Financial Time Series Forecasting Research Based On Neutrosophic Sets

Fuzzy time series models in handling data which has the uncertainties inside,and the random influence outside have a unique advantage,traditional fuzzy time series models are mostly build within the statistical features of time series and development regularity of the problem.In fact,the nonlinear dynamic system of the stock market is affected by multiple uncertainties,uncertain and inconsistent which lead to produce noise.In the processing of forecasting,describing the changes of stock market has important theoretical and realistic significance.This paper takes the financial time series as the research object,and according to the characteristics of the financial time series,using the neutrosophic set and fuzzy time series to predict the future.Finding the inherent rules and patterns of a time series by eliminating disturbances without losing important details has long been a research hotspot.The true degree,uncertainty degree and false degree of neutrosophic set can solve the problem,and provide a more detailed representation of the uncertainty performance of stock index fluctuations,so as to improve the prediction accuracy.Therefore,this paper constructs three time series prediction models based on neutrosophic set.TAIEX from 1998 to 2010 and HSI and SHESCI time series data are selected for case analysis.The similarity measurement method and the influence of different orders on the model are discussed.The models proposed are compared with typical traditional prediction models to verify the validity of the model constructed in this paper.The main conclusions are summarized as follows:(1)Stock market is a non-linear dynamic system with full of noise.The multi-dimensional information expression of multi-valued neutrosophic sets and neutrosophic soft sets can be able to retain the valuable information of historical data.(2)Considering the relevant factors and inherent attributes of the stock market.Multi-factor prediction rules are established from the consideration of external factors and multi-factor high order time series are constructed.In order to improve the prediction accuracy,the multi-attribute high order fuzzy time series was constructed by integrating the variables related to the current model.(3)Considering the uncertainty of the stock market.On the basis of the prediction model of time series based on neutrosophic soft set,information entropy is used to depict the uncertainty degree of time series at a time in financial time series,so as to further express the internal rules of stock market accurately.This paper not only considered the inside and outside the complex changes in the stock market,also considering the uncertainty due to factors such as investor sentiment,policy uncertainty on the share price.Thus,proposed model using information entropy to describe the relationships forecast the stock market trends,which improve the prediction precision.