Research On Stock Forecasting Based On Bacterial Foraging Optimization Algorithm

Stock investment is an important aspect of modern economic activity,the stock market plays an important role in the capital financing,the effective allocation of resources,the price discovery of financial assets and the redistribution of social wealth.Because the stock market is affected by various factors,stock prices showed significant non-linear,non-deterministic characteristics,stock market so great turmoil in financial markets,a direct impact on the stability and development of financial markets,but also on investment those gains have an important impact.How to find a method to accurately predict the stock price,and reasonable control and health guidance on the stock market and the interests of our future national economic growth and investors have important reference value.Neural network technology has a strong nonlinear approximation ability and self organization,self-learning ability,which has been widely applied in stock prediction.But the single neural network model is slow and easy to fall into local optimum.This article is based on the BP neural network,introduction of bacterial foraging algorithm(BFO)to optimize the network,then select model parameters which makes the network performance to achieve optimal.For BFO slow training speed,low global optimization efficiency,introduction of PSO algorithm thought in BFO to optimize network learning function,put forward a new model of PSO-BFO-BP.In order to verify the effect of the prediction model,using the BP neural network model with the same structure,BFO-BP model and the PSO-BFO-BP model.And apply to the stock price prediction,choose China mobile,China unicom,China telecom three stocks from January 3,2011 to November 30,2015,1209 days of data as the training sample,data from December 1,2015 to December 31,2015 data as the test sample,using MATLAB software simulation.The empirical results show that: compared with single BP neural network model,BFO-BP model and PSO-BFO-BP model greatly improve the prediction precision of the stock price,and PSO-BFO-BP Model to improve the prediction accuracy but also speed up the training speed.