A novel time series behavior matching algorithm for online conversion algorithms.

This work presents a novel time series behavior matching algorithm for analyzing behavior (trend) similarity between two given time series. Unlike traditional approaches, our dynamic programming based approach "Behavior Matching (BM)" is based on trends and behavior rather than absolute distance as similarity measure. In order to compare the effectiveness of our proposed algorithm, we conduct an experimental study on real world stock data (DAX30). We compare our proposed algorithm with state-of-the-art algorithm Euclidean Distance, V-Shift and Dynamic Time Warping. The experimental results validates the performance guarantee and consistency of our proposed scheme. [ABSTRACT FROM AUTHOR]