The Method Of Financial Time Series Prediction Based On State Space

In the financial field, Time Series Data is the important one of this kind, which is very great meaningful for the analysis and research. However, in the Practical application, since the particularity in the financial field and limitation of traditional method, the error can be happened in the prediction of time series data. So this thesis introduces the new kind of analysis method of Time Series Data, namely State Space Model Method.It can provide the most reasonable analysis frame for the financial time series model. This thesis make the research on the basis of the forecasting method of the time series data of State Space Model Method.1. Firstly introduced the theories such as State Space Model Method, Kalman Filter, Kalman Precursor, Kalman Smoother, The most superior smooth fixed sector, Stable state of Kalman Filter, and Edges the stability.2. The Recursion Iteration Method can be introduced in detail which make use of the combination of the Kalman Filter, the most superior smooth fixed sector, The forecast error decomposes the likelihood function and EM algorithm, so it can make full use of the total time series data, which can obtain the parameter prediction.3. By way of the contrast forcasting results, Separately use the Product season model and State Space Model to establish the model for the non-steady time series, and Research object about the monthly deposition Remaining sum data of dwellers both in the city and the country.4. Through the specific P= or Q = options are the series of time-gap of the Parenthesis in the Product season model, which time length can control the specific parameter, so the obtained Subset model is much better and precise than the traditional model.