The ratio against Prost is 70% (14-6). In qualifying I have the same as you.
Yep, my mistake. I accidentally used all 32 races as the divisor instead of the number of joint races finishes.
"So do these statictics extrapolate the quality of the teammates completely accurate in your opinion?" Completely accurate no and not the quality of the teammates, but more how they performed in that specific time together.
The intention is clear, but i still doubt, that the numbers against all teammates combined (and also sometimes isolated against a single teammate) are precise enough to represent the real hierarchy between all these drivers as there are still many potential distorting factors (that must not necessarily be true for "Senna vs. teammates" or "MSC vs. teammates", but my critique was more related to this kind of statistic in general).
- What if one driver had just 3 joint race finishes in a season with some trash can, whereas another driver had 10 instead? That would polish up the statistic (overall ratio against all teammates) of the latter more than that of the former, despite the fact, that both would´ve beaten their respective teammates 100% of the time.
- What if all of the races, in which driver X was slower than his teammate, were also joint race finishes at the same time, whereas driver Y had more luck as in 2-3 races, in which he was slower than his teammate, either he himself or his teammate retired?
- Does this statistic really show all aspects of the abilites in the races or qualifying? Driver X reaches 110% of a certain performance level in 85% of the time, but has also 2 or 3 off days and gets beaten by his teammate. Driver Y on the other hand reaches 100% all of the time and has zero off days. Which driver did better? According to the statistic driver Y as it focuses on consistency (even if the data would be completely acurrate).
To demonstrate this on our two examples:
MSC has just 3 joint races finishes against his poor teammates of 1994. But Senna has 5 against Dumfries and 7 against Nakajima.
But the more striking example is this: Schumacher has a pretty poor ratio against Martin Brundle (4-3 or 57%), but that doesn´t represent their respective race performances at all. Out of all the season´s races Brundle was only ahead in two occasions due to his pace/strategy (Silverstone and Canada) or 3 if include also MSC´s mistake/mishap in Monza. As you can see Brundle was pretty lucky statisticwise, that his "wins" are overrepresented.
Of course one general idea of such a statistic would be, that such things even out over the course of a whole career, but these numbers are still pretty small from a mathematical p.o.v.:
How many joint races finishes did Senna have e.g.? I guess not even 100, probably more like 80 or lower. So, if the statistical chance had altered just 5 of all these joint race finishes, the ratio would already deviate by more than 5 percentage points...
About the problems. Thats why I count the % of retirements not related to car failures. That way you are penalized in one place or another depending if you kept racing or retired.
About the field together. I'm not very sure, after 2004 I certainly can say that, but before that is hard to say.
About the completely normal circumstances I would say Senna only lost twice in his entire career. Mexico 88 and France 88 against Prost.
- Alright, but that doesn´t change the fact, that these incidents still have a influence on the other statistic, if some are terminal and some only cause, that you fall behind your teammte.
- Ok. This point is not really important, so to make it short: It can perhaps be seen best on the basis of the gaps in qualfying. They gaps got closer throughout the late 90s and early 2000s...already before 2004.
- I guess, you meant examples, that were also considered in the statistic (not also Adelaide 88?), cause i remember some (not many of course) races, in which e.g. Berger was better, but either he or Senna didn´t make it to the finish (of course that´s true for every other driver, as i pointed out above).
Edited by LiJu914, 06 December 2012 - 12:57.