Using economic and financial information for stock selection

A major inconvenience of the traditional approach in portfolio choice, based upon historical information, is its inability to anticipate sudden changes of price tendencies. Introducing information about future behavior of the assets fundamentals may help to make more appropiate choices. However the specification and parameterization of a model linking this exogenous information to the asset prices is not straightforward. Classification trees can be used to construct partitions of assets of forecasted similar behaviour. We analyze the performance of this approach and apply it to different sectors of the Samp;P 500