Font Size: a A A

Study On The Long-term Memory Of Chinese Stock Market And Its Trend Prediction

Posted on:2018-06-21Degree:MasterType:Thesis
Country:ChinaCandidate:Q ZhangFull Text:PDF
GTID:2359330536483878Subject:applied economics
Abstract/Summary:PDF Full Text Request
This paper uses of the new Two-step Exact Local Whittle estimation for the first time in the domestic,estimating the long term memory parameters of C hinese stock market,also suggesting that the threshold and the effective conditions of the decline of memory parameters when the stock market trend changes by analyzing the characteristics of time-varying memory parameters.In order to estimate the time-varying memory parameters,the optimal parameters of the above estimation method,which is the long-term memory cycle and the bandwidth,determined by the Monte Carlo simulation method.In addition,this paper also examines the existence of long-term memory authenticity.To begin with,based on the characteristics of nonlinear correlation of the Shanghai Composite Index return series,and test results prove that the sequence also has long-term memory and structural change,then proved that the long-term memory of the seq uence is not caused by structural mutationsby using the method of sample partition test and d order difference stationarity test.In addition,after determining the average cycle length of the Chinese stock market,and using Monte Carlo simulation method to determine optimal parameters of long-term memory cycle and bandwidth required to estimatethe time-varying parameters.The simulation results show that when n=260,m=[2600.65],long memory parameter estimation d which has consistency and asymptotic normality.At last,this paper achieved that nine times upward or downward trend conversion signal,eight times is correct,only one error signals is presented during the period from October 8th,2004 to December 30 th,2016.The time-varying memory parameter d not only captures the important turning points,but also captures some secondary turning pointsduring the sample period.Meanwhile,this paper identifies that the threshold value of the memory parameter decreased when the stock market trend changes based on the technical analysis,and conditionfor effective decline of memory parameters when the stock market local trend has change.
Keywords/Search Tags:ARFIMA(p,d,q) Model, Long-term Memory, ELW estimations, Trend
PDF Full Text Request
Related items