Research & Application Of Combination Forecasting Method On Stock Rate Of Return

Based on genetic algorithm, this paper proposes a nonlinear combination forecast model consisted of improved GM(1,1) and ARCH models. This model's basic thought is to set parameters of this combination forecast model by the learning ability of genetic algorithm to decrease the difficulties of constructing nonlinear combination forecast function. It is confirmed that this combination forecast model based on improved genetic has effectiveness and feasibility by theoretical analysis and application instances. One of the good examples is that some uncertain nonlinear system combination and model such as stock index average rate of return.With the guidance of system theory and engineering, this paper has a research on current theories and methods and builds a responding mathematical model by the combinational use of econometrics, technology economics, statistics, investment combination and so on. This research presents the technical line of theory, practice and application and the concrete contents includes: the improvement and application of GM(1, 1) in financial time data forecast. By the EGARCH model, this paper tries to forecast the object's average rate of return, analyze VAR model and impulse response function. This paper tries a new nonlinear combination form based genetic algorithm. This paper calculates and analyzes the whole and some single market average rate of return, including the model and variable selection, data collection and processing, statistical data arrangement and results analysis.