A Financial Approach to Machine Learning with Applications to Defaulted Debt Recovery Rates

C. Friedman, S. Sandow, and P. Chang (USA)

Keywords

: Knowlege Discovery, Valuation, Machine Learn ing, Recovery Rates of Defaulted Debt

Abstract

We review a coherent, utility based approach for measur ing model performance and building probabilistic models that learn from data. We apply this methodology to finan cial problems: estimating the conditional probability den sity of trading prices of defaulted debt, given side informa tion and estimating the conditional distribution of ultimate discounted recovery values for defaulted debt, given side information.

Important Links:

• DOI:
• From Proceeding (437) Financial Engineering and Applications - 2004

Go Back