Bad math hurts...

An engineer's guide for understanding the subprime mortgage mess.

Investors love these higher interest low risk bond packages. So the banks start to put together more of them. And they realize that they can cobble together even higher yield bundles if they cascade them: they take a bundle of 10 different kinds of s--- loan bonds, and turn them into high risk loan meta-bonds. Then bundle 10 different high risk loan meta-bonds (which were formed from the s--- loans) into medium risk meta-meta-bonds. Then bundle 10 different medium into a low. And so on. So now you've got something labeled as "high quality, low risk" which is formed completely from a collection of s---. But, by golly, it's marketed as high quality, and it's got a nice rate of return! And it's really low risk, because you've distributed the risk into a bunch of independent places, and you're only in trouble if they all fail. (This is bad math point number one: false independence. This scheme only works if there is truly no connection between the failures of the components.)

You bundle them together according to a complicated scheme called tranching to make them look like they're really, truly safe. And you take out insurance, so that if something goes wrong, it's all insured. [More of a helpful - albeit scatological - explanation of the credit crunch trigger]

I have read several articles trying to explain simply what happened as mortgages were massaged into more exotic investment packages. This one helped more than most.