The Research About Combined Model In The Stock Market Forecasting

Stocks has become an indispensable part of financial markets， the ups and downsof stock prices, and the changes of stock market, is not only the impact by the internallaw, but also by the external environment such as political, economic and other factors,but investors still want to be able to analyze and predict stock prices accurately in orderto obtain lucrative income. So how to build a more ideal forecasting model has been theresearch content by many scholars.select two combined model is selected to analysis, forecasting and compare theShanghai Composite Index in this article，which is Grey-Markov model and geneticalgorithm optimization of BP Neural Network model. First, create a gray system modelGM (1,1), which describes a gray variables, and is more suitable for system objects withlittle data， short-term, small fluctuations, but is not ideal for relatively large fluctuationsin the data series, the fitting is relatively poor, the prediction accuracy. So Markovprocess is introduced because of this-disadvantage against gray model, then use thecombination method Grey-Markov model to predict the Shanghai Composite Index;Secondly, through the modeling and analysis of BP Neural Network, found there aresome problems in a local minimum point and on the convergence rate，and introduce theanalysis and research of genetic algorithm, establish a second combined model based ongenetic algorithm BP Neural Network, and made a single comparison with BP NeuralNetwork, then get the conclusion that the combined model prediction is better; Finally,by comparing two combinations methods to analyze their respective characteristics of thedata presented, in the end of the article，predict the data of the Shanghai index relativelylarge fluctuations.The experimental results showed that: On the whole， GM (1, N)-Markov isprediction accuracy higher than the GA-BP model，in the case of stable data and the data fluctuation，and from the relative prediction error measure, but in otherperformance indicators, the difference between the two models is not great.