The joint's a little more complex than that. Consider this from the American Academy of Orthopaedic Surgeons website:
"For simplicity, the knee is considered a hinge joint because of its ability to bend and straighten like a hinged door. In reality, the knee is much more complex because the surfaces actually roll and glide as the knee bends."
The joint is not a pure hinge, because it allows a fair amount of rotation and gliding of the tibia on the femur. Also, the upper surface of the tibia is more flat than concave.
There are some slight concave depressions on the top of the tibia for each of the femural condyles to fit into and the femur does glide across the top of the tibia a little bit, but there really is no "track" from a bone standpoint, as the tibia can still rotate slightly.
The reason the knee is the most injured joint in the body is because it is essentially a round femural end sitting on top of a mostly flat tibialar cartilage surface. The only thing holding it in place firmly is ligament and muscle.
The sense of fitting into a track in flexion is actually caused by a tightening of the knee ligaments. What happens is that at parallel, the knee's ligaments are at their slackest. As the knee passes 90 degrees into greater flexion, the ligaments tighten up, greatly stabilizing the knee and reducing the amount of rotation experienced between the fibia and the femur.
This also happens as the knee extends and approaches full lock out. The knee ligaments tighten up and actually lock out the knee with a slight twist, usually referred to as the "screw home" mechanism.
So this is the general reasoning that most people cite for going beyond parallel, saying that the knee is much more stable past 90 degs due to the ligaments tightening up. And it's maybe also a reason why you're not supposed to fully lockout your knee at the top of a squat, because of that last slight bit of rotation at full lockout and hyper extension.
I found a decent site that shows how the surfaces of the knee glide and rotate across each other during flexion and extension. http://moon.ouhsc.edu/dthompso/namics/kneeak.htm