Modeling the CDX Index using neural networks
Abstract
Because of their ability to learn nonlinear relationships and the ability to deal with uncertain or insufficient data, artificial neural networks have become important tools in finance. They have been successfully applied to areas such as predicting the stock market performance which inspired this thesis. The objective of this thesis is to see whether an artificial neural network is a good method to model credit derivatives like the credit derivative index. The final model built uses tranche quotes on the standardized tranches 3-7%, 7-10% and 10-15% as input variables to predict the CDX Index. With limited data, the model has a training root mean square error at 5.43 and out of sample testing root mean square error at 6.82 which are both quite small. Also, we demonstrate that the model can accurately grasp the future trends of the index.
Degree
M.S.I.E.
Advisors
Morin, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.