Stock Market Forecasting Using ant Colony Optimization Based Algorithm

Abstract

Due to the importance of forecasting the capital market earnings in finance, recently the aspect of stock market prediction has been a major research area that has generated a lot of attention involving various machine learning algorithms. In the recent presentations, it has been indicated that neural networks have some drawbacks in learning the data patterns or that they may perform inconsistently and unpredictable because of the complexity of the stock market data. However, due to the distributive nature of the capital market, a computational intelligence technique called Ant Colony Optimization (ACO) which is suitable for solving distributed control problem was applied in this paper, to get the most optimal solution from three technical analysis strategies. The obtained optimal prediction of the next day closing stock price the ACO algorithm performs better than the other three approaches (Price Momentum Oscillator, Stochastic and Moving Average). Our algorithm (ACO based) was evaluated to have the accuracy of 0.812500, Sensitivity of 0.907407 and Specificity of 0.690476. The ACO based technique have the highest accuracy, Sensitivity and Specificity than the other three (3) technical indicators in predicting the next day closing stock price. Therefore, the optimal prediction of our ACO Agent provides a better forecast than the three initial strategies.