Evaluating investment values of stocks using a fuzzy TOPSIS approach

Abstract

This study presents a fuzzy TOPSIS (technique for order preference by similarity to ideal solution) approach to evaluate the investment values of stocks, where ratings of stocks versus qualitative attributes and weights of all attributes are assessed in linguistic values represented by fuzzy numbers. The quantitative attributes are transformed into dimensionless indices to ensure compatibility with the linguistic ratings of qualitative attributes. The membership function for weighted rating of each stock versus each attribute can be clearly presented. To avoid complicated calculations of fuzzy numbers, all weighted ratings are defuzzified into crisp values. The ideal and negative-ideal solutions can then be easily obtained to derive the relative closeness in order to complete the fuzzy TOPSIS model. The relative closeness is the evaluation value of a stock. A higher evaluation value implies a better performance. A P/Ez.star index is further defined to facilitate decision making, in which the lower P/Ez.star implies a lower risk. The strategy of selecting a stock with a higher forecasted evaluation value and a lower P/Ez.star is recommended. Finally, an empirical example of sixteen commercial banks demonstrates the effectiveness of the proposed model