A novel portfolio selection model based on fuzzy goal programming with different importance and priorities

• A novel fuzzy portfolio selection approach is proposed.
• This approach is based on the fuzzy goal programming techniques.
• Portfolio return, risk and beta coefficient are defined as the fuzzy goals.
• This approach allows defining certain importance and priority among fuzzy goals.
• The proposed approach considers the risk preferences of investor and market trend.

Despite the risk–return tradeoff is main concern of financial theory; the rational investment decisions requires considering many criteria simultaneously. In addition to determining a certain importance and priority among these criteria, modeling the investor behaviors in accordance with market trends provides much more realistic approach. However, the researchers mostly overlook to evaluate these concepts simultaneously. This article introduces a novel fuzzy portfolio selection model that takes into accounts the risk preferences in accordance with the market moving trends as well as the risk–return tradeoff, and allows the decision makers to define a certain importance and priority among their objectives. To construct this model, firstly the portfolio return, risk and beta coefficient are assumed as main objectives including the possibilistic uncertainties. To define possibilistic uncertainty, the specific fuzzy membership functions are constituted for these objectives with respect to the risk preferences of investors and market moving trends. By means of the fuzzy goal programming techniques, a novel portfolio selection model is developed using these specific fuzzy membership functions. In the application section, three investment terms are examined in the Istanbul Stock Exchange National 30 Index. While ISE30 index has the upward (bullish) and the downward (bearish) moving trends in the first two implementations, the third implementation includes a scenario in which the investors desire to chase the ISE30 index. In the analyses, the proposed model is compared with the classical Mean–Variance, Mean-Absolute-Deviation and Maxmin models in terms of their portfolio returns based on the selling prices in the test periods. As a result, the proposed model gives superior performance than the classical models because it takes into account the investor preferences in accordance with market moving trend.