Optimum Return on Share Market Investment through Genetic Algorithms

Predicting stock investments with traditional time series analysis and other statistical methods have proven to be difficult, since the vast amount of data and scripts are handled. Moreover, the data available in the Indian stock trading are not fixed and always uncertainty and ambiguity in nature. Therefore, the large number of researchers has proposed many new methods/models, which are based on the non-traditional tools like Fuzzy Logic, Neural Networks etc., The neural network is used to find the optimum number of iterations is required to train FIS rules. Again, for increasing the efficiency of the investing model, we trained these FIS rules by Neural Network (called ANFIS) and to reduce the scripts by a smaller number of scripts, so that one can manually picked the optimum rule. The disadvantage of this method is that the invested amount goes only to a limited number of scripts, it resulted that it will be automatically increased the risk. In this chapter, all the rules established from ANFIS are taken into account and genetic algorithm is used.

To find the order (sequence of preference) in which one can invest the amount, if the investors want to invest their amount in a limited subindexes and which is different from the approach given by the authors.[Abraham (2003), Alemdar (1998)].

To find the sequence of preference together with the investing amount preference like high, medium, and low. (In these cases the investors have to invest their amount to the entire sub-indexes with the amount preference).

In this contest, two genetic algorithms are simultaneously executed, one will take care of arranging the sub-index preferences and the other will take care of amount of investment like high, medium, and low. This will help the investors to invest their amount, based on the preference given by genetic algorithm, to all type of sub-indexes. Another advantage of this chapter is to establishing the fitness function, which include the ‘weight value of the sub-indexes’ and the ‘weight value of the risk’. Finally, the designed model is tested and implemented through BSE stock market index.