Security Analysis Platform And The Study Of Long-term Trend Prediction

In recent years, predicting the data of the bond has already shown more and more important value of the research caused by the important in the management of finance and decision.It is great significance to predict the trend of bond as an important research project of nonlinear time series analysis. The data of the bond can be regarded as the data of nonlinear time series, and we can use the way of nonlinear time series to predict. But it is difficult to predict the short the trend of the data of the bond because of complicated and uncertain factors which include the policy of the government, company situation, important news and investor sentiment.Now we can use the way of time series predication to predict the trend of the bond accurately under in the long period condition because the market of bond shows obvious nonlinear characteristics. In order to make up the uncertain in the prediction in the short period and make people understand the stock market with the angle of macroscopic to get the guiding of temporary investment, this dissertation presents to build the model of nonlinear time series to predict the trend of stock market in the long run. From view of essence it is the design of the system of the trend of bond. For the function we should firstly settle in how to collect the data to build a simple analysis interface to observe and judge. The core task is to design long-term analysis system of the trend of the bond based on relevant selection algorithm nonlinear time series. This dissertation consists of the four parts, which will be specified in following sections.Firstly, to protect intellectual property, the corresponding module is designed in this dissertation. The two algorithms, DES and RAS, for encryption are introduced at first, and then the algorithm used in this system is specified. This system combines the characters of hardware's physical series and logical series into one unite to create the software ID.. The method is specified as follow: use the Windows API function making programs to obtain the computer's hardware's physical series and logical series, and then intercept and combine into the software's ID. Then the register password can be gotten by transferring the software's ID according to the time encryption algorithm, at last, the register machine is gotten to register conveniently for users. There are 10 English letters in this software, and there are 26 probabilities for the password, thus, there are 141 trillion probabilities for the software's password in the system, so, resolve this problem is so hard. After time-encrypting, the capability for defending destroying can be improved again. Endless search, for example, this system is time-encrypted, because of this, different date should be coupled with different passwords. If the endless search does not make work, the system's password must be changed in the next date, if you should destroy it, the similar work must be done as the same process.The second, the data interface platform is set in this dissertation. The data interface is used to connect to the exterior data. Because the quantities of the data is very large, and they are updated with time changing and are emerged, how to introduce the data into the system is the key function to make this system operate. The data is formatted into binary data in this system, and introduce these data into the system. The next step is, based on period analyzing requested, and several periods analysis methods are integrated into one, and become certain data analyzed, then the corresponding curve can be gotten. Based on this process, the generalization rules for bonds changing are analyzed, it is conclude that the categories of the bonds have being prepared for selecting is the key basic standard for long period investment.The third using relevant algorithm of time series to compute relevant time series to find same sample which has known with the trend mode of bond data to predict the trend of sample. Relationship degree can be regarded as a reference in predicting bond changing. Relation degree can denote the relation degree between two serials and can analyze the bond changing, and can analyze the history changing and restore them into modules such that can be used to analyze the similar situation, which can be regarded as predict in some degree. It is shown from examinations that, the high similarity for the data between in the process sampling and in the module, the small relevant error and the more similar changing between them, the small kinds bond can be selected, and the precise prediction for the bonds changing can be gotten. Anyway, this predicting method can make us satisfaction.Lastly, the algorithm about Vector Machine (SVM) is applied in this dissertation to predict the long-scale changing for bond. Here, the core mapping is used in Vector Machine (SVM), and the input samples is mapped into a certain space H. In this character space, the samples can be separated into several sets according to the linear methods. At first, the data is conducted to normalization, then select the data normalized take part in training, and category examination. The result shows that this method can predict the bond and its changing more accurately, and can provide some standard for deciding.The whole software system is easy and applied based on VB6 and MatLab. Especially it is guiding significance in reality.