Financial time series segmentation based on Turning Points

Segments extracted from financial time series are widely used in trend analysis as well as in predicting future tendency of the price movement. Recent approaches for time series segmentation often rely on an arbitrary threshold value and segments are generated at only one level. In this paper, we propose a novel time series segmentation method based on Turning Points which are extracted from the maximum or minimum points of the time series. The proposed segmentation method generates segments at different levels of details and achieves satisfactory results in preserving higher number of trends compared to an existing segmentation approach.