The Research Of Financial Time Series Mining And Modeling Based On Idiosyncratic Volatility

With the merging and development of computer technology, artificial intelligence,learning machines and the method of statistical analysis, the technology of data mining hasbeen developed rapidly. Besides, with the advent of Big Data, the traditional financial analysismethods have not been right for the needs and applications of the financial data analysis. Sousing data mining to analyze financial time series data is in tune with the times.Under this background, this paper use data mining technology to research and modelingthe time series of stock prices. In the meantime, the research on the trend forecasting whichused Idiosyncratic Volatility is relatively small. Therefore, this paper use the self-definingmethod which called TBUD to set partitioning of the time series of stock prices, and useSupport Vector Machine for modeling, researching the prediction ability between IdiosyncraticVolatility, one of the attributes of set, and the stock prices trend. The empirical research of thispaper found that Idiosyncratic Volatility which is the attribute of the collection set partitioningby the TBUD method is not significant among the collections. Also Idiosyncratic Volatility isunable to make an accurate prediction of the trend of share prices.This paper proposes an partitioning method of the inflection point on the time series set,getting rid of from the regression equation to forecast the time sequence, but from theperspective of data mining, studying the relativity between inflection point set and trend,providing a new direction for future research.