Forecasting Stock Exchange Index Using Particle Swarm Optimization Comparing to Traditional Models

Abstract

The stock market is one of the most attractive investment choice from which a large amount of profit can be earned. This study presents a PSO-based methodology to deal with Stock market index prediction. The study showed superiority in applicability of the proposed approach by using Tehran Stock Exchange Index (TSEI) and comparing the outcomes with conventional method such as Simple Exponential Smoothing (SES), Hoelt-Winters Exponential Smoothing (HWES), Auto Regressive (AR), Moving Average (MA), Auto Regressive Integrated Moving Average (ARIMA). Experimental results clearly showed that PSO approach meaningfully outperforms all of the conventional method in terms of MAD, MSE, RMSE and MAPE. Additionally, evaluation statistics of the proposed approach significantly decrees variance of the errors compared to the conventional method.

Keywords

• particle swarm optimization algorithm
• simple exponential smoothing
• hoelt-winters exponential smoothing
• auto regressive
• moving average
• auto regressive integrated moving average