Security Price Studying With Combination Forecasting

Securities price is one of the hot issues in the field of economic and system science. Because of the variability in securities market, this paper tried to find the law of the fluctuant stock price and forecast it for stock investment. We can hardly get the completely real model due to the uncertainty and rapid change in securities market, so it is undoubtedly that there be great value to seek a model which is more close to the reality. But in most forecasting, because of the objective and timeliness, it can hardly wait for finding the real model. The new uncertainty may instead of the old one, so single model which is highly sensitive or complicated will face the risks of the false model assumptions. The starting point of combined forecasting accepts the difficulties of structuring real model. Combined forecasting model put every single model as different pieces of information. Through the integration of the information, combined forecasting can reduce the uncertainty and improve the accuracy of forecasting. As a result, constructing the model of combined forecasting to predict the fluctuation of stock price has a theoretical value and a strong guidance.In this paper, ARIMA model, GM (1,1) model and RBF neural network model are respectively used to forecast stock price. Then, by integrating the useful information of the three models, we can construct the best model of combined forecasting for empirical research of single securities pricing. This paper includes three parts. The first part is to systematically analyze single model and its application in the financial field, in order to lay a theoretical foundation for constructing the model of combined forecasting. The second part is to construct the model of combined forecasting based on three single forecasting models. The third part is to build single forecasting model and combined forecasting model for single securities price to empirical research, and compare the forecasting accuracy of single forecasting model with combined forecasting model. The results showed that the forecasting accuracy of combination forecasting model forecasts is better than the single models participated in the combination forecasting model.In this paper, the innovation point is to construct the optimal combinative forecasting model based on single forecasting models of ARIMA model, GM (1,1) model and RBF neural network model, which uses the least error square of forecast as the objective function to respectively give the single forecasting models weights, concluded that combination forecasting method combines each information effectively. Therefore, combination forecasting model is more comprehensive than single model, in some extent, compensate for the limitations of single model.