A Study On Granularity Optimization Of SAR Patterns For Financial Time Series

Financial time series analysis is an important research direction in the field of time series analysis.One of the financial time series analysis methods is to find a specific pattern to determine the operation law of financial time series,and SAR pattern is an important one.SAR is the acronym for stop and reverse,which provides the location of stop loss.When the SAR pattern is applied to the period of market shock,because of its lag,it will lead to delayed decision-making and operation signals,sharp decline in revenue and even negative revenue.To solve this problem,this paper adopts the idea of granularity adjustment to complete the following research work:(1)Financial time series data acquisition and preprocessingObtaining relevant data is the basis of the work.Firstly,Sohu platform is selected as the data download channel of this paper;Secondly,set the interface parameters;Finally,select the public stock time series data to build the data set.(2)Granularity expression of SAR pattern and its defect analysisFirstly,according to the application principle of SAR index,a series of decision points are determined to obtain SAR pattern;Secondly,the decision points in SAR pattern are fitted into a series of pattern segments by piecewise linear representation method,which are transformed into particles with different lengths and sizes;Finally,a series of ideal time series are constructed from the two angles of earthquake amplitude and decision point interval to test the SAR pattern.It is analyzed that the defect is that the decision point interval is too small.(3)A posteriori granularity optimization algorithmAiming at the defects of SAR pattern,two kinds of a posteriori granularity optimization algorithms are proposed,one is fixed parameter and the other is non fixed parameter.The principle of the algorithm is to filter out the decision points with too short interval by adjusting the granularity,so as to improve the income of SAR pattern.The experimental results show that the algorithm with fixed parameter 7 has the best revenue promotion effect.(4)Real time granularity optimization algorithmFor real-time SAR pattern optimization,a real-time optimization method of unbalanced strategy and equalization strategy is proposed.Real time data processing and revenue optimization are realized by adjusting the buy and sell decision rules.The experimental results show that the best strategy is "6-1" disequilibrium strategy.After comparison,the physical meaning of the optimal granularity parameter is different from that of the optimal strategy.(5)Granularity optimization algorithm combined with statistical average SAR patternAiming at the inconsistency of the above physical meaning,a granularity optimization algorithm combined with statistical average SAR pattern is proposed.The plate index is the statistical average of the data in the plate.This algorithm uses its SAR pattern to optimize the granularity of the included data.The experimental results show that the optimization algorithm is universal.The average revenue of the four sectors accounts for 73%,and the average revenue accounts for 64% from 40%.