| Summary: | 碩士 === 國立彰化師範大學 === 企業管理學系 === 98 === The well-known mean-variance model cannot satisfy investors’ request for different investment preference and risk diversification. Consequently, we consider genetic algorithms for portfolio selections which consider risk preference including return, risk, liquidity, return distribution and transaction cost. Further, we try to improve the Markowitz model by multi-objective genetic algorithms (MOGAs). Why we used MOGAs? Because of MOGAs have considered all the objectives in the same time with solving quadric programming problem and optimized the solution in globally pareto optimal. Moreover, Multiobjective functions are prior than single objective because of solving the conflicts exquisitely in complex objections. Multiobjective genetic algorithms (MOGAs) can explain the trade-off between return and risk which behavior finance investigates. This paper proposed method which incorporate different risk measures, skewness, entropy, liquidity and transaction cost. A trading example is also illustrated to compare with the proposed method. On the basis of the numerical results, the method we proposed can provide a higher return on asset and having better risk diversifications.
|
|---|
