@Monzstar83Hey mate,
I always use my race pratice program (RPP from now on) to both calculate tyre and fuel for the race and the AI level - so I do them a lot. Like, a real lot haha
What I feel doing this since F1 2019 is that the output depends strongly on the track's length and average speed. For example:
Tracks with regular length and average speed (Bahrain, Spain) or fast short tracks (Brazil, Austria) if I let the delta to be +0.3s on the RPP them the race would be ok. Long fast tracks (Spa) this delta needed to be raised to ca. +0.5s. Slow tracks (Monaco, Singapore) I could let this time fall to +0.1s and I'd still do alright.
On F1 2021 I did it meticulously plotting tracks' length x AVG speed, so I kept the AI fixed throughout an entire season and felt if it was easy, ok or hard. That's how I came to those deltas. It's tiring and one likely wastes a season doing so but it's rewarding from there on. Comparing throught 19-22, I noticed that I could keep the deltas though the AI absolute values would change.
This graph is lost somewhere in GDrive, if I find it I bring it to you.
Bottom line, I feel it depends heavily on expected cornering efficiency (braking and exiting) and its cumulative effect on the track layout (ammount of corners and their speed and tracks' length).
So, I know I don't directly approach what you asked but I hope it brings some insight into it.
Cheers,
Dan
edit: @SexyBrigadeiro, I added small info since your reply.
Found it.
