Jun 13, 2005

Facts Emerge About Spring Hedge Fund Losses

From the FT:
GLG Partners, Europe's largest hedge fund manager, has admitted that flaws in its trading models were partly to blame for a 14.5 per cent drop last month in the value of its Credit Fund.

In particular, it has acknowledged that the mathematical model it used to price complex credit derivative products failed to foresee market swings after last month's ratings downgrades of General Motors and Ford.

In a private letter to investors, a copy of which has been obtained by the Financial Times, the hedge fund argues that it has now rejigged these trading models. The admission is significant because other banks and hedge funds appear to have been using similar trading models, meaning that they may also have suffered big derivatives losses.

GLG warns that conditions in the credit market could remain difficult for some time because banks and hedge funds are trying to get out of loss-making derivatives positions at the same time.

"Segments of the hedge fund community and a substantial number of investment banks are nursing material losses in structured credit trades as a result of recent market conditions," the letter says.

I recommend this as some background reading on the current state of hedge fund risk models. The comments about Mandelbrot and Eckhardt seem most relevant.

The FT goes on to say that while the GLG's models considered a move like last month's (May?) an 8 sigma event that could be "ignored". This quote in particular seems like the root of problems: "GLG blamed the model's shortcoming on the fact that this CDO market had only traded since last year...". Personally I bet the model did exactly what it was built to. Generally they do that.

Maybe the CDO market is small enough and has had its time in the sun but with the VIX still tripping along sub 15% I find this kind of explanation troubling.

1 comment:

  1. eight sigma events consistently seem to happen more often than people expect

    brad setser