can we agree that it looks like Ferrari has the best starts/first-lap performance, McLaren are neutral, and Red Bull the worst starters amongst those 3 teams?
Or is it coincidence that they pair up like that and it is the drivers who are responsible?

Yea looks to me Red Bull have trouble keeping their positions at the start despite having a faster car.

Care to substantiate how Red Bull is a faster car on starts?