Stock price prediction using genetic algorithms and evolution strategies

Abstract

Stock market is a very challenging and an interesting field. In this paper, we are trying to predict the target prices of the stocks for the short term. We are predicting the target priceof script individually for eight different scripts. For each script, six attributes are used which help us to find, whether the prices are going up or down. The evolutionary techniques used for this experiment are the genetic algorithms and evolution strategies. By using these algorithms, we are trying to find the connection weight for each attribute, which helps us in predicting the target price of the stock. An input for each attribute is given to a sigmoid function after it is amplified based on its connection weight. The experimental results show that using this approach, predicting the stock price is promising. In each case, the algorithms were able to predict with an accuracy of at least 70.00.