Research Of Stock Price Fluctuation Forecasting Based On Fuzzy Time Series And Swarm Intelligence Algorithms

Financial time series is a kind of important data type in financial market.The price fluctuation of financial market is a financial time series,especially the stock price,which reflects the economic development of the operating conditions and operating conditions of enterprises,and is an economic barometer that is the major concern of investors.According to the fluctuation of the price of financial market,investors can adjust their investment portfolio to achieve the purpose of avoiding risks and improving the profits.Therefore,the research on modeling and forecasting of stock price time series is particularly important.Since the financial market,scholars have been studying the price changes in the stock market(such as stock,fund,exchange rate,etc.).They believe that stock price time series has high noise,non-stability and long term non-predictability.According to these characteristics of the financial time series forecasting,many economists put forward mathematical models to predict the financial time series from the point of view of statistics.Theoretically,it has a certain reference value to the real financial market.But these models have statistical assumptions limiting,such as normality and stability.The financial time series is a complex,random,uncertain and nonlinear system,so these traditional methods with many assumptions to predict the stock price volatility is clearly not accurate and objective.So this kind of method is restricted in the application of financial time series forecasting.In recent years,the rapid development of computer science,data mining,data analysis,machine learning and artificial intelligence technology,the swarm intelligence algorithm among them has been paid more and more attention by people.It does not need to carry on the statistical hypothesis to the data,mostly uses the metadata,and thus has guaranteed the data objectivity.Many scholars also try to combine these theories with the financial time series forecasting model,andthe forecast effectfrom the results is also very good.At the same time,the fuzzy set theory is proposed,which provides a new angle for the modeling of financial time series with the characteristics of randomness and uncertainty.Based on the above problems,this paper combines the swarm intelligence algorithm with two kinds of fuzzy time series model,and puts forward two enhanced financial fuzzy time series models to forecast the fluctuation of stock price.In this paper,we focus on two types of financial multivariate-high order fuzzy time series model,including Type-1 and Type-2 fuzzy time series forecasting model of development and construction.In order to solve the two problems of how to divide the interval for the Type-1 and Type-2 fuzzy time series models and how to optimize the order for Type-2 fuzzy time series,the swarm intelligence algorithm is applied to solve the problems.The innovations of this paper are as follows:(1)We proposed a new fuzzy time series model based on genetic fuzzy clustering and multi-factors BP neural network.This paper combines genetic algorithm and fuzzy C means clustering algorithm to solve the partition of universe of discourse.Genetic algorithm has the advantage of global search.(2)This paper proposed a higher-order and multi-factor Type-2 financial fuzzy time series model based on acuckoo search algorithm and the improved adaptive harmony search algorithm.Type-2 fuzzy time series model is a three-factor forecastingmodel;this paper puts forward a four-factor one.(3)The Shanghai Stock Exchange Composite Index is used as both training and test data to verify the performance of the proposed method,and compare it with several benchmark models.The experimental results show that the forecasting accuracy of these two types of financial fuzzy time series model is higher than the benchmark models,and fully verifies the applicability and effectiveness of the both proposed models.This study combined the swarm intelligence algorithm and fuzzy theory to construct and improve the two kinds of fuzzy time series model to solve the prediction problem is mixed with different orders,different factors of the stock price.It provides two new methods for solving this kind of problem and a model of choice for investors to avoid risk and improve income.