Partial Order Based Approach to Preference Assessment in Portfolio Selection

The paper presents a new approach to encompassing investors' subjective preferences in the portfolio selection problem. The method is based on dividing the investor's assessment into separate perspectives and assigning a single linguistic score for each of them. Investment assets, represented as vectors of such scores, are compared using proposed operator. This induces a partial ordering of assets for the portfolio. Local Partial Order Method (LPOM) is used to obtain average heights in partial order that are used for decision making. The proposed model differs significantly from the average value approach since it allows for non-numeric interpretation of scores and encompassing behavioural factors as well as subjective knowledge and experience of the investor.