A Short-term Stock Price Based On Artificial Neural Network Prediction Model

The artificial neural network is a kind of network made of by a great deal of neural units topologically, acting as a miniature of human's brain. Owing to its excellent ability for processing information, it possess popularity in many fields, such as calculating, automation, finance and so forth. In this paper, we will probe its usage in the forecasting of short-term price of stocks.The artificial neural network surmounts the traditional forecasting methods in two aspects—it can take advantage of the combination of many factors that can affect the price and probing the law of movement of the price dynamically. Here we use the B-P network and RBF network in predicting the price of the stock of Thompson Corporation. We chose the input variants cording to the relations between them not abiding by the traditional method—the principal component analysis, using the unitary data, in order to accelerate the convergent speed.In this article we pay more attention to the accuracy estimation , the role of rational expectation in predicting the short-term stock price, the feasibility of predicting the abnormal price and the reliability of forecasting the figure of reversion, and at reach the conclusion that :the neural network can predict normal stock price with accuracy ,when abnormal price with inaccuracy especially in predicting the lowest price, rational expectation can impose great effect on the stock price, and it can forecast the price reverse successfully. At last, we give some advice on further studies.