The Research Of Stock Trend Prediction Based On Kuramoto Influence Transmission Model

With the rapid development of China’s economic health, people’s living standards gradually improved, to reinvest in the stock market for surplus property becomes an important financial means. However, the stock market is a highly complex financial system, due to stock price volatility has strong mutation and susceptible to external factors, the stock market is often accompanied by high-risk and high-profit co-exist. It is of great significance to improve the major stock prediction accuracy, but also challenging.In the research, the analysis of trends and data characteristics in the stock market, found that it is inherent correlative between the stock market trends and the transmission relationship of different time-line. On account of the characteristic of mutable and the problem of interference data redundancy from big data during the stock forecast, via the perspective of stock price synchronicity, two models are proposed, the first one is Kuramoto stock market prediction model based on influence transmission(IT-KFM), the other is stock market trend prediction model based on hierarchical influence(HI-TPM), research work are as follows:(1)Firstly, from the feature of stock price synchronicity, analyzes the meaning of stock price synchronicity and its common analysis methods. In the stock market synchronization phenomenon between the different time-sharing line can partly reveal to the basic mechanism of market trends. Based on the synchronous vibrator Kuramoto model, IT-KFM model utilizes the bayesian network to build structural relationships between the oscillators, formalizes time-line Kuramoto stock prediction model, then influence transmission and quantitative method of transmission factor are introducd, the transmission parameters of transmission factor are added to the original Kuramoto model. At last according to the covariance between different oscillator phase trend analysis and forecast stock market trends. Experimental results, obtained by running on datasets taken from The Shanghai Market, show that the performance of our method is better than the standard SVM algorithm on stock trend forecast.(2)Secondly, IT-KFM model only uses the moving average data transmission, which has the characteristics of time delay, may leading to reduced accuracy of the prediction model. Improved on the IT-KFM model, HI-TPM model, starts from the influence of the transmission factor, excavates hierarchical influence via several suitable technical indicators as coupling, using analytic hierarchy process to weight influential factor, adopts analytic hierarchy process to determine the type and weight ratio of Transmission factor between different levels. Tested on the same datasets, the effect has improved.Experimental testing and comparative analysis show that, IT-KFM and HI-TPM model has good predictive accuracy; at the same time, experimental results show that the stock market trends and influence transmission of stock market index are inherent correlative, it has important value on the research of stock market’s prediction model.