As I said earlier, there are exceptions. But, in the case of teen driving, relatively few. And, you never really know if any particular kid is one of these exceptions until he's on the backside of adolecense and makes it to 21/25.
But, the individual kid isn't the only issue here. How they interact with other teens makes a significant difference too.
As a group, teens are the ONLY driving group where carrying a passenger actually increases the risk that they'll have a crash. For ALL other groups, the crash risk decreases whenever they are carrying passengers. Those are the statistics.
Not only that, but teens are much more likely than other groups to be carrying passengers, and more likely to have other teens as passengers than any other group. Studies have found that their natural propensity for bad decision making actually compounds itself when groups of teens are together. Look at the M5 incident where there were 4 teens involved. How many of those passengers spoke on the side of caution vs how many who urged the driver on? Actually, I hate to label their decision making as bad, a better word is undeveloped.
Not only do teens lack an experience database on which to measure their decisions, but studies have shown that the development of the frontal lobes of their brains (critical in decision making) isn't fully completed until in their 20's. In many ways, they are simply incapable of consistent good decision making. This is especially true when those decisions are split-second, as is often the case in driving. They may in fact recognize a threat, and are armed with acute reflexes. But the whole process breaks down when they are forced to analyze and formulate corrections in a timeframe adequate to avoid the threat.
Sorry, IMHO, it's a mistake to give a kid the keys to a Boxster (or any car whose capabilities or flavor suggests fast driving and risky performance) and let them go unsupervised, and history and statistics backs me up.