Fixed vs open setups

Andrew Bakke made a rough analysis about Open and Fixed setups and their influence or suitability for an overall mid pack driver. Please bear in mind his main intent is to open a debate about this question.

Prompted by the questions in this thread, I asked myself whether we had any data as to what the effect of fixed setups was on the gaps between drivers. I think I found the answer.
Until now a prevalent thought was:

“I think people prefer fixed racing because their lack of setup knowledge does not become a hindrance in these races. Fast drivers will always be fast drivers, but the stats show that a fast driver with a fast setup is faster than the same fast driver with a slower setup.”

I saw “the stats show” and “I don’t have hard data”, so that triggered the engineer in me to see if I could find data.

Careful, I’m going to do math!

So we have two questions to answer:

1. Is “a fast driver with a fast setup is faster than the same fast driver with a slower setup”? (the gap between the fast drivers is bigger in open than fixed)
2. Is “the difference between the best and worst is smaller with a fixed setup”?

We can’t get easy Qualifying data anymore, and actual race laps are hard to parse, but we can get TT data. It’s not perfect data, but it might be good enough. To compare, I figured Ruf GT3 and Blancpain Sprint would be a good comparison, two series with GT3 cars, one of which is fixed setup. Possibly the Open numbers could be affected by the BoP, and the GT3 results may not apply for all cars, but these numbers are more for personal interest, rather than incredible detail.

Our dataset then is 89 drivers setting a TT time in a fixed setup GT3, and 117 drivers setting a TT time in the open setup GT3 series. The attached box and whisker plot below shows the results. There were 11 outlier times with the fixed setup (12.4%), and 8 outlier times with the open setup (6.8%).

This data points to two conclusions:

1. Yes, the gap between the good drivers is larger with the fixed setup. The difference between the fastest driver and the first quartile driver increased from 1.68s in fixed to 2.21s in open. The gap at the front of the field increases with open setups.
2. No, the gap between the best and the worst drivers is not smaller with a fixed setup. This one intrigued me, but somewhat matches the anecdotal evidence of fixed setups not matching a player’s driving style. While the gap in the first quartile increased with an open setup by about half a second, the gap in the second quartile decreased in open setup slightly (0.07s) and the gap in the third quartile decreased by 0.75s (from 1.90s to 1.15s).

This is interesting data, because it means that for approximately half of drivers they are more competitive when they can modify the setup than when they can’t, presumably because their gains in consistency with a comfortable setup outweigh the the gains of fast guys with fast setups. To put it another way, Fixed setups make the front-runners more competitive, but Open makes for a more competitive mid-field.

I think that makes for a very interesting twist in the debate about Fixed setups. Both because it seems to indicate that Fixed isn’t benefiting the people it’s proposed to help (the slower drivers without the ability to make setups), and because it indicates that fixed may hurt the slowest percentage of drivers more than it helps them (again, potentially because they can’t pick a stable, comfortable setup). The argument I keep seeing is that it’s better for the slower drivers who can put in less effort, but it seems to me that if drivers in a series are sharing setups with the newbies that it’s making them faster and it’s actually only the fast but time-crunched drivers who really benefit from fixed (again, at least for GT3 road racing, oval might have a completely different answer). Perhaps more to the point, fixed setups only potentially gain you something if you think the only reason you’re not winning in the top splits is because they’ve spent more time on setups.

So, use this information for what it’s worth. If anyone has an alternative dataset they’d like to see, let me know and I’ll try to run those numbers as well.