I'll start this with the caveat that results vary from car to car, obviously, but I've been experimenting with changes in track friction and angle to mitigate spinouts (leading to cars reversing down the track) and encourage drift cornering.
The biggest challenge is that cars are travelling under gravity rather than any self-motivation, therefore weight distribution in the car has the greatest effect on their cornering stability. That said it is very possible to alleviate spinouts (I'm using it as a catch-all term for a car changing its orientation by 90° or more during a run) by altering either the friction of the track surface, the direction of gravity or a combination of the two. By applying a higher-friction surface texture to the outside edge of a corner you can reduce the likelihood that the back end of the car takes the easier route, spinning forward. Typically, the rear of the car comes forward when corning if the rear has less resistance than the front - in banked corners this will frequently assert itself if the incline of the bank causes the front of the car to make contact with the inside of the curve.
On flat corners however you can force this effect by increase resistance on the inside of the curve, providing a low friction outer surface, forcing a "drift" effect. The challenge here is in the entering and exiting of the corner to ensure that momentum remains pointing forward. Obviously, grooved tracks in the desired direction of travel help keep cars pointing the right way, but changes in surface texture can have a more profound effect in cornering. I've been using sandpaper for these experiments so far as different grades of paper yield different results. I'm interested in whether anyone else has experimented in a similar way.