PI project-Now Testing!

[fndrbndr]

I tried another car today.  I'm developing a C-class small sportscar/roadster spec, in which every car is under 200hp and 2000 lbs.  I tried to build a Ferrari Dino 246 GT for the spec, and it fit, but came to a top PI of 537.  I ran it, though, and it beat all of the other cars in the QR, and did so quite handily.  Anyway, I also ran Suzuka, so I could use it for this project.  The time was 138.450.  It beat my Renault Turbo 5 ghost rather handily, and is close enough to the top of the class that I may try to run it online.  BTW, it was the top car (among the ones I've run QR for, mostly grip/flowing cars) at Sebring Short and was competitive at Tsukuba and RAS.  Very interesting....

[fndrbndr]

So far, PI tends to predict laptimes pretty well.  The PI alone explains 96.2% of the variance in the laptimes in the sample thus far.  That leaves about 4% to driver skill and variation within the cars that doesn't fit into PI.  It's not a full list by any stretch, but not too shabby either!