Abstract:

No-Arbitrage pricing theory has its uses, but recently it has been helpless to explain real prices of market securities. In this talk I will explain some of the sources of market inefficiency and illustrate the current problem with an example fixed-income arbitrage trade that was available recently.

While inefficiencies have always been present in illiquid markets, the recent credit crunch has produced mispricing in even the most liquid markets (e.g swap spreads). To a large extent, this broad market failure followed from unwarranted faith in the models used by banks and rating agencies to value and account for the risk of CDOs. Too much leveraged money was invested in the AAA tranches that weren't there.

In this talk I present a different approach to the problem of analyzing correlation products (including CDOs) that makes use of techniques from discrete geometry. I compare the industry-standard copula approach with an alternative maximum entropy model. The question of which model should be used to analyze CDOs is not settled, but the approach presented here seems promising.