Stock Prediction Model Based On ANN Method

The stock market plays a decisive role in the current economic activity and is the"barometer" and "alarm" in a national or regional economic and financial activities.Study on the trend of the stock market has a decisive significance for countries andpeople.The stock market is affected by many factors and is a highly nonlinear system. Neuralnetwork can extract the nonlinear relationship from data and simulate it, so commonlyis used to analyze the time series data such as stock price. The paper conducts atheoretical and empirical research on stock price prediction. BP neural network is easyto fall into local optimal value, has a slow convergence speed, and its accuracy is nothigh, so can be optimized by using genetic algorithm with global searching optimization.We establish the GA-BP algorithm model, take Chang’an stock for the empiricalanalysis, and use error evaluation index of MAPE to compare accuracy of twoalgorithm model. The empirical results show that the prediction accuracy of the BPnetwork has been significantly improved by using the GA algorithm, but the problemlike how to select the number of hidden layer’s nodes still need to be study.