Icelandic CDS Datapoints of the Day

The CDS auctions for Iceland's failed banks are all over now, and the results are up at creditfixings.com.

For an explanation of what this all means, see here. But these numbers are definitely weird. Here's just a few questions thrown up by them:

1. After the first round of the auction, the open interest is always to sell -- there are more sellers than buyers. Yet after the second round, in three of the six cases the settlement price has gone up rather than down, as you would expect. Why would it do that?
2. The clearing price for Kaupthing's debt -- which, we can assume, is the market's best guess as to its risk-adjusted recovery value -- is significantly higher than the other two, despite the fact that Kaupthing was the most leveraged of the three banks. How come? All I can think of is that the market was concentrating on Kaupthing's UK assets, such as investment bank Singer and Friedlander, while in the case of Landsbanki the market was looking more at the UK liabilities, from the bank's Icesave operation.
3. Why was the inside market midpoint for all three banks' subordinated debt so high -- and how come there were people willing to pay 2.375 cents on the dollar for Kaupthing's sub debt in particular, when the senior debt is trading in single digits? After all, the senior debt has to be paid out in full, at par, before subordinated bondholders get anything at all.

I suspect that the answer to all of these questions is something I like to think of as the Law of Small Numbers: in financial markets, assets always behave very weirdly when their price approaches zero. Clearly, the CDS market is no exception.