Discovering the Lead-Lag Relationships in Financial Markets: A Method based on DTW

摘要

The lead-lag relationships of stock prices are of great importance for its effectiveness in the fields of risk control and the portfolio investments. In this paper, we propose a new method to analyze the inherent lead-lag relationships of stocks in financial market. Considering the “golden crosses” and the “death crosses”. we represent multivariable stock time series as dynamic competitivity graph series to capture the meaningful signals of the changing trends. Inspired by the corresponding relation of optimal path in the process of calculating Dynamic Time Warping, we propose a new method called LLR to detect the dynamic time evolution of the lead-lag structure between two dynamic competitivity graph series. Finally, we verify the obtained relationships in Chinese A-share market. The experimental results demonstrate that it is effective to use the relationships revealed by LLR to make prediction of stock trend and explain the financial law of stock market.