I really shouldn't be doing this now.

What with my CPDO arch-nemesis off communing with the penguins, it seems downright ungentlemanly. And I really ought to be working for the man, you know, the one who actually pays me, just now.

But there were some particularly interesting comments to some of the recent posts on CPDOs, and I thought them worth highlighting.

Responding to an earlier post, commenter P. K. Koop notes:

...I would expect the barrier implicit in the 15X leverage limit to act as a target or safety net for those trading against the CPDOs.

This reminded me of an interesting post from Cassandra Does Tokyo, "Amaranth: Was It The Market?":

But there is a... possibility that is understandably NOT discussed in the mainstream media, but surprisingly is not discussed in the trade press either. And this is the possibility that [Amaranth's] clumsy and quasi-public long Natty position was the subject of predatory trading by those with material non-public information about the Fund and it's positions...

Roger Lowenstein's account, When Genius Failed reconstructed the scenario pretty well. Essentially, if you're very leveraged, once someone sees your positions, you're a target. Hillenbrand was seemingly the only one who really understood this risk. He made sure they used multiple Prime Brokers, swapped positions between leverage providers to insure no one saw the full extent of their leverage or their positions. If one cannot be certain as to whether one has an offsetting position at another shop, the risk-reward equation for "gunning" is greatly reduced. After LTCM started to take a hit, and needed either new capital or bigger lines, anyone who might supply the credit that was needed also needed to see "the position". All the Positions. He fought it, but there was recourse, and that was the precise point at which Hillenbrand knew they were dead.

Suppose that ABN's pioneering CPDO issues grow popular and are widely emulated, as investors snap up what seems to be no-risk extra yield. Suppose also that the imitators are not innovators, that they all adopt the same basic strategy in structuring their instruments. Then won't CPDO-owners collectively be something like a large hedge fund whose portfolio, strategy, and response to changing market conditions is fixed and published in advance? If so, what would prevent other funds from taking advantage of the information assymmetry to intentionally break these structures?

It wouldn't be cheap, and it wouldn't be easy. But it might be possible. If so, we'd have an illustration of the signature irony of finance, what Patrick Hynes and David Post have dubbed the "reverse tinkerbell effect". The fact that so many people believe the rating agencies' models will have created the conditions under which those models prove to be unreliable.

The wisest commenter on the internet, the prolific "Anonymous", made an interesting point in response to one of Felix Salmon's posts:

If AAA covers all bonds with a default probability below X, "natural" AAAs will be randomly distributed within the range while synthetics will likely be skewed upwards toward X because the banks have more choice in achieving a rating than governments or corporations. There is some evidence that synthetics have higher default rates than similarly rated naturals. It's a bit like Goodhart's law.

Goodhart's Law could be considered an application of the reverse tinkerbell effect. Anonymous' observation is a financial analog to this recent result regarding political-science academia. [Okay, that's a blogging of the result. The original paper is here.]

Some other CPDO links:

• CPDO For Dummies on the excellent Alea.
• Mark Gilbert at Bloomberg is much harsher than I was: CPDOs Are Derivatives Market's New Alchemy Trick
• A nice Financial Times article: Questions lie behind CPDO hype
• More from the FT, fairly harsh: John Dizard: Challenging nature's forces [Hat tip: Felix Salmon]
• A very insightful thread on Wilmott [Hat tip: P. K. Koop]
• A very insightful thread on NuclearPhynance
• An ABN-AMRO presentation of a CPDO product. [Hat tip to commenter Holmes on the NuclearPhynance thread above]

Relevant to nothing: Last week while I was writing about CPDOs, the other project I was working on rhymed. With CPDO. Not so easy.