I said mortgage backed securities like mortgage CDOs are bespoke because the assets behind them aren't really fungible. Two CDOs might both derive their returns from home mortgages in the general Las Vegas area, but their return flows won't come from the same houses. This is very different from stocks, where any two shares in some company are equivalent (provided they're of the same share class). When pricing ABS, people try to pretend they are interchangeable, which is where the fancy computing and dodgy modelling comes in.
My use of the word 'bespoke' here is somewhat poetic. CDO issuers are more likely to build a CDO and try sell it than they are to make them to custom specifications. (Investment banks did make custom CDOs & CDSs, of course. If an investor calls up and says "can you find X, and sell me some", you find X and sell him some.
And yes, you can't just buy an MBS through your broker (unless you're incredibly rich). You can invest in mutual funds that own a wide variety of MBS, of course. But these things just aren't sold to retail investors. This is not so much because of the size of the assets involved -- you can slice off and sell as small a slice of the assets as you want -- but because the fixed costs of arranging a sale are high. You need to pay lawyers who know what the hell they're doing to set up the legal arrangements for these contracts. It doesn't make sense for Uncle Stocktrader to drop $$$$$ just so he can buy a $$$ slice of a CDO. Much easier to pool the costs in a mutual fund or a money market.
I guess my central confusion is that I thought the entire reason for bundling mortgages into some kind of aggregate object (I'm going to drop the financial terminology, as I'd probably use it imprecisely) was so that they could transform it into a fungible asset and therefore have access to a broader market of customers who were interested in buying it.
In other words, while retail investor might not be interested (either through a mutual fund or something else) in owning these 3 mortgages, they might be interested in owning a smaller share of 1,000 aggregated mortgages. And furthermore, that you could alter the criteria those 1,000 mortgages were selected by to create different products (offering different rates and risks). And furthermore, that you could attempt to value add (and I thought this was where it went off the rails) by packaging junk mortgages in some statistically magical way where you could sell the aggregate for more than the total value of its individual components.
So, in essence my understanding was something like (start at top, trickle to bottom):
Actual mortgage
Basic mortgage backed security (no value add)
Complex mortgage backed security (value add)
Retail mutual funds / investment bank consumers
(Which, as you've said, still has nothing to do with HFT, as given the nature of these things there probably isn't much pricing variance day-to-day, and therefore has little interest to that community)
My use of the word 'bespoke' here is somewhat poetic. CDO issuers are more likely to build a CDO and try sell it than they are to make them to custom specifications. (Investment banks did make custom CDOs & CDSs, of course. If an investor calls up and says "can you find X, and sell me some", you find X and sell him some.
And yes, you can't just buy an MBS through your broker (unless you're incredibly rich). You can invest in mutual funds that own a wide variety of MBS, of course. But these things just aren't sold to retail investors. This is not so much because of the size of the assets involved -- you can slice off and sell as small a slice of the assets as you want -- but because the fixed costs of arranging a sale are high. You need to pay lawyers who know what the hell they're doing to set up the legal arrangements for these contracts. It doesn't make sense for Uncle Stocktrader to drop $$$$$ just so he can buy a $$$ slice of a CDO. Much easier to pool the costs in a mutual fund or a money market.