If you want, I can run stock premium cars along trial mountain (best test circuit in my opinion because it has elevation, curves, straights, and most importantly short) all with racing soft tires and share lap times, maybe run a regression between performance points and lap time or performance points and power to weight ratio, etc. Other variables that I think could be used: wheelbase, peak horsepower to peak torque ratio (a measure of power-band efficiency), torque to weight ratio, cost in credits. I am guessing a multivariate model would be best.