Prospect Theory And The Stock Price Short-term Prediction

According to the efficient market hypothesis, stock market is unpredictable, butbehavioral finance studies have shown that stock returns have a certain degree ofpredictability. The traditional finance theory with explicit mathematical expressionsdescribing the relationship between stock price and its influence factors, but this isbuilt on the basis of strict "economic man" hypothesis, at variance with reality.Investors are bounded rationality, according to the traditional financial theory topredict stock price, the effect of course will not be too idealistic. The stock market is ahighly complex and nonlinear dynamical systems, the economic development quality,market fluctuations, changes in policy, and even a key conversation are likely to havea greater impact on the stock market. Thus using the traditional statistical time seriesmodel to predict stock prices wouldn’t achieve satisfactory results.For the above, this paper will take the non-validity of the market as a precondition,take the prospect theory as prediction basis, take the BP neural network as theprediction model. The purpose of the study is to explore a more effective method forquantitative analysis of stock market. Under the framework of prospect theory, thispaper analyzes the investment decision-making process of investors and points outthat the cost of buying and the stock price time series will affect the investmentdecisions of investors, the cost of distribution and the stock price time series has animportant impact on the market outlook. Based on the prospect theory, this article willanalyze the cost of distribution of investors and construct the input index system,finally, take BP neural network model for short-term forecast of the ShanghaiComposite Index. In the forecasting process, this paper will set up two control groups.The first control group will take technical analysis indicators as input variables, thesecond control group will take exponential smoothing model as the predictive model.The experimental results show that compared with the first control group, theprediction based on prospect theory indicators is superior on fitting results and theprediction accuracy, and its "malfunction" phenomenon is less than the first controlgroup, this suggests that taking prospect theory as prediction basis is more reasonable.Compared to the second control group, in prediction accuracy and stability, theprediction of BP neural network model are better than the prediction of exponentialsmoothing prediction model. This indicates that the BP neural network model in the prediction of stock market has obvious superiority.