A Fast Similarity Search in Stock Market Time Series
Xiaoye Wang · Journal of Hebei University of Technology · 2005
This paper presents a new method of the similarity of stock market time series.Comparing with the existent method,it is robustness and fastness.We represent the stock market time series based on the piecewise linear,which reduces the quantity of the data and the complexity of compute.In the stock market analyses,a usual tool is the chart pattern.The valleys and the peaks of these chart patterns obey some rules about their relative position.We sort the valleys and the peaks based on value,and we sort time series based on the value also.Thus,the position of the data will be permuting.The pattern similarity is to decide if the given pattern is similar to the chart pattern by looking at if they have the same permutation of the data.