> For example in 2008 the CDS market alone was at some point "valued" at close to $36 trillion [1], while for comparison the US GDP back then I'd say was around $14-15 trillion
People say this all the time, but regardless of whether or not the CDS market was or wasn't a shitshow at the time, comparing the outstanding notional to GDP is meaningless.
At the time, the CDS market worked like most derivatives markets: you face a small set of players and you are constantly offsetting existing transactions and subjecting them to netting rules on a master isda.
So like, let's say Foocorp has 100m bonds out there. I want to buy protection on 1m of this (basically, I want to buy default insurance). You agree to sell this to me. Ok, there's 1m in CDS contracts' notional value outstanding.
Now, our views, other risks, whatever changes. Maybe I sell the bonds, and holding this CDS is too risky for me now because it's just a naked short of bonds I'm not long. Maybe you buy a bunch of these bonds, and now you're not only facing the default risk of the bonds but paying the default insurance to me.
So you decide to buy another CDS offsetting your existing one. I'm also trying to flatten out here so I sell a CDS offsetting my existing one. Now (mutatis mutandis) we're both at 0 CDS risk. We each have entered into 2 swaps, they're directly cancelling.
But the total notional? Now it's 2m. Contracts expire, so the number doesn't strictly go up, but given how the products are bought and sold, it mostly does. It would be like looking at the stock market not in terms of the market cap of all the companies, but as the dollar value of all the trades made in a certain time period. It's not that this number is meaningless, it's just that it does nothing to explain how much is "at risk".
Of course in reality this is all much more complicated, but the basic principle holds.
Anyway, I know this isn't your main point.