Talk:Latent variable portfolio credit models

From FinMath

Some notes.

1. I'm not aware of a really good taxonomy of credit models, as many popular terms differentiate between mathematically equivalent or similar frameworks. To be trite, these are all special cases of Bayesian networks.

2. Practical demands suggest the use of conjugate priors for latent variables or frameworks admitting efficient updating via the Kalman filter and its generalizations. Probably these belong in Dynamic portfolio credit models

Pcotton 22:29, 24 July 2007 (UTC)Peter